402
FlexFrame™ for SAP ® Version 5.1A Administration and Operation Edition July 2012 Document Version 1.2

Administration and Operation - Fujitsumanuals.ts.fujitsu.com/file/10478/FF51A00_Administration_Operation.pdf · FlexFrame™ for SAP® Version 5.1A Administration and Operation Edition

  • Upload
    others

  • View
    27

  • Download
    0

Embed Size (px)

Citation preview

FlexFrame™ for SAP®

Version 5.1A

Administration and Operation

Edition July 2012

Document Version 1.2

Fujitsu Limited

© Copyright Fujitsu Technology Solutions 2012

FlexFrame™ and PRIMERGY™ are trademarks or registered trademarks of

Fujitsu Limited in Japan and other countries.

SAP® and NetWeaver™ are trademarks or registered trademarks of SAP AG in

Germany and in several other countries

Linux® is a registered trademark of Linus Torvalds

SUSE® Linux is a registered trademark of Novell, Inc., in the United States and

other countries

Oracle™ and Java™ are trademarks of ORACLE Corporation and/or its affiliates

Intel® and PXE

® are registered trademarks of Intel Corporation in the United States

and other countries

MaxDB® is a registered trademark of MySQL AB, Sweden

MySQL® is a registered trademark of MySQL AB, Sweden

NetApp® and the Network Appliance

® logo are registered trademarks and Network

Appliance™ and Data ONTAP™ are trademarks of NetApp, Inc. in the U.S. and

other countries.

EMC®, CLARiiON

®, Symmetrix

® and Celerra™ are trademarks or registered

trademarks of EMC Corporation in the United States and other countries

VMware®, ESX

®, ESXi, VMware vCenter, VMware vSphere are registered

trademarks or trademarks of VMware, Inc. in the United States and/or other

jurisdictions.

Ethernet® is a registered trademark of XEROX, Inc., Digital Equipment Corporation

and Intel Corporation

Windows® and Word

® are registered trademarks of Microsoft Corporation

All other hardware and software names used are trademarks of their respective

companies.

All rights, including rights of translation, reproduction by printing, copying or similar

methods, in part or in whole, are reserved.

Offenders will be liable for damages.

All rights, including rights created by patent grant or registration of a utility model or

design, are reserved.

Delivery subject to availability. Right of technical modification reserved.

Administration and Operation

Contents

1 Introduction ..................................................................................................... 1 1.1 Requirements .................................................................................................... 1 1.2 Notational Conventions ..................................................................................... 1 1.3 Document History .............................................................................................. 2 1.4 Related Documents ........................................................................................... 2 1.5 Special Hints for FlexFrame 5.1A ...................................................................... 2 1.5.1 Incompatibilities (command scripts) .................................................................. 3

2 FlexFrame Architecture .................................................................................. 5 2.1 General Notes on FlexFrame ............................................................................ 5 2.2 Hardware and Software ..................................................................................... 6 2.3 Shared Operating System ................................................................................. 9 2.3.1 Shared OS Boot Concept .................................................................................. 9 2.3.2 Control Nodes ................................................................................................. 10 2.3.3 Application Nodes ........................................................................................... 10 2.4 FlexFrame Structure in LDAP.......................................................................... 11 2.4.1 Working with LDAP ......................................................................................... 12 2.5 Linux-HA Cluster on Control Center ................................................................ 12 2.5.1 Terminology..................................................................................................... 12 2.5.2 Simple Resources ........................................................................................... 14 2.5.2.1 Constraints ...................................................................................................... 14 2.5.2.2 Score ............................................................................................................... 14 2.5.2.3 Stickiness ........................................................................................................ 15 2.5.2.4 Resource Groups ............................................................................................ 17 2.5.3 FlexFrame Specific Configuration ................................................................... 17 2.5.3.1 Configuration of Limits for Cluster Resources ................................................. 19 2.5.3.2 LDAP Configuration ......................................................................................... 20 2.5.3.3 Ressource ff_manage ..................................................................................... 22 2.5.3.4 Resource Netbooot ......................................................................................... 22 2.5.3.5 STONITH......................................................................................................... 22 2.5.3.6 Configuring the Cluster .................................................................................... 24 2.5.3.7 Starting Behavior of the Cluster ....................................................................... 25 2.5.3.8 Status Information of the Clusters ................................................................... 25 2.5.4 Linux-HA CLI Commands ................................................................................ 26 2.5.5 Linux-HA Logfile .............................................................................................. 27 2.5.6 Linux-HA GUI .................................................................................................. 27 2.6 Network ........................................................................................................... 27 2.6.1 LAN Failover.................................................................................................... 28 2.6.2 Segments ........................................................................................................ 28 2.6.3 Network Switches ............................................................................................ 29 2.6.4 Network Speed ................................................................................................ 30 2.6.5 Network Switch Groups ................................................................................... 30

Contents

Administration and Operation

2.6.6 Network Switch Ports ...................................................................................... 30 2.6.7 Automounter Concept ...................................................................................... 31 2.7 Storage Systems ............................................................................................. 34 2.7.1 NAS Support Architectural Overview ............................................................... 34 2.7.1.1 Network Appliance Filer ................................................................................... 34 2.7.1.2 EMC NAS (Celerra) ......................................................................................... 36 2.7.2 SAN Support Architectural Overview ............................................................... 40 2.7.2.1 SAN Basic Layers ............................................................................................ 41 2.7.2.2 Scope of the FlexFrame SAN Integration ........................................................ 42 2.7.2.3 Rules and Restrictions ..................................................................................... 43

3 FlexFrame Basic Administration .................................................................. 44 3.1 Accessing a FlexFrame Landscape (Remote Administration) ......................... 44 3.2 Powering up the FlexFrame Landscape .......................................................... 44 3.3 Powering off the FlexFrame Landscape .......................................................... 45 3.4 Reactivating ANs after Power Shutdown by FA Agents .................................. 48 3.5 Displaying the Current FlexFrame Configuration State ................................... 49 3.5.1 FlexFrame Web Portal ..................................................................................... 50 3.5.2 FA Autonomous Agents ................................................................................... 51 3.5.2.1 State of Pools .................................................................................................. 51 3.5.2.2 State of Application Nodes .............................................................................. 52 3.5.2.3 State of SAP Systems ..................................................................................... 52 3.5.2.4 State of SID Instances ..................................................................................... 52 3.5.3 Networks.......................................................................................................... 52 3.5.4 ServerView Operations Manager ..................................................................... 53 3.5.5 Cluster Status .................................................................................................. 61 3.5.6 Supported Hardware ....................................................................................... 61 3.5.7 Management Blades ........................................................................................ 61 3.6 FlexFrame Backup with Tape Library .............................................................. 62 3.6.1 NetWorker ....................................................................................................... 62 3.6.2 ARCserve ........................................................................................................ 63

4 Pools and Groups .......................................................................................... 64 4.1 Adding a Pool .................................................................................................. 64 4.2 Removing a Pool ............................................................................................. 69 4.3 Listing Pool Details .......................................................................................... 70 4.4 Listing All Pools ............................................................................................... 74 4.5 Changing Pool DNS Domain ........................................................................... 76 4.6 Changing Pool Default Router ......................................................................... 77 4.7 Adding a Group to a Pool ................................................................................ 77 4.8 Removing Pool Group ..................................................................................... 78 4.9 Changing Group Assignment of Application Nodes ......................................... 79 4.10 Changing Group and Pool Assignment of Application Nodes .......................... 79 4.11 Hosts Database ............................................................................................... 79 4.11.1 Script: ff_hosts.sh ............................................................................................ 80

Contents

Administration and Operation

5 User and Groups Administration ................................................................. 82 5.1 Create, Modify, Delete, or List User(s) for Application Nodes ......................... 82 5.2 Creating, Modifying, Deleting or Listing Group(s) for Application Nodes ......... 84 5.3 Creating, Modifying, Deleting or Listing Service(s) for Application Nodes ....... 87

6 Pool-independent Spare Nodes ................................................................... 89 6.1 Creation of Spare Nodes in the ADMINPOOL ................................................. 89 6.2 Moving of a Spare Node .................................................................................. 89 6.3 Listfunction for Spare Nodes ........................................................................... 90 6.4 Handling Pool-independent Spare Nodes with FA Agents .............................. 90

7 Application Nodes Administration ............................................................... 91 7.1 Listing Application Nodes ................................................................................ 91 7.1.1 Displaying Information on a Specific Application Node ................................... 91 7.1.2 Displaying Information of all Application Nodes............................................... 95 7.2 Adding Application Nodes ............................................................................... 96 7.3 Removing Application Nodes ........................................................................ 101 7.4 Renaming Application Nodes ........................................................................ 101 7.5 Moving Application Nodes between Pools .................................................... 102 7.6 Application Nodes and SAN .......................................................................... 104 7.7 Administrating Blade Server Cabinets ........................................................... 104 7.7.1 Listing Blade Server Cabinets ....................................................................... 105 7.7.1.1 Displaying Information on a Specific Blade Server Cabinet .......................... 105 7.7.1.2 Displaying Information on all Configured Blade Server Cabinets .................. 106 7.7.2 Adding Blade Server Cabinets ...................................................................... 107 7.7.3 Removing Blade Server Cabinets ................................................................. 109 7.7.4 Changing Switch Blade Type ........................................................................ 111 7.7.5 Changing Switch Blade Name ....................................................................... 112 7.7.6 Changing Switch Blade Password ................................................................. 113 7.7.7 Getting Switch Blade Configuration ............................................................... 114 7.7.8 Change Switch Blade Uplink ......................................................................... 115 7.7.9 Move a Blade Cabinet to Another Switch Group ........................................... 116 7.8 Administrating ESX Servers and Virtual Machines ........................................ 118 7.8.1 Getting Started with ESX Servers and VMs .................................................. 118 7.8.2 ESX related global FlexFrame parameters.................................................... 120 7.8.2.1 System Code for ESX Servers and VMs ....................................................... 120 7.8.2.2 vCenter Server .............................................................................................. 121 7.8.2.3 Control LAN Default Router for ESX Servers ................................................ 123 7.8.3 Adding ESX Servers ...................................................................................... 124 7.8.4 Completing ESX Server Configuration .......................................................... 126 7.8.5 Removing ESX Servers ................................................................................. 127 7.8.6 Displaying Information about ESX Servers and VMs .................................... 128 7.8.7 ESX Servers and Pools ................................................................................. 131 7.8.8 Special Functions for Virtual Machines ......................................................... 132 7.8.9 Virtual Machine Properties and ESXi Resources .......................................... 134

Contents

Administration and Operation

7.8.10 Using vSphere Functions for FlexFrame Objects .......................................... 136 7.9 Script for Power on/off/Reboot of a Computing Node in FF4S ...................... 137 7.9.1 Synopsis ........................................................................................................ 137

8 Storage Systems Administration ............................................................... 138 8.1 NAS Systems Configuration (EMC and NetApp) ........................................... 138 8.1.1 Adding a New NAS ........................................................................................ 138 8.1.2 Removing a NAS ........................................................................................... 141 8.1.3 Configuring SNMP Traps for NetApp Filers ................................................... 142 8.1.4 Displaying all configured NAS ....................................................................... 143 8.1.5 Displaying NAS Configuration ....................................................................... 144 8.1.6 Adding a Pool to a NAS ................................................................................. 145 8.1.7 Removing a Pool from a NAS ........................................................................ 146 8.1.8 Adding a Blade (Data Mover) to an EMC Celerra NAS ................................. 147 8.1.9 Removing a Blade (Data Mover) from an EMC Celerra NAS ........................ 148 8.1.10 Create NAS Cluster Partnership .................................................................... 149 8.1.11 Move a NAS to another Switch Group ........................................................... 149 8.1.12 Switching a NetApp Filer between 1Gbit and 10Gbit ..................................... 150 8.1.13 Changing NAS Command Shell .................................................................... 150 8.1.14 Changing NAS LinkAggregate Ports ............................................................. 151 8.1.15 NAS Disk Free ............................................................................................... 152 8.1.16 Celerra SRDF-NAS High Availability ............................................................. 154 8.1.16.1 Syntax............................................................................................................ 154 8.1.16.2 Background Processes .................................................................................. 155 8.1.16.3 Diagnosis ....................................................................................................... 155 8.1.16.4 Return Codes ................................................................................................ 156 8.1.16.5 Used File Ressources ................................................................................... 157 8.2 SAN Configuration in a FlexFrame Environment ........................................... 158 8.2.1 Setting Up the SAN Configuration ................................................................. 158 8.2.1.1 Configuring Storage ....................................................................................... 158 8.2.1.2 General Remark for the Use of Navisphere ................................................... 158 8.2.1.3 Storage System Access................................................................................. 158 8.2.1.4 LUN Creation ................................................................................................. 159 8.2.1.5 Recording LUN Information ........................................................................... 161 8.2.2 Configuring Application Nodes ...................................................................... 162 8.2.3 Connecting the Storage to the Application Nodes ......................................... 163 8.2.3.1 Creating Zones on the Fibre Channel Switches ............................................ 164 8.2.3.2 Checking Visibility of the Storage System on the Application Node .............. 165 8.2.3.3 Registering Host Initiators with a CLARiiON/FibreCAT CX............................ 165 8.2.3.4 Mapping LUNs to the Application Nodes ....................................................... 166 8.2.3.5 Checking Visibility of the LUNs on the Application Node ............................... 168 8.2.4 Creating Volumes and File Systems for a SAP System ................................ 169 8.2.4.1 Creating a Linux LVM2 Volume Group for FlexFrame Usage ........................ 169 8.2.4.2 Completing the Configuration and Testing Usability of SAN for an SID ......... 170 8.3 Dynamic LUN Masking .................................................................................. 172

Contents

Administration and Operation

8.3.1 Using StorMan to Reconfigure SAN .............................................................. 172 8.3.1.1 Installation of SMI-S Provider ........................................................................ 172 8.3.1.2 Installation of StorMan ................................................................................... 174 8.4 SRDF Support in FlexFrame ......................................................................... 179 8.4.1 Storage System Configuration....................................................................... 180 8.4.2 Configuring Application Nodes for SAN SRDF Usage ................................... 180 8.4.3 FlexFrame SAN Configuration for SRDF ....................................................... 184 8.4.4 SAN SRDF Usage in FlexFrame ................................................................... 187 8.5 FlexFrame SAN Configuration ....................................................................... 189 8.5.1 Script: ff_san_ldap_conf.pl ............................................................................ 189 8.5.2 FlexFrame SAN Configuration File ................................................................ 191 8.5.3 SAN Support Scripts ..................................................................................... 198 8.5.3.1 Script: ff_san_mount.sh ................................................................................. 198 8.5.3.2 Script: ff_san_info.sh ..................................................................................... 199 8.5.3.3 Script: ff_qlascan.sh ...................................................................................... 201 8.5.3.4 Script: ff_san_srdf.pl ...................................................................................... 201 8.5.3.5 Script: ff_san_luns.pl ..................................................................................... 204

9 Switch Administration ................................................................................ 208 9.1 Adding a Switch to a Switch Group ............................................................... 208 9.2 Removing a Switch from a Switch Group ...................................................... 211 9.3 Listing a Switch Group Configuration ............................................................ 212 9.4 Generating a Switch Group Configuration ..................................................... 214 9.5 Changing the Password of a Switch Group ................................................... 215 9.6 Changing the Host Name of a Switch Group ................................................. 216 9.7 Displaying/Changing Common Network Configuration Parameters .............. 218 9.8 Adding a Switch Group .................................................................................. 220 9.9 Adding an Expansion Module ........................................................................ 223 9.10 Removing an Expansion Module ................................................................... 224 9.11 Removing a Switch Group ............................................................................. 225 9.12 Adding an Uplink to Switch Group ................................................................. 226 9.13 Extend an Uplink of Switch Group ................................................................. 227 9.14 Delete an Uplink of Switch Group .................................................................. 228 9.15 Migrating Switches of a Switch Group ........................................................... 230 9.16 Adding a Switch Port Configuration ............................................................... 233 9.17 Removing a Switch Port Configuration .......................................................... 235 9.18 Displaying a Switch Port Configuration ......................................................... 236 9.19 Displaying the Complete Switch Port Configuration ...................................... 237 9.20 Moving Device Connection to Core Switch.................................................... 239 9.20.1 Move Control Center to Core Switch ............................................................. 239 9.20.2 Move Client LAN to Core Switch ................................................................... 240 9.20.3 Move NAS System to Core Switch ................................................................ 240 9.20.4 Move Application Node to Core Switch ......................................................... 241 9.20.5 Move ESX Server to Core Switch .................................................................. 242 9.20.6 Move BX Cabinet to Core Switch .................................................................. 242

Contents

Administration and Operation

10 SAP System Handling ................................................................................. 244 10.1 Listing SAP SIDs and Instances .................................................................... 245 10.2 Updating System Configuration Files ............................................................ 247 10.3 Adding/Removing/Modifying SAP SIDs and Instances (Classic SAP Services)247 10.4 Removing SAP SIDs and Instances .............................................................. 253 10.5 Adding/Removing SAP SIDs (addon services) .............................................. 254 10.5.1 BOBJ – Business Objects.............................................................................. 254 10.5.2 Content Server (CMS) ................................................................................... 255 10.5.3 MDM – Master Data Management ................................................................. 259 10.5.4 SMD – Solution Manager Diagnostics ........................................................... 263 10.5.5 TREX (Search and Classification Service) .................................................... 265 10.6 Cloning a SAP SID into a Different Pool ........................................................ 268 10.6.1 Script: ff_clone_sid.pl .................................................................................... 268 10.6.2 Changing User and Group IDs after Cloning ................................................. 269 10.7 Cloning a SAP system into a different Pool ................................................... 269 10.7.1 Script: ff_clone_sap.sh .................................................................................. 269 10.7.1.1 Cloning SAP systems with MaxDB ................................................................ 273 10.7.1.2 Post cloning steps – DB2............................................................................... 274 10.7.1.3 Post cloning steps – Oracle ........................................................................... 274 10.7.1.4 Post cloning steps – MaxDB .......................................................................... 274 10.7.1.5 Cloning a SAP system – example session .................................................... 275 10.8 Relocating sapdata/saplog by ff_sid_mnt_adm.pl ......................................... 276 10.9 Multiple NAS Systems and Multiple Volumes ................................................ 279 10.9.1 NetApp Filer ................................................................................................... 280 10.9.2 EMC Celerra .................................................................................................. 281 10.10 Upgrading a SAP System .............................................................................. 282 10.10.1 Service Port ................................................................................................... 282 10.10.2 FA Agents ...................................................................................................... 283 10.10.3 Support SAP Upgrade ................................................................................... 284 10.11 SAP Kernel Updates and Patches ................................................................. 286 10.12 Unloading volFF ............................................................................................ 287 10.12.1 Status Quo/Solution ....................................................................................... 287 10.12.2 ff_relocate_sid_data.pl .................................................................................. 287 10.12.3 LDAP ............................................................................................................. 291 10.12.4 Move Data from volFF (how to) ..................................................................... 293 10.12.4.1 Specify Volume Before Installing SAP ........................................................... 293 10.12.4.2 Moving an Existing /usr/sap/SID .................................................................... 294 10.12.5 Delete Entries from LDAP.............................................................................. 294 10.13 Administrating SAP Services ......................................................................... 295 10.13.1 Displaying Status of SAP Services ................................................................ 295 10.13.1.1 myAMC.FA WebGUI ..................................................................................... 295 10.13.1.2 List SAP Services .......................................................................................... 296 10.13.2 Starting and Stopping Application Services ................................................... 297 10.13.2.1 Starting and Stopping SAP Services Without Root Privileges ....................... 298 10.13.2.2 SAP Service Scripts ...................................................................................... 299

Contents

Administration and Operation

10.13.2.3 SAP Service Script Actions ........................................................................... 300 10.13.2.4 SAP Service Script logging............................................................................ 303 10.13.2.5 SAP Service Script User Exits ....................................................................... 303 10.13.3 Return Code of the SAP Service Script ......................................................... 304 10.13.4 Starting and Stopping Multiple SAP Services ................................................ 304 10.13.5 Removing an Application from Monitoring by FA Agents .............................. 305 10.13.5.1 Stopping and Starting an Application for Upgrades Using r3up.................... 307 10.13.6 Service Switchover ........................................................................................ 307 10.13.7 Use ServicePings from FA Agents ................................................................ 309

11 Software and Hardware Update and Maintenance ................................... 311 11.1 Upgrading the Entire FlexFrame Landscape ................................................. 311 11.2 Software Upgrade on the Control Node ........................................................ 311 11.2.1 FlexFrame Control Node Software ................................................................ 313 11.2.2 Updating/Installing a New Linux Kernel ......................................................... 313 11.2.2.1 Software Stage .............................................................................................. 313 11.2.2.2 Install the New Kernel ................................................................................... 314 11.2.2.3 Reboot the Control Node ............................................................................... 314 11.2.3 Backup/Restore of FlexFrame Control Nodes ............................................... 315 11.2.3.1 Backup of a Control Node ............................................................................. 315 11.2.3.2 Restore of a Control Node ............................................................................. 315 11.3 Maintenance of the Control Node - Hardware ............................................... 317 11.3.1 Exchanging a Control Node........................................................................... 317 11.3.1.1 Hardware Failed, Hard Disk and Installed OS are not Affected ..................... 317 11.3.1.2 One Hard Disk is Defect, the Other One is Undamaged ............................... 318 11.3.1.3 The Control Nodes OS is Damaged .............................................................. 318 11.3.2 Replacing a Network Card – Control Node ................................................... 318 11.4 Maintenance of Application Nodes - Software ............................................... 318 11.4.1 Introduction.................................................................................................... 318 11.4.2 Schematic Overview ...................................................................................... 319 11.4.3 Installing Application Node Images from Installation Media .......................... 321 11.4.3.1 Installing the Application Node Image ........................................................... 321 11.4.3.2 Understanding the Application Node Image .................................................. 323 11.4.4 Step #1: Creating a Maintenance Base Image .............................................. 323 11.4.5 Step #2: Assigning the Maintenance Image, booting and maintaining .......... 325 11.4.5.1 Choosing a Node ........................................................................................... 326 11.4.5.2 Assigning ....................................................................................................... 327 11.4.5.3 Booting .......................................................................................................... 328 11.4.5.4 Maintaining .................................................................................................... 329 11.4.6 Step #3: Reverting the Maintenance Image .................................................. 330 11.4.7 Migrating remaining Application Nodes ......................................................... 333 11.4.8 Re-Using the Maintenance Image ................................................................. 335 11.4.9 Maintaining Use Cases ................................................................................. 337 11.4.9.1 Service Packs ................................................................................................ 338 11.4.9.2 Updating/Installing a New Linux Kernel ......................................................... 338

Contents

Administration and Operation

11.4.9.3 ServerView Update ........................................................................................ 340 11.4.9.4 Upgrading the Application Software .............................................................. 341 11.4.9.5 Updating RPM Packages on an Application Node ......................................... 342 11.4.9.6 Updating vmware-open-vm-tools ................................................................... 343 11.4.9.7 Upgrading the Application Software .............................................................. 344 11.4.9.8 Updating RPM Packages on an Application Node ......................................... 345 11.4.9.6 Updating vmware-tools .................................................................................. 346 11.5 Maintenance of the Application Nodes - Hardware ........................................ 347 11.5.1 Changing BIOS Settings for Netboot ............................................................. 347 11.5.2 Replacing a Network Card – Application Node .............................................. 347 11.5.3 Replacing a Switch Blade .............................................................................. 349 11.5.4 Replacing Power Control Hardware .............................................................. 349 11.6 Maintenance of ESXi Servers ........................................................................ 350 11.6.1 BIOS Updates ................................................................................................ 350 11.6.2 Replacing a Network Card - ESXi Server ...................................................... 350 11.6.3 ESXi Updates and Patches ........................................................................... 350 11.7 Maintenance of Other FlexFrame Components ............................................. 351 11.7.1 NetApp Storage ............................................................................................. 351 11.7.2 EMC Storage ................................................................................................. 351 11.7.3 Cisco Switches and Switch Blades ................................................................ 352 11.7.3.1 Firmware Update ........................................................................................... 352 11.7.3.2 Backup of Switch Configurations ................................................................... 352 11.7.3.3 Restore of Switch Configurations .................................................................. 353 11.7.4 DNS Servers .................................................................................................. 354 11.7.5 Third Party Products ...................................................................................... 354 11.8 MyAMC Agents .............................................................................................. 355 11.9 Project specific changes ................................................................................ 355 11.9.1 The ff_projectinfo.sh command ..................................................................... 355

12 Troubleshooting .......................................................................................... 359 12.1 Locking of Flex Frame Administration Commands ........................................ 359 12.2 Script Logging in FlexFrame .......................................................................... 360 12.3 Log Files ........................................................................................................ 362 12.4 Network Errors ............................................................................................... 363 12.5 NFS Mount Messages ................................................................................... 363 12.6 LDAP Error Codes and Messages ................................................................. 363 12.7 LDAP and Cache Coherence ........................................................................ 365 12.7.1 Linux .............................................................................................................. 365 12.8 Start/Stop Script Errors .................................................................................. 365 12.8.1 Severity ‗INFO‘ .............................................................................................. 366 12.8.2 Severity ‗WARN‘ ............................................................................................ 366 12.8.3 Severity ‗ERROR‘ .......................................................................................... 366 12.8.4 Severity ‚DEBUG‗ .......................................................................................... 366 12.9 Script Debugging ........................................................................................... 366 12.9.1 Shell Scripts ................................................................................................... 366

Contents

Administration and Operation

12.9.2 Perl Scripts .................................................................................................... 367 12.10 Debugging the Linux Kernel .......................................................................... 367 12.10.1 Netconsole .................................................................................................... 367 12.10.2 Capturing Crash Dumps ................................................................................ 368 12.10.2.1 Common Restrictions for Taking Crash Dumps ............................................ 368 12.10.2.2 „Kdump― Kernel Crash Dump Capturing ........................................................ 368 12.10.2.3 Forcing a Crash Dump .................................................................................. 369 12.11 Activate core dumps on CN or AN ................................................................. 369

13 Abbreviations .............................................................................................. 371

14 Glossary ....................................................................................................... 375

15 Index ............................................................................................................. 381

Administration and Operation 1

1 Introduction

This document provides instructions on administrating and operating an installed

FlexFrame™ 5.1A environment. It focuses on general aspects of the architecture as well

as on software updates, hardware extensions and FlexFrame-specific configuration.

It does not cover the installation of an entire FlexFrame environment. Please refer to the

―Installation of a FlexFrame Environment‖ manual for information on initial installation.

1.1 Requirements

This document addresses administrators on FlexFrame environments. We assume that

the reader of this document has technical background knowledge in the areas of

operating systems (Linux®), IP networking and SAP

® basis.

1.2 Notational Conventions

The following conventions are used in this manual:

Additional information that should be observed.

Warning that must be observed.

fixed font Names of paths, files, commands, and system output.

<fixed font> Names of variables.

fixed font User input in command examples (if applicable using <> with variables).

Command prompt: # The notation

control1:/<somewhere> # <command>

indicates that the command <command> is issued on the first

Control Node in the directory /<somewhere>.

The reader may need to change into the directory first, e.g.

control1:~ # cd /<somewhere>

control1:/<somewhere># <command>.

Introduction Document History

2 Administration and Operation

1.3 Document History

Document

Version

Changes Date

1.0 First Edition 2011-11-23

1.1 Support Nexus 5548, 5596

Support Catalyst 3750X

Support BOBJ 4.0 (SBOP)

New hint in Chapter 7.8.9 ―Virtual Machine

Properties and ESXi Resources‖

2012-05-04

1.2 Update of chapter 3.3 ―Powering off the FlexFrame

Landscape‖

2012-07-04

1.4 Related Documents

FlexFrame™ for SAP® – Administration and Operation

FlexFrame™ for SAP® – HW Characteristics Quickguides

FlexFrame™ for SAP

® – Installation ACC 7.3

FlexFrame™ for SAP® – Installation Guide for SAP Solutions

FlexFrame™ for SAP® – Installation of a FlexFrame Environment

FlexFrame™ for SAP® – Management Tool

FlexFrame™ for SAP

® – myAMC.FA_Agents Installation and Administration

FlexFrame™ for SAP

® – myAMC.FA_Messenger Installation and Administration

FlexFrame™ for SAP

® – myAMC.FA_LogAgent Installation and Administration

FlexFrame™ for SAP® – Network Design and Configuration Guide

FlexFrame™ for SAP® – Security Guide

FlexFrame™ for SAP® – Technical White Paper

FlexFrame™ for SAP® – Upgrading FlexFrame 4.2B or 5.0A to 5.1A

ServerView Documentation

SUSE Linux Enterprise Server Documentation

1.5 Special Hints for FlexFrame 5.1A

In this document, you often will find console output, configuration data and installation

examples which are based on earlier FlexFrame versions. Please keep in mind that these

are examples and may look slightly different on the new operating systems introduced in

FlexFrame 5.1A.

The two Control Nodes (CN) of FlexFrame for SAP are also named as the FlexFrame

Control Center (CC). In this documentation the notation Control Node (CN) is used as a

synonym for Control Center (CC) and the other way round.

Special Hints for FlexFrame 5.1A Introduction

Administration and Operation 3

1.5.1 Incompatibilities (command scripts)

There are some changes with supplied scripts

The script ff_change_id.pl is not further available.

With the script ff_user_adm.pl resp. ff_group_adm.pl you can change UIDs

or GIDs of a specific user/group

The command line syntax of ff_setup_sid_folder.sh changed to

ff_setup_sid_folder.sh –s <SID> -p <pool>

With script f f _nas_adm. pl , the operation mode conf 10gb is no longer supported.

Use operation mode move instead.

Administration and Operation 5

2 FlexFrame Architecture

The FlexFrame solution is a revolutionary approach to run complex SAP solutions with

higher efficiency. At the same time some major changes of the configuration paradigms

for infrastructures have been implemented. These changes are:

A shared operating system booted via IP networks for the SAP Application Nodes.

Decoupling of application software and operating system, called virtualization of SAP

software or Adaptive Computing.

Shared Network Attached Storage from Network Appliance® providing Write

Anywhere File Layout (WAFL™) and sophisticated snap functionality.

Shared Network Attached Storage by EMC® Celerra.

FlexFrame Autonomous Agents (FA Agents) providing revolutionary mechanisms to

implement high-availability functions without cluster software

SAN storage

The concept of FlexFrame for SAP consists of several components, which implement

state-of-the-art functionality. Together with new components, such as the FlexFrame

Autonomous Agents, the whole solution is far more than just the sum of its components.

A major part of the benefits consist in a dramatic reduction in day to day operating costs

for SAP environments.

It is, of course, possible to use parts of the FlexFrame solution in project-based

implementations. However, they can not be called FlexFrame.

2.1 General Notes on FlexFrame

FlexFrame was designed and developed as a platform for SAP Applications. Its major

purpose is to simplify and abstract the basic components to enable the administrator of a

SAP system landscape to focus on SAP and not worry about servers, networking and

storage.

The FlexFrame Control Nodes are seen as an appliance. Like a toaster or a micro wave

oven, they have a well defined purpose and the build-components must work together to

achieve that purpose. It happened that Fujitsu picked SUSE® Linux Enterprise Server

(SLES) as the operating system for the Control Nodes; however it is not intended that the

customer is using it as a regular server, meaning installing additional software on it and

applying patches to it is not wanted, unless Fujitsu support line instructs so. Upcoming

versions of the Control Node's operating system may be totally different and may not

allow modifying anything at all. The installation and backup/restore functionality of a

Control Node are based on fixed images which are delivered on DVD. Modifications of

installed images will not be taken care of, if a new version is installed.

FlexFrame Architecture Hardware and Software

6 Administration and Operation

Modifications may even lead to errors which may be hard to find. Therefore we strongly

recommend to the customers not to install any software or patches onto the Control

Nodes without confirmation of Fujitsu support.

Similar to the Control Nodes are the Application Nodes. Fixed images are shipped for

installation of Linux Application Nodes in the FlexFrame landscape. Another aspect of

FlexFrame is the reduction of the TCO (Total Cost of Ownership). A static approach

(once started, never touched) will not be very efficient. To achieve the best savings, it is

recommended to actively manage where a certain SAP application instance or database

instance is running. If, as an example, a SAP instance requires the power of two CPUs of

an Application Node during most days of a month and eight CPUs during month-end

calculations, it is best to move it back and forth to the appropriate Application Node with

the right size. During the time where the application instance is running on the two CPU

Application Node, another SAP instance can use the bigger eight CPU Application Node

and therefore saving the need to have more eight CPU Application Nodes as in a static

approach.

2.2 Hardware and Software

The FlexFrame solution consists of both, hardware and software. To grant proper

function of the whole landscape, the entire software set is strictly defined. Anything other

than the software components listed below is not part of FlexFrame. This applies

unchanged if software from the list below is missing, is installed in other versions than

below, or if software other than the actual SAP components is added.

For detailed information about the hardware supported in a FlexFrame environment, see

the FlexFrame 5.1A Configuration Guide. Any other functions, such as backup, can be

implemented separately as an add-on to FlexFrame and need dedicated hardware,

operating system, high availability, professional service and support etc.

Hardware and Software FlexFrame Architecture

Administration and Operation 7

No. Hardware OS Software Services

1

Control Nodes:

2 x PRIMERGY RX300 S7 or

2 x PRIMERGY RX300 S6 or

2 x PRIMERGY RX300 S5 or

2 x PRIMERGY RX300 S4 or

2 x PRIMERGY RX300 S3

SLES 10 SP4 (x86_64)

FA Agents (Control Agents) V9.0,

FlexFrame 5.1A File System Image

CN,

ServerView etc.

TFTP, DHCP, LDAP,

(SAPROUTER), etc.

2

Network switches:

n*m Cisco Catalyst 3750 1 n 1, m 2

n*2 Cisco Nexus 50xx 1 n 0

n*2 Cisco Nexus 55xx 1 n 0

IOS (proprietary)

NX-OS (proprietary)

NX-OS (proprietary)

(as delivered)

3

Network Attached Storage:

one or more NetApp Filer heads (FASxxxx),

disk shelves as required 2

hosting shared OS file systems,

and application data

or

ONTAP 3

ONTAP 3

NetApp Tools

NFS, optional:

cluster components,

FlexClone,

SnapVault,

SnapMirror,

SnapRestore

one or more EMC Celerra NSxxx,

disk shelves as required 2,

hosting shared OS file systems,

and application data.

DART 3 DART

3

EMC Tools

NFS, …

1 allowed Types according to FlexFrame Support Matrix

2 The amount of disks required for customer-specific FlexFrame configurations can be determined together with Fujitsu's Customer Support Filer Sizing Team

3 supported versions according to FlexFrame Support Matrix

FlexFrame Architecture Hardware and Software

8 Administration and Operation

No. Hardware OS Software Services

4 SAN Storage Multipath SW

SLES 11 / SLES 10:

– DM-MPIO integrated Multipath SW

HA Services

5 SAN Storage Volume Manager

SLES 11 / SLES 10:

– LINUX Volume Manager LVM2

Volume Management

Services

6 Intel- or AMD-based PRIMERGY Servers

(standard rack or blade server)

SLES 11 SP2 (x86_64)

and / or

SLES 10 SP4 (x86_64)

FlexFrame 5.1A File System Image,

FA Agents (Application Agents),

SAP Applications, Database

SAP & DB Services

Administration and Operation 9

2.3 Shared Operating System

One major aspect of FlexFrame is its shared operating system. Sharing in this case

means that the very same files of essential parts of the underlying operating system are

used to run multiple Application Nodes. This part of the file system is mounted read-only,

so none of the Application Nodes that run the actual applications can modify it. Server

specific information is linked to a file system area that is server-specific and mounted

read-write. The shared operating system is kept to a NAS system (Network Attached

Storage system from Network Appliance or EMC Celerra).

2.3.1 Shared OS Boot Concept

The chart below shows the boot process of a FlexFrame Application Node

(PRIMERGY/Linux):

FlexFrame Architecture Shared Operating System

10 Administration and Operation

2.3.2 Control Nodes

A productive FlexFrame landscape always includes two Control Nodes. Their purpose is

to be a single point of control for the Application Nodes, as well as to check and manage

the proper function of the Application Nodes.

Control Nodes are not running SAP software (with the exception of saprouter, as an

option). They exclusively run SUSE Linux Enterprise Server Version 10 (SLES10, SP3),

installed on local disks. Control Nodes provide and run services such as:

Linux-HA high availablity cluster framework

Timeserver for the complete FlexFrame landscape

Control Agents

Web server to provide the Control Agents user interface

DHCP for assignment of IP addresses and TFTP for the boot process of the

Application Nodes

saprouter (optional)

Control Nodes have to be of the type PRIMERGY RX300 S7, RX300 S6, RX300 S5,

RX300 S4 or RX300 S3.

2.3.3 Application Nodes

Application Nodes run database and SAP services on the SUSE Linux Enterprise Server

shared operating system. Application Nodes can be physical servers that offer CPU and

memory, or virtual servers built on top of physical servers, using the physical server's

CPU and memory through a virtualization layer.

For FlexFrame Version 5.1A, the principal types of Application Nodes are PRIMERGY

servers running Linux directly and virtual servers on top of PRIMERGY servers using

VMware ESXi as a virtualization layer.

Admissible PRIMERGY servers have to be approved for SAP on Linux by Fujitsu.

During the boot process using Intel's PXE® technology, each Application Node will be

identified using the hardware address of its boot interface (MAC address). The Control

Node will assign an IP address to it and supply the operating system via the network.

File systems, especially the root file system (/), are mounted via the network in read-only

mode. If, for any reason, an Application Node needs to be replaced or added, only a

handful of settings need to be adjusted to integrate it into the FlexFrame landscape.

Intel's PXE technology is implemented in Fujitsu PRIMERGY servers and allows booting

via the network. DHCP will be used with static MAC address relationship for all the

Application Nodes.

FlexFrame Structure in LDAP FlexFrame Architecture

Administration and Operation 11

2.4 FlexFrame Structure in LDAP

LDAP is used as the central information service for all shared OS nodes within a

FlexFrame environment. The Control Nodes are used as LDAP servers. The LDAP

database is located on shared file systems mounted from the NAS storage. The

Application Nodes are configured as LDAP clients. LDAP requests from Application

Nodes are restricted to the data of their own pool.

LDAP provides host-related network information such as:

net boot

automount

user authentication

groups

host names and IP addresses

shared services

networks and netmasks

LDAP client profiles

The FlexFrame LDAP tree roughly looks as illustrated here:

FlexFrame Architecture Linux-HA Cluster on Control Center

12 Administration and Operation

Additional information about configuration data is only applicable for Control Nodes.

These are used FlexFrame-internally to add, remove or modify the configuration of

Application Nodes or SAP services.

FlexFrame utilizes LDAP for two different purposes:

(1) for operating naming services (such as host name resolution, user/password

retrieval, tcp/udp service lookup, etc.) and

(2) for storing FlexFrame specific data on the structure of the installed environment

Application Nodes are only able to search in area (1). It is separated into pool specific

sections in order to protect pools from accessing other pools' data. Each of them contains

pool specific network information service (NIS) like data. The LDAP servers have access

lists to prevent searches outside of the own pool.

The other main DIT part contains FlexFrame configuration data (2). It should only be

accessed through maintenance tools from one of the Control Nodes. This part of the DIT

contains a lot of cross references, which need to be kept in sync. Do not try to change

this data, unless you are explicitly instructed by Fujitsu support to do so.

2.4.1 Working with LDAP

The usage of LDAP specific commands like ldapadd or ldapmodify is limited to very

few actions. One is to create or remove a PUT service for a SAP system copy. This

action is described within the ―Installation Guide for SAP Solutions‖ manual. Other direct

interaction through LDAP commands is limited to service issues.

No other interventions have to and should be done. The FlexFrame maintenance tools

provide the necessary functionality.

2.5 Linux-HA Cluster on Control Center

2.5.1 Terminology

Some of the Linux-HA terms used below are explained here in order to promote broader-

based understanding:

Node

Every computer that is part of a cluster is a node

Resource

Everything than can be administered by heartbeat is referred to as a resource. For

example, an IP address that is administered by the cluster is a resource.

Linux-HA Cluster on Control Center FlexFrame Architecture

Administration and Operation 13

Resource Agent (RA)

The RA is the connection between heartbeat and the programs that are started

when the RA is called. They are shell scripts, which have to provide a standardized

interface to heartbeat, so that they can be started, monitored and stopped by

heartbeat.

Supported standards:

● Linux Standard Base (LSB)

All scripts under /etc/init.d correspond to this standard

● Open Cluster Framework (OCF)

For examples see /usr/lib/ocf/resource.d

● Stonith

Designated Coordinator (DC)

Every cluster has precisely one DC as the central instance in the cluster. It is elected

from all the nodes. It alone is responsible for all the actions in the cluster and has the

only valid cluster information base. All the other nodes only have a duplicate.

STONITH

STONITH is an abbreviation for "Shoot The Other Node In The Head". This is the

name of a method that stops any nodes no longer accessible from "causing

damage" in the cluster by switching it off or alternatively causing a reboot.

heartbeat

The cluster can be started and stopped with the /etc/init.d/heartbeat script.

Cluster Resource Manager (CRM)

The CRM manages the resources, decides which resource is to run where and

ensures that the required status of the cluster is achieved based on the current

status. For this purpose the CRM distributes the work to the LRM and receives

feedback from the latter.

Local Resource Manager (LRM)

The LRM manages the resources on "its" local nodes. The CRM tells the LRM which

resources they are. The LRM communicates with the CRM.

Cluster Information Base (CIB)

The CIB is the central information file for all the resources and nodes of the cluster.

It not only contains information about the static configuration but also about a

dynamic part of the current status of all resources. The data is stored in XML format.

FlexFrame Architecture Linux-HA Cluster on Control Center

14 Administration and Operation

ha.cf

This configuration file controls the behavior of the cluster software. It is needed to

start heartbeat. For example, it defines the interface via which communication is to

take place in the cluster and the nodes which belong to the cluster. It must be

available on every node under /etc/ha.d.

authkeys

The file authkeys is used to authenticate the nodes of a cluster for one another. This

file must also be available on every node under /etc/ha.d. Its access mode

should be set to 600.

2.5.2 Simple Resources

2.5.2.1 Constraints

Simple resources can be linked with constraints, which are triggered when the resource is

started. These are:

Ordering:

specifies the relationship of two resources to each other. This means that you can

only start resource B after resource A.

Collocation:

ensures that two resources are started on the same node.

Location:

defines the node where a resource is to preferably run.

2.5.2.2 Score

Constraints are defined by rules. A system of points is used to decide on which node a

resource is to run. Each applicable rule is linked to a score. In addition to normal inte-

gers, it can also accept special values –INFINITY and INFINITY.

Example:

Resource A is to preferrably run on CN1.

The rule: NODE eq CN1

Score: 10

If A would run on CN1, the rule would return ―true‖ and CN1 would get the score 10.

If A would run on CN2, the rule would return ―false‖ and CN2 would not get any score,

thus 0.

After all possibilities being evaluated by the cluster, CN1 has a greater score than CN2,

thus CN1 would be chosen to run A.

Linux-HA Cluster on Control Center FlexFrame Architecture

Administration and Operation 15

The extremes:

INFINITY: The resource is started in any case on the node selected by an applicable rule.

If it cannot be started on the specified node, an attempt is made to start it on another

node.

-INFINITY: The resource is not started on the node selected by an applicable rule. If it

cannot be started on any other node than the selected one, it remains stopped.

In other words:

INFINITY ~ is applicable if possible

-INFINITY ~ is never applicable

2.5.2.3 Stickiness

Another value that plays an important role in the decision about the resource where the

node is to run is stickiness. You can regard this value as the level of adhesion, with

which a resource aims to remain on the node on which it is currently running. This value is defined via the global settings of the CIB by the variable default-resource-

stickiness. The following values are possible.

0 :

Linux-HA attempts to optimally place the resource in the cluster.

In other words, the resource is redistributed in the cluster if Linux-HA finds a more

"suitable" node.

> 0 :

Linux-HA attempts to optimally place the resource in the cluster.

In other words, the resource is redistributed in the cluster if Linux-HA finds a more

"suitable" node.

< 0 :

The resource tends to leave the current node. Lower values reinforce this tendency.

INFINITY :

The resources remain on the current node until they are compelled to leave it, e.g.

when their node is shut down.

-INFINITY :

The resources do not want to remain on the current node.

The two values, score and stickiness, determine the node on which a resource is then

actually started. For this purpose, they are determined and added for each node of the

cluster. The resource is then started on the node that has the highest score.

FlexFrame Architecture Linux-HA Cluster on Control Center

16 Administration and Operation

Example:

Default_resource_stickiness : 100

Resource A is to be preferably run on CN1

Rule: NOTE eq CN1

Score: 10

Case 1: A already runs on CN1

Points for CN1 :

Location score =10

Stickiness score = 100

Resulting score CN1 =110

Points for CN2 :

Location score = 0 (rule not met)

Stickiness Score = 0 (because it is currently not running on CN2)

Resulting score CN2 = 0

Result: A remains on CN1, score CN1 > score CN2

Case 2: A already runs on CN2

Points for CN1 :

Location score = 10

Stickiness score = 0 (as not running on CN2)

Resulting score CN1 = 10

Points for CN2 :

Location score = 0

Stickiness score = 100

Resulting score CN2 = 100

Result: As score CN2 > score CN1, resource A remains on CN2, although its location

score expresses the wish to move to CN1.

Linux-HA Cluster on Control Center FlexFrame Architecture

Administration and Operation 17

Case 3: A does not run on any node (e.g. during cluster start)

Points for CN1 :

Location score =10

Stickiness score = 0

Resulting score CN1 = 10

Points for CN2 :

Location score = 0 (rule not met)

Stickiness score = 0

Resulting score CN2 = 0

Result: A starts on CN1, score CN1 > score CN2

In FlexFrame almost all the resources are configured in such a way that they remain on

their current node after a move, even when the other control node is available again after

a downtime. This avoids unnecessary switching processes.

2.5.2.4 Resource Groups

Simple resources can be put together to form groups. It is also possible to create so-

called clones, in which simple resources may run repeatedly in the cluster. Constraints,

such as ordering or colocation, which are defined for the group, apply for all the re-

sources belonging to the group.

2.5.3 FlexFrame Specific Configuration

The basic configuration for heartbeat is created during the initial configuration of

FlexFrame, i.e. the file ha.cf is set up, a key to authenticate the nodes CN1 and CN2 is

created for the file authkeys and the CIB is created.

FlexFrame Architecture Linux-HA Cluster on Control Center

18 Administration and Operation

It contains the following resources (type and name of the RA in brackets):

Resource Group: network_CN1_server

ip_CN1_server_<poolname-m> (ocf::heartbeat:IPaddr2)

Resource Group: network_CN2_server

ip_CN2_server_<poolname-m> (ocf::heartbeat:IPaddr2)

ldap_master (ocf::fsc:ff_ha_ldap)

ldap_replica (ocf::fsc:ff_ha_ldap)

slurpd (lsb:ff_ha_slurpd)

Resource Group: netboot

dhcpd (lsb:ff_ha_dhcpd)

tftpd (lsb:ff_ha_tftpd)

Resource Group: ff_manage)

mysql (lsb:ff_ha_mysql)

myAMC.FA_Messenger (lsb:ff_ha_myAMC.MessengerSrv)

tomcat (lsb:ff_ha_tomcat)

myAMC.FA_CtrlAgent (lsb:ff_ha_myAMC.CtrlAgent)

Clone Set: clone_ClusterMon

ClusterMon:0 (ocf::heartbeat:ClusterMon)

ClusterMon:1 (ocf::heartbeat:ClusterMon)

stonith_ipmi_CN1 (stonith:external/ipmi

stonith_ipmi_CN2 (stonith:external/ipmi)

Clone Set: clone_stonith_meatware

Stonith_meatware:0 (stonith:meatware)

Stonith_meatware:1 (stonith:meatware)

Linux-HA Cluster on Control Center FlexFrame Architecture

Administration and Operation 19

2.5.3.1 Configuration of Limits for Cluster Resources

In some large FlexFrame environments the default number of open files

(1024) is not sufficient to solve all requests to ldap or tomcat.

/var/log/messages show entries like:

... slapd[2393]: warning: cannot open /etc/hosts.allow: Too many open

files

... java.rmi.RemoteException: VI SDK invoke

exception:java.net.SocketException: Too many open files

In such a case the number of open files has to be increased before the

cluster resources are (re)started.

The problem only occurs in very large configurations (many pools with many Application

Nodes and a lot of big SAP applications with many active users permanently logging in).

Therefore instead of system-wide increasing of the default number of open files a

FlexFrame specific configuration template is provided. This template can be used to

create a custom specific file limiting the number of open files.

The mentioned template is the

/opt/FlexFrame/etc/ff_ha_settings.conf.template file. You can use it to

create your own configuration file with the settings suitable for your system:

At first copy the template file to /opt/FlexFrame/etc/ff_ha_settings.conf:

control1:~ # cp /opt/FlexFrame/etc/ff_ha_settings.conf.template

/opt/FlexFrame/etc/ff_ha_settings.conf

Now uncomment the line with the 'FF_ULIMIT_SETTINGS' variable. The example

below shows an entry to increase the number of open files up to 8192 (8K):

## Template for config file /opt/FlexFrame/etc/ff_ha_settings.conf

## Used to set special values for our linux ha resources like ldap or

tomcat

## The resource scripts /opt/FlexFrame/lib/ha/scripts/ff_ha_* will look

## for active variables in the config file

##

## Copy the template file to /opt/FlexFrame/etc/ff_ha_settings.conf and

## uncomment the needed variables

## increase number of open files for resource ldap and tomcat

FF_ULIMIT_SETTINGS="-n 8192"

FlexFrame Architecture Linux-HA Cluster on Control Center

20 Administration and Operation

Finally the resources tomcat and ldap_master have to be restarted, if they are

already running.

control1:~ # ff_ha_cmd.sh restart tomcat

control1:~ # ff_ha_cmd.sh restart ldap_master

You may verify the results of the changes by checking the limits before (should be 1024)

and after (should now be 8192) the restart of the resources:

cat /proc/<pid_of_tomcat>/limits

cat /proc/<pid_of_ldap>/limits

The configuration file /opt/FlexFrame/etc/ff_ha_settings.conf will be backed

up automatically by ff_cn_backup.sh during a FlexFrame upgrade, but has to be

restored manually after the FlexFrame upgrade has finished.

2.5.3.2 LDAP Configuration

The FlexFrame LDAP concept is implemented by the resource groups net-

work_CN1_server and network_CN2_server as well as the simple resources

ldap_master, ldap_replica and slurpd. Since this concept has damatically

changed with the introduction of Linux-HA, it is explained in more detail here.

Each network_CN<n>_server (n=1..2) group consists of the simple resources

network_CN<n>_server_<poolname-m> (n=1..2)4. Each of these simple

resources is exactly one server LAN IP of a pool, i.e. the appropriate resource agent

(IPaddr2) accurately monitors a server LAN IP and, in the event of a fault, moves it to the

surviving node of the cluster. Accordingly, there are as many resources in the cluster as

there are server LAN IPs.

This ensures that the entire server LAN IPs of all the pools can always be accessed in

the network.

The resource ldap_master is initially started on the first control node. It starts the LDAP

server process slapd on the same control node with the server LAN IPs of all the pools

of the first control node as defined by the management tool. Analog to this, the resource

ldap_replica on the second control node is started with the appropriate server LAN

IPs of all the pools of the second control node.

4 "n" is an enumeration consisting of 2 elements. Element 1 for CN1 and element 2 for CN2.

Linux-HA Cluster on Control Center FlexFrame Architecture

Administration and Operation 21

By applying the constraint "Colocation" (see above) the resource slurpd is forced to

always start on the node, on which the resource ldap_master could also be started.

This concept ensures that:

1. ldap_master und ldap_replica can always access precisely defined IP

addresses, i.e. the address of the server LAN

2. these IP addresses are managed by the cluster and are thus always available on a

cluster-wide basis

3. ldap_master und ldap_replica can also run in parallel on a node.

In other words, if a node fails, the cluster ensures that its IP addresses and the

appropriate ldap resource are switched over to the other node and are then available

again on a system-wide basis

Example:

2 Pools, pool1, pool2:

Server LAN IPs :

cn1-pool1-se 192.168.12.12

cn1-pool2-se 192.168.102.16

cn2-pool1-se 192.168.12.13

cn2-pool2-se 192.168.102.17

Ressource group network_CN1_server contains the two simple resources:

network_CN1_server_pool1 manages 192.168.12.12

network_CN1_server_pool2 manages 192.168.102.16

Ressource group network_CN2_server with

network_CN2_server_pool1 manages 192.168.12.13

network_CN2_server_pool2 manages 192.168.102.17

The resource ldap_master starts on CN1 with the IP addresses 192.168.12.12 and

192.168.102.16

The resource ldap_replica starts on CN2 with the IP addresses 192.168.12.13 and

192.168.102.17

FlexFrame Architecture Linux-HA Cluster on Control Center

22 Administration and Operation

Since ldap_master could be started on CN1, the resource slurpd is on account of the

"colocation" constraint also started on the node CN1.

In comparison with other resources and resource groups the rule for the "location" con-

straint of the resource groups network_CN<n>_server (n=1..2) is given a score

(=100000) higher than the default-resource-stickiness value (=100). After a

move to the other nodes (e.g. after a control node failure or reboot) this causes the server

LAN and LDAP resources to return to their original nodes as soon as possible.

2.5.3.3 Ressource ff_manage

The resource ff_manage is a group of simple resources that is initially started on the

first control node. These are:

mysql

myAMC.FA_Messenger

tomcat

myAMC.FA_CtrlAgent

Prerequisite for the start of myAMC.FA_messenger is the successful start of the resource

mysql. This is compelled by the constraint "orders".

Since the start of ff_manage can take longer, an attempt is made as a result of an "or-

ders" constraint to start the resources ldap_master and ldap_replica before the

start of the group ff_manage.

2.5.3.4 Resource Netbooot

The resource group netboot contains the simple resources.

dhcp

tftpd

An attempt is also made for this group – as for the group ff_manage – to only start it if

the resources ldap_master and ldap_replica run successfully. The "orders"

constraint is also used for this.

2.5.3.5 STONITH

The stonith agents' purpose is to prevent the cluster entering a non-defined status in the

event of a fault in communications. The corresponding stonith resources are.

stonith_ipmi_CN2

stonith_ipmi_CN1

as well as the clone

Linux-HA Cluster on Control Center FlexFrame Architecture

Administration and Operation 23

stonith_meatware:0

stonith_meatware:1

Due to the "location" constraint stonith_ipmi_CN2 may – under no circumstances –

be started on CN2 and stonith_ipmi_CN1 may not be started on CN1.

These resources monitor the other node respectively. If, after a time interval set during

configuration, a node finds that communication with the other node is no longer possible,

a hard "reboot" or a "reset" is triggered on the inaccessible node via the IPMI interface.

The IP addresses of the IPMI interface as well as user and password were transferred to

the resource during configuration in order to enable this step.

For the reboot to take place via the ipmi interface the latter must be accessible. Therefore

stonith_ipmi_CN<n> (n=1..2) monitors the respective interface every 30 seconds. If it

is determined that the interface does not answer (e.g. in the event of a power failure), stonith_ipmi_CN<n> is ended and stonith_meatware:<n> is started. This re-

source communicates with the operator: it creates a message as to which node is af-

fected and requests it to restart the node. When this is done, feedback is expected from

the operator. The message about the affected node and the command to provide feed-back after the start are written in /var/log/ha-log.

Example after power failure of CN2 :

Using the command crm_mon -1 -r -f -n provides an overview of the error counters

and inactive resources:

control1:~ # crm_mon -1 –r –f –n

Inactive resources:

stonith_ipmi_CN2 (stonith:external/ipmi): Stopped

Clone Set: clone_stonith_manual

stonith_meatware:0 (stonith:meatware): Started cn1

stonith_meatware:1 (stonith:meatware): Stopped

Failcount summary:

* Node cn2:

* Node cn1:

stonith_ipmi_CN2: fail-count=1000000

Failed actions:

stonith_ipmi_CN2_start_0 (node=cn1, call=699, rc=1): complete

FlexFrame Architecture Linux-HA Cluster on Control Center

24 Administration and Operation

The following entry is written in the log file /var/log/ha-log :

stonithd[6462]: 2009/05/28_10:47:46 CRIT: OPERATOR INTERVENTION REQUIRED to reset

cn2.

stonithd[6462]: 2009/05/28_10:47:46 CRIT: Run "meatclient -c cn2" AFTER power-

cycling the machine.

The operator is thus requested to execute the command meatclient –c –cn2 on CN1

after ensuring that CN2 has been completely switched off or that CN2 has been manually

"reset" or restarted.

Otherwise, no further actions are performed on the resources by the cluster to avoid data

loss; the cluster does not know its status due to the lack of a connection to the partner.

control1:~ # meatclient -c cn2

WARNING!

If node "cn2" has not been manually power-cycled or disconnected from all shared

resources and networks, data on shared disks may become corrupted and migrated

services might not work as expected.

Please verify that the name or address above corresponds to the node you just

rebooted.

PROCEED? [yN] y

Meatware_client: reset confirmed

The process which asks the operator to run the meatclient tool is respawned

every 30 seconds.

If this respawn occurs after the operator started the meatclient tool and be-

fore he entered "yes", the request will be ignored. Therefore it may be re-

quired to call it several times and confirm it with a reasonable small delay.

2.5.3.6 Configuring the Cluster

The cluster is automatically configured by the script ff_post_conf.sh during the instal-

lation of FlexFrame. If, in certain situations, it is necessary to subsequently configure the

cluster, the command ff_ha_tool.sh [-f] -i must be used. During the setup the

IPMI user and IPMI password are queried if they are not yet known or access is not pos-

sible with the known user and password.

Syntax :

ff_ha_tool.sh [-f] -i

-i: initial configuration of linux-ha for FlexFrame

Linux-HA Cluster on Control Center FlexFrame Architecture

Administration and Operation 25

-f: force execution, purge old configuration

2.5.3.7 Starting Behavior of the Cluster

The cluster is started:

automatically when FlexFrame is installed via the installation scripts

with each reboot

with the command /etc/init.d/hearbeat start

Two cases should be observed both when rebooting and during a manual start:

4. One node is affected – either it has failed or the service heartbeat was stopped

manually. The second node subsequently took on the resources and has the role of

the DC.

5. Both nodes are affected.

In case 1 there is still nothing to be considered. After the node is online again or the clus-

ter could be successfully started on this node, the resources are redistributed according

to their score values and started.

If the cluster has to be restarted on both nodes (case 2) – either through rebooting or

manually – the following must be taken into consideration:

During the start of a node the service heartbeat attempts to communicate with the other

nodes. If there is no answer after a configurable time interval, the stonith agent is

activated, which via the IPMI interface then causes the second node to reboot. The time interval is configured in the file ha.cf by the parameter initdead and is currently 60

seconds.

Particularly when switching on an entire FlexFrame environment it is

essential to ensure that both control nodes have to be switched on very

quickly one after each other. If the delay is too great, the control node

switched on last may be reset hard by the stonith agent, but this usually has

no negative effects.

If in this situation the IPMI interface of the other node is not accessible, the stonith IPMI

agent would - after a vain attempt to trigger a reboot on the node - activate the stonith

meatware agent. Then the operator must proceed as described in the section "STONITH"

so that the resources monitored by the cluster can be started.

2.5.3.8 Status Information of the Clusters

Various options can be used to find out the current status of the cluster. One of which is implemented by the resource agent ClusterMon. It is the task of this RA to regularly

transfer the current status of the cluster to a HTML file. For this purpose, the clones

FlexFrame Architecture Linux-HA Cluster on Control Center

26 Administration and Operation

ClusterMon:0 and ClusterMon:1 are configured which are each started on a

control node. They write the status in /srv/www/htdocs/ClusterMon.html, which can be

accessed via the local web server at http://localhost/ClusterMon.html and is also linked

from the default homepage of the web server.

2.5.4 Linux-HA CLI Commands

Various options are available to achieve a required action. For this purpose, FlexFrame

provides a simple CLI interface to Linux-HA. The command ff_ha_cmd.sh, which is

internally based on the Linux-HA CLI commands, was provided for the most important ac-

tions, such as status display, starting and stopping a ressource.

Action Command

Display the status of all resources ff_ha_cmd.sh status

Display the status of one resource ff_ha_cmd.sh status <resource>

Start a resource ff_ha_cmd.sh start <resource>

Stop a resource ff_ha_cmd.sh stop <resource>

Migrate (move) a resource to another

node – this creates a special location

constraint

ff_ha_cmd.sh migrate <resource>

<node>

Undo the migration – the special

location constraint is deleted

ff_ha_cmd.sh unmigrate <resource>

Cleanup a resource, delete all status

flags – this will restart the resource if

possible

ff_ha_cmd.sh cleanup <resource>

Remove a resource from the

administration using heartbeat.

Resources will not be automatically

started, stopped or restarted

ff_ha_cmd.sh unmanage <resource>

Reintegrate a resource in the

administration using heartbeat

ff_ha_cmf.sh manage <resource>

Output the CIB (XML format) cibadmin -Q

List all the resources of the cluster crm_resource -L

crm_resource -l

List a resource (XML format) crm_resource -x -r <resource>

Network FlexFrame Architecture

Administration and Operation 27

Action Command

Output the node, on which a resource

runs

crm_resource -W -r <resource>

ff_ha_cmd.sh status resource

Regular status display (interval: 15

sec)

crm_mon

Regular status display, interval <n>

sec

crm_mon -i <n>

Once-only status display crm_mon -1

[Once-only] status display with output

of the error counters of all resources

as well as all offline resources

crm_mon [-1] -f -r

As above, but grouped by nodes crm_mon [-1] -f -r -n

If a resource is to be migrated to another node, it is essential that this action creates a

"Location" constraint with the score INFINITY. However, this means that this resource

also remains permanently on this node even after a complete restart of the cluster and is

therefore not moved – unless the node has a problem to the effect that no resources can

run any more. This constraint can be removed with the command ff_ha_cmd.sh un-

migrate <resource>, which nevertheless does not cause the resource to automati-

cally move back; it merely removes the constraint of the preferred node as a result.

2.5.5 Linux-HA Logfile

All messages that concern Linux-HA are written in the log file /var/log/ha-log .

2.5.6 Linux-HA GUI

Linux-HA has its own GUI which can be launched using the "hb_gui" command. However, it is strongly recommended not to use this interface to change configuration parameters or control resources, because each change hides the risk of negatively influencing or even severely damaging the FlexFrame cluster configuration. There also may be situations where it is not possible to use the command line interface after having used the "hb_gui" interface. Please use the command line interface ff_ha_cmd.sh exclusively to control resources.

2.6 Network

The network is the backbone of the FlexFrame solution. Communication between the

various nodes and storage devices is done exclusively via the IP network infrastructure. It

FlexFrame Architecture Network

28 Administration and Operation

serves both, communication between servers and clients as well as delivering IO data

blocks between the NAS (Network Attached Storage) and the servers.

The IP network infrastructure is essential for every FlexFrame configuration. FlexFrame is

designed with a dedicated network for connections between servers and storage that is

reserved for FlexFrame traffic only. One network segment, the Client LAN (see below)

can be routed outside the FlexFrame network to connect to the existing network.

2.6.1 LAN Failover

The term „LAN failover― describes the ability of a FlexFrame environment to use a logical

network interface that consists of several physical network interface cards (NICs), which

in turn are using redundant network paths (cables and switches). When a network

component (NIC, cable, switch, etc.) fails, the network management logic will switch over

to another network interface card and path.

2.6.2 Segments

FlexFrame uses a network concept providing high availability as well as increased

flexibility in virtualizing the whole FlexFrame landscape.

The FlexFrame network concept relies on VLAN technology that allows running multiple

virtual networks across a single physical network. Additionally, in order to ensure high

network availability, LAN bonding is used on every node. This includes a double switch

and wiring infrastructure, to keep the whole environment working, even when a network

switch or cable fails. Within FlexFrame there are four virtual networks. These networks

run through one logical redundant NIC, using bonding on Linux.

Client LAN

The purpose of the Client LAN segment is to have dedicated user connectivity to the

SAP instances. This segment also allows administrators to access the Control

Nodes.

Control LAN

The Control LAN segment carries all administrative communication, such as IPMI or

similar interfaces.

Server LAN

The Server LAN segment is used for the communication between SAP instances

among each other and the databases.

Storage LAN

The Storage LAN segment is dedicated to NFS communication for accessing the

Application Nodes' shared operating systems, the executables of SAP and the

RDBMS as well as the IO of the database content and SAP instances.

Network FlexFrame Architecture

Administration and Operation 29

The following figure outlines the basic network segments of a typical FlexFrame

landscape with Application Nodes.

2.6.3 Network Switches

Network switching components play a very important role within FlexFrame. Therefore,

only designated switch types are tested and supported (see FlexFrame Support Matrix

https://partners.ts.fujitsu.com/com/products/infrastruc-solutions/FlexFrame).

All supported switch models support VLAN technology for a flexible configuration for the

various network segments.

Switches of the Cisco Catalyst 3750-E Series are only allowed to be used with 2 X2 10

Gigabit Ethernet uplinks. If you want to use TwinGig SFP Converter Modules you have to

file a request for special release and add the switches to the FlexFrame configuration

using the FlexFrame switch type cat3750g-24ts resp. cat3750g-48ts.

Ports of Cisco Nexus switches are only allowed to be connected to 10GbE Ports of other

devices with the exception of 1GbE SFP ports of 3750g models or 3750x 1GbE module

for uplink purposes which are allowed to be connected to dual speed ports.

FlexFrame Architecture Network

30 Administration and Operation

2.6.4 Network Speed

FlexFrame supports network connections for data communication with the following

network speeds (see also: FlexFrame™ for SAP® – Network Design and Configuration

Guide):

1Gbit/sec (1GbE)

10Gbit/sec (10GbE)

To determine the 10GbE use case for end systems which are able for both, 1GbE or

10GbE network connection, use the --10gbit option when adding the device to the

FlexFrame configuration.

For further details about supported end systems see the FlexFrame Support Matrix.

2.6.5 Network Switch Groups

In FlexFrame network switches are grouped. Ports of an end system are building a

redundant connection and are connected to ports of different members of the same

switch group (see also: FlexFrame™ for SAP® – Network Design and Configuration

Guide).

The switch groups within the entire FlexFrame environment are numbered starting with 1.

Also the members within a switch group are numbered starting with 1 in each case. In

special cases the use of switch group number 0 (zero) is allowed. This switch group is

neither registered in nor managed from FlexFrame. It helps to identify network devices

outside but with relation to the FlexFrame environment (see also next section Network

Switch Ports).

2.6.6 Network Switch Ports

In FlexFrame network switch ports are uniquely identified within the entire FlexFrame

environment by:

the number of the switch group

the number of the member within the switch group

the port ID within the switch

The portID is

a single number as for Cisco Catalyst 3750 switch fixed ports

a prefixed number

o <slot number>/<port number> as for Cisco Nexus switch ports

o G1 ... G4 as for Cisco Catalyst 3750x 1GbE module

o TE1, TE2 as for Cisco Catalyst 3750x 10GbE module

Network FlexFrame Architecture

Administration and Operation 31

In case of Cisco Catalyst 3750e the numbers 1 and 2 are used both for GigabitEthernet

Ports and TenGigabitEthernet Ports. To distinguish both cases a --10gbit Option is used

on input and on output (10G) is appended.

Switch ports belonging to the unspecific switch group 0 (zero) are meant to be

somewhere outside the FlexFrame environment.

2.6.7 Automounter Concept

The Automounter Concept is based on the ability of Linux to mount file systems

automatically when their mount points are accessed.

During the boot process of an Application Node some file systems are mounted. For Linux these are the root file system (read-only) as well as the /var mount point (read-

write). These are the basic file systems which must be accessible for the Application

Node to function properly. There is no data in the two file systems that is specific for a

SAP service.

Data which is specific for a SAP service, a database or a FlexFrame Autonomous Agent

is found in directories which are mounted on first access. Some of the mounts will stay as

long as the Application Node is operational. Others will be unmounted again, if directories

or files below that mount point have not been accessed for a certain period of time.

Within the LDAP database there are two types of data which relate to the automounter

configuration: automountMap and automount.

An automountMap is a base for automount objects. Here's how to list the

automountMaps:

control1:~ # ldapsearch -x -LLL '(objectClass=automountMap)'

dn: ou=auto.FlexFrame,ou=Automount,ou=pool2,ou=Pools,ou=FlexFrame,

dc=flexframe,dc=wdf,dc=fujitsu,dc=com

objectClass: top

objectClass: automountMap

ou: auto.FlexFrame

...

The base directory looks like this:

dn: cn=/FlexFrame,ou=auto_master,ou=Automount,ou=pool1,ou=Pools,

ou=FlexFrame,dc=flexframe,dc=wdf,dc=fujitsu,dc=com

objectClass: top

objectClass: automount

cn: /FlexFrame

automountInformation: auto_FlexFrame

Further on there are entries like:

FlexFrame Architecture Network

32 Administration and Operation

dn: cn=myAMC,ou=auto_FlexFrame,ou=Automount,ou=pool1,ou=Pools,

ou=FlexFrame,dc=flexframe,dc=wdf,dc=fujitsu,dc=com

objectClass: top

objectClass: automount

cn: myAMC

automountInformation:

-rw,nointr,hard,rsize=32768,wsize=32768,proto=tcp,nolock,

vers=3 filpool1-st:/vol/volFF/pool-pool1/pooldata/&

There are two things that have to be pointed out.

First, the ou=auto.FlexFrame refers to the base directory as shown before.

The second notable aspect in this entry is the use of the wildcard &. If the folder

/FlexFrame/myAMC is accessed, the autofs process tries to mount it from the path

filpool1-st:/vol/volFF/pool-pool1/pooldata/myAMC. If the folder myAMC is

found and permissions allow the clients to access it, it will be mounted to

/FlexFrame/myAMC/<name>. If myAMC is not found or the client does not have the

permissions, the folder will not be mounted. In such a case, try to mount the folder

manually on a different folder, e.g. like:

an_linux:~ # mount filpool1-st:/vol/volFF/pool-pool1/pooldata/myAMC /mnt

If you get an error message like Permission denied, check the exports on the NAS

system and the existence of the directory myAMC/ itself.

Other entries in LDAP make use of platform specifics. With Linux you can find a number of variables like ${OSNAME}/${OSDIST}/${ARCH} to make a distinction between

different platforms.

dn: cn=/,ou=auto.oracle,ou=Automount,ou=pool2,ou=Pools,

ou=FlexFrame,dc=flexframe,dc=wdf,dc=fujitsu,dc=com

objectClass: top

objectClass: automount

cn: /

description: catch-all for Linux automount

automountInformation:

-rw,nointr,hard,rsize=32768,wsize=32768,proto=tcp,nolock,

vers=3 filpool2-st:/vol/volFF/pool-pool2/oracle/${OSNAME}/${OSDIST}/${ARCH}/&

On Linux, the automount mount points can be read using the following command:

an_linux:~ # mount

rootfs on / type rootfs (rw)

/dev/root on / type nfs (ro,v3,rsize=32768,wsize=32768,reserved,

hard,intr,tcp,nolock,addr=192.168.11.206)

192.168.11.206:/vol/volFF/os/Linux/FSC_4.2A00-000.SLES-9.X86_64/var_img/var-

c0a80b36

Network FlexFrame Architecture

Administration and Operation 33

on /var type nfs (rw,v3,rsize=32768,wsize=32768,reserved,

hard,intr,tcp,nolock,addr=192.168.11.206)

192.168.11.206:/vol/volFF/os/Linux/FSC_4.2A00-000.SLES-9.X86_64/var_img/var-

c0a80b36

/dev on /dev type nfs (rw,v3,rsize=32768,wsize=32768,reserved,

hard,intr,tcp,nolock,addr=192.168.11.206)

192.168.11.206:/vol/volFF/os/Linux/pool_img/pool-c0a80bff on /pool_img type nfs

(rw,v3,rsize=32768,wsize=32768,reserved,

hard,intr,tcp,nolock,addr=192.168.11.206)

proc on /proc type proc (rw)

devpts on /dev/pts type devpts (rw)

shmfs on /dev/shm type shm (rw)

/dev/ram on /var/agentx type ext2 (rw)

automount(pid1750) on /FlexFrame type autofs (rw)

automount(pid1772) on /saplog/mirrlogA type autofs (rw)

automount(pid1752) on /home_sap type autofs (rw)

automount(pid1788) on /saplog/saplog1 type autofs (rw)

automount(pid1766) on /sapdata/sapdata5 type autofs (rw)

automount(pid1778) on /saplog/origlogA type autofs (rw)

automount(pid1762) on /sapdata/sapdata3 type autofs (rw)

automount(pid1758) on /sapdata/sapdata1 type autofs (rw)

automount(pid1784) on /saplog/saparch type autofs (rw)

automount(pid1764) on /sapdata/sapdata4 type autofs (rw)

automount(pid1786) on /saplog/sapbackup type autofs (rw)

automount(pid1760) on /sapdata/sapdata2 type autofs (rw)

automount(pid1754) on /myAMC type autofs (rw)

automount(pid1796) on /usr/sap type autofs (rw)

automount(pid1780) on /saplog/origlogB type autofs (rw)

automount(pid1768) on /sapdata/sapdata6 type autofs (rw)

automount(pid1792) on /saplog/sapreorg type autofs (rw)

automount(pid1776) on /saplog/oraarch type autofs (rw)

automount(pid1774) on /saplog/mirrlogB type autofs (rw)

automount(pid1770) on /sapdb type autofs (rw)

automount(pid1756) on /oracle type autofs (rw)

automount(pid1790) on /saplog/saplog2 type autofs (rw)

automount(pid1794) on /sapmnt type autofs (rw)

filpool2-st:/vol/volFF/pool-pool2/pooldata on /FlexFrame/pooldata type nfs

(rw,v3,rsize=32768,wsize=32768,reserved,hard,tcp,

nolock,addr=filpool2-st)

The cn: parts show the mount points.

FlexFrame Architecture Storage Systems

34 Administration and Operation

2.7 Storage Systems

Concerning the storage aspect of a FlexFrame landscape today there is the basic and

mandatory NAS (Network Attached Storage) part and (as of version 4.0) the additive,

optional SAN (Storage Area Network) part. The central NAS storage can be a FAS

System from Network Appliance (NetApp) or a Celerra Network Server from EMC2. SAP

database data may reside on either NAS or SAN attached Storage from either Network

Appliance or EMC2.

Both the Network Appliance and EMC2 implementations of the NFS (Networked File

System) allow the same data files to be shared by multiple hosts and thus provide a built-

in cluster file system.

Fujitsu is working jointly with the two partners Network Appliance and EMC2 in the deve-

lopment of the FlexFrame concept. The Network Appliance product class Filer or FAS

System is an essential part of the FlexFrame infrastructure solution, and as of FlexFrame

4.0 the same is true for the EMC2 product class Celerra. Both product classes allow the

so called nested export function, which is important for the FlexFrame realization.

While data areas like operating systems, application software and commonly used data of

FlexFrame infrastructure solutions have comparatively low need for data throughput,

these areas remain on NAS and thus still benefit from the flexibility of the present

solution, the central data areas of type database and loggings (sapdata and saplog)

needing high data throughput may be shifted to SAN storage.

2.7.1 NAS Support Architectural Overview

2.7.1.1 Network Appliance Filer

In FlexFrame the storage for all Application Nodes can be one or more NAS systems

(Filers) from Network Appliance (or EMC, see 2.7.1.2). The Filer can be connected with

10Gbit/sec or 1Gbit/sec NICs to the FlexFrame system. The parallel use of 1Gbit/sec and

10Gbit/sec NICs of one Filer is not possible. But you can use on one filer 1Git/sec NICs

and on another 10Gbit/sec NICs.

The operating system of the Filer is called "ONTAP". The disks will be grouped into RAID

groups. A combination of RAID groups will make a volume. Starting with ONTAP 7, also

aggregates can be created. FlexVolumes can be created on top of aggregates. A Filer

volume contains a file system (WAFL - Write Anywhere File Layout) and provides

volumes or mount points for NFS (for UNIX systems) or CIFS (for Windows systems).

The Filer has NVRAM (Non Volatile RAM) that buffers committed IO blocks. The contents

of the NVRAM will remain intact if the power of the Filer should fail. Data will be flushed to

the disks once power is back online.

The minimal FlexFrame landscape has at least the following volumes:

vol0 (ONTAP, configuration of Filer)

Storage Systems FlexFrame Architecture

Administration and Operation 35

sapdata (database files)

saplog (database log files)

volFF (OS images of Application Nodes, SAP and database software,

pool related files)

In FlexFrame, the volume volFF separates FlexFrame data (file system of Application

Nodes and other software) from the Filer's configuration and ONTAP. In larger

installations, multiple sapdata and saplog volumes can be created (e.g. to separate

production and QA etc.).

2.7.1.1.1 Built-in Cluster File System

The Network Appliance implementation of the NFS (Networked File System) allows

sharing of the same data files between multiple hosts. No additional product (e.g. cluster

file system) is required.

2.7.1.1.2 Volume Layout

The FlexFrame concept reduces the amount of "wasted" disk space since multiple SAP

systems can optionally share the same volume of disks. As the data grow, it is easy to

add additional disks and enlarge the volumes without downtime.

2.7.1.1.3 Snapshots

When a snapshot is taken, no data blocks are being copied. Just the information where

the data blocks are located is saved. If a data block is modified, it is written to a new

location, while the content of the original data block is preserved (also known as ―copy on

write‖). Therefore, the creation of a snapshot is done very quickly, since only few data

have to be copied. Besides that, the snapshot functionality provided by NetApp is unique,

because the usage of snapshots does not decrease the throughput and performance of

the storage system.

Snapshot functionality will allow the administrator to create up to 250 backup-views of a

volume. The functionality ―SnapRestore‖ provided by NetApp significantly reduces the

time to restore any of the copies if required. Snapshots will be named and can be re-

named and deleted. Nested snapshots can be used to create e.g. hourly and daily

backups of all databases. In a FlexFrame landscape, a single backup server is sufficient

to create tape backups of all volumes. Even a server-less backup can be implemented.

Off-Line backups require a minimal down-time to the database because the backup to

tape can be done reading form a quickly taken snapshot.

2.7.1.1.4 Filer Cluster

A Filer can be clustered to protect data against the failure of a single Filer. Switching from

one Filer to its cluster counterpart is transparent to the Application Nodes.

FlexFrame Architecture Storage Systems

36 Administration and Operation

Filer clustering is a functionality of Network Appliance Filers.

2.7.1.1.5 vFiler

With activation of a Multistore license a Filer can only support vFiler. Every vFiler may

have its own administrator. A special vFiler named vfiler0 is automatically created and

cannot be deleted. Initially all resources are owned by vfiler0. Only the vfiler0

administrator can create or delete vFiler, volumes and interfaces and assign volumes and

interfaces to vFiler.

With FlexFrame 5.1 vfiler0 is allowed to be used as NAS System in the FlexFrame

environment. Other vFiler on the same storage host can be used for other purposes.

2.7.1.2 EMC NAS (Celerra)

As of FlexFrame for SAP V4.0A you can use an EMC NAS (Network Attached Storage of

EMC = Celerra) instead of or additive to a NetApp Filer. In the following we use ―NAS

system‖ or ―NAS‖ to indicate that we mean NetApp Filer and/or EMC NAS. Several times

we use ―Celerra‖ for EMC NAS. In FlexFrame for SAP EMC NAS is always connected

with 1Gbit/sec cabling technology.

An EMC Celerra consists of control station(s), data mover(s) and a storage system, which

is a CLARiiON® or a Symmetrix

®, depending on whether it is an integrated or a gateway

variant.

The EMC NAS family is composed of two - for FlexFrame relevant - product lines:

The NS line - available in both integrated and gateway models offering choice of 1, 2

or 4 data movers that attach to CLARiiON, or in the case of a gateway, to

CLARiiON/FibreCAT or Symmetrix storage subsystems.

The high end CNS (Celerra Clustered Network Server) with a high performance fault

tolerant cluster of data movers attached to enterprise class Symmetrix storage

subsystems and/or modular CLARiiON subsystems.

The minimal FlexFrame landscape has at least the following volumes:

sapdata (database files)

saplog (database log files)

volFF (OS images of Application Nodes, SAP and database software, pool related

files)

2.7.1.2.1 Control Station

The control station is a management computer which controls components of the Celerra

such as data movers (see below). The control station works with RedHat Linux as opera-

ting system, which is customized to its needs.

Storage Systems FlexFrame Architecture

Administration and Operation 37

The control station is able to connect to each data mover in the Celerra and send

commands to them. After the data movers are booted, they do not depend on the control

station for normal operation. In the unlikely event the control station fails, the data movers

continue to serve files to clients. Depending on the model, your Celerra Network Server

may include an optional standby control station that takes over if the primary control

station fails.

2.7.1.2.2 Data Mover

The data movers are the Celerra components that transfer data between the storage

system and the network client. The data mover operating system is DART (Data Access

in Real Time). You do not manage a data mover directly. You use the control station to

send commands to the data mover.

A Celerra can have from 1 to 14 data movers. Normally, the data movers are named

server_n, from server_2 to server_15.

A data mover can be active (or primary) or, in the case of a Celerra Network Server

model that supports multiple data movers, can be a standby for other data movers.

However you must configure one or more data movers as standby to support high

availability.

A data mover communicates with the data storage systems using Fibre Channel

connections. In the Celerra, this is done by a Fibre Channel HBA card, that connects to

the host through a Fibre Channel switch. In the Celerra NS series, storage processing is

provided by two storage processors.

2.7.1.2.3 Cluster Architecture

The EMC Celerra product family has distributed clustered architectures that give these

products competitive level of scalability. The EMC Celerra NS family supports one to four

data movers in a single system and the EMC Celerra CNS can support up to 14 data

movers. This scalability allows customers to grow their EMC NS and CNS environments

by adding data movers to enhance bandwidth, processing power and cache memory. The

EMC Celerra implementation of the NFS (Networked File System) allows sharing of the

same data files between multiple hosts. No additional product (e.g. cluster file system) is

required.

The Celerra Network Server protects against failures (service interruption or data loss)

with an additional control station and one or more data movers that can take over the

operations of a failed component.

Each data mover is a completely autonomous file server with its own operating system.

During normal operations, the clients interact directly with the data mover, not only for

NFS access, but also for control operations such as mounting and unmounting file

systems. Data mover failover protects the Celerra Network Server against total or partial

hardware failure of an individual data mover. You can define one data mover as standby

for the others (see Installation Guide ―Configuration of Standby Data Mover‖). This

FlexFrame Architecture Storage Systems

38 Administration and Operation

standby data mover has no own MAC address and IP address and no hostname. In the

case of a failover it takes all these from the failed data mover, so that the client cannot

decide, whether he is connected with the original or the standby data mover. In the same

time of the failover, the control station alerts the EMC service center via modem, and an

EMC technician can diagnose the problem and do corrective actions.

2.7.1.2.4 Snapshots

Snapshot images of a file system are used to minimize backup windows and to enable

quick restores of accidentally deleted or corrupted files. The EMC product for this

purpose is SnapSure™. It refers to these snapshot images as checkpoints. The amount

of storage consumed for each checkpoint is only determined by the amount of data that

was written since the last checkpoint was taken. Checkpoint images can be created on

demand or scheduled according to a time or change based policy.

As an example, consider a SAP database. A checkpoint at midnight is used to quickly

create a consistent point in time image of the file system for backup to tape. Checkpoints

during business hours as well are used to enable quick restores of accidentally deleted or

corrupted files and minimize the window of possible data loss compared to recovery from

last night's tape backup.

2.7.1.2.5 Two Side Mirror Architecture

In FlexFrame environments at the moment only SRDF active/passive configurations with

one control station for the R1 side and the R2 side are supported. R1 side is the active

Celerra where the NAS service is running for normal operation. R2 side is the passive

Celerra to which will be switched in case of errors to continuing the NAS service. The

configuration of both Celerra and both storage systems will be done by EMC. In general a

switchback to the R1 side after a switchover to the R2 side must be done by EMC,

because after NAS errors a check of Celerra environment and a following repair of the

error must be done for a successful switchback to the R1 side.

Beside SRDF mirrored active data movers an SRDF active/passive configuration can

have also not-mirrored local active data movers. These will be stopped in the case of a

switchover and so no data access is further possible for volumes which were mounted on

these data movers. There can be also not-mirrored active local data movers at the R2

side, these won't be stopped in a case of switchover and so they can serve further their

active mounted volumes. R1 side SRDF mirrored active data movers should have a local

standby data mover to protect a local failure. For each remote mirrored data mover at the

R1 side must exist a dedicated linked R2 side RDF standby data mover. In a minimal

remote high available SRDF active/passive configuration there are four data movers.

For the supported and released Celerra DART software version please consult the

FlexFrame support matrix.

Storage Systems FlexFrame Architecture

Administration and Operation 39

For the functionality of FlexFrame Celerra NAS high availability an ssh connection will be

used. The configuration of these ssh connections is described in the FlexFrame

installation manual.

Only volumes, which are mirrored through R1 devices to R2 devices, can be switched

over at all. Data volumes, which are not protected through an SRDF-R1-R2 mirroring,

can't be switched over. At the moment with FlexFrame NAS high availability only the

central volFF volume will be protected and only for the volFF volume a manually initiated

or an automatic switchover can be done. All other volumes, which are located on that

SRDF active/passive configuration, will be switched implicitly with the volFF volume.

There is no possibility for other data volumes in that SRDF active/passive configuration to

switch them separately.

For an active data mover with an assigned local standby data mover local standby

activation is controlled by a parameter policy. Three different policies exist: manual,

retry and auto. With manual a standby activation won't be done automatically, it must

be done manually with an administrative command. With retry at first a reboot of failed

data movers will be tried and afterwards only standby data mover activation will be

started. With auto standby data mover activation will be done immediately. In FlexFrame

auto must be set.

FlexFrame Celerra NAS high availability checks data access also with Celerra internal

commands. For that reason no switchover is initiated for healthy data movers with a

disturbed data access. Such events can result from an error on the IP data mover access

or wrong defined exports.

For automatic switching an access to both control stations must be possible. The R1

control station is needed to stop a controlled R1 side data mover to avoid data

inconsistencies in a split brain scenario.

In some special configuration situations no switchover will be done, e.g. standby data

mover policy = manual or SRDF disk group isn't set to synchronous mode.

FlexFrame Celerra SRDF-NAS high availability is completely implemented in the FlexFrame command ff_nas_ha.pl. This command has following operation modes:

init, list, check and switchover. Each operation mode must be specified with the

option –-op. e.g. ff_nas_ha.pl --op list. This FlexFrame command is only

installed on the Control Nodes and only users with the user ID = 0 are privileged to

execute this command.

After the above Celerra SRDF-NAS base configuration as a first step, a FlexFrame

Celerra SRDF-NAS initialization with the operation mode init must be done. With that

initialization in the main on each Control Node a local parameter file that is independent

from volFF will be created.

The state of a FlexFrame Celerra SRDF-NAS configuration can interactively determined

with option list.

FlexFrame Architecture Storage Systems

40 Administration and Operation

For an automatic control the operation mode check can be used. This operation mode

can also be executed manually, however the main user is the myAMC FrameAgent. The

FrameAgent checks periodically the volFF volume. In the case of an volFF error an snmp

trap will be generated. The FrameAgent is on both control nodes active and so from both

FrameAgents a check will be done concurrently, this can lead to multiple traps for one

error event.

A wanted automatic switchover must be configured in the FrameAgent different from the

default configuration of manual intervention after a volFF error. An automatic switchover

will be initiated if a prior executed check delivers an appropriate numeric return code.

An SRDF-NAS switchover from the R1 side to the R2 side will be done through the

operation mode switchover. This operation mode will be executed automatically by the

FrameAgent after an appropriate check return code and the configuration of automatic

switchover in the FrameAgent. The operation mode switchover can also be executed

manually if this was introduced by an error or as an administrative task, e.g. before any

maintenance work for the R1 side Celerra. A manual switchover will be done interactively

through the switchover caller.

Before performing a manual switchover or switchback the FrameAgents on both Control

Nodes must be stopped, otherwise the FrameAgents will be active in parallel and the

result may be not predictable!

For the configuration and in-depth information of the FrameAgent please refer to the

corresponding FlexFrame manual (FlexFrame™ for SAP® - myAMC.FA_Agents

Installation and Administration).

2.7.2 SAN Support Architectural Overview

This section provides an architectural overview on the integration of SAN-based storage

in the FlexFrame environment. It describes the level of integration provided with the

actual release of FlexFrame and the components which are considered supported and

those which are not explicitly supported, but can be used due to the approach described

here. Reading this section requires a basic technical understanding of the different SAN

components (e.g. fibre channel adapter, fibre channel switches and storage

components).

Storage Systems FlexFrame Architecture

Administration and Operation 41

2.7.2.1 SAN Basic Layers

Configuring a SAN environment for one or multiple SAP databases can be a very

complex task. As with many complex tasks they become easier if you break it down into

little pieces. In order to make you understand the various layers of a SAN environment

we divide it along the path as seen from the disk to the database:

Level Description

DISK The real disk(s) contained in the storage subsystem (aka

storage system).

RAID-GROUP One or multiple disks are grouped together based on RAID

mechanisms provided by the storage subsystem.

ARRAY-LUN LUNs as seen by the storage subsystem.

SAN-FABRIC A SAN fabric to connect one or multiple storage subsystems

with one or multiple hosts. It consists of fibre channel cabling

and switches.

HBA The HBA (host bus adapter) which connects the host to the

SAN fabric.

OS-DRIVER Basic OS drivers to make the HBA(s) work.

HOST-LUN LUNs as seen by the host (Application Node).

MULTIPATHING OS specific multipathing software to make multiple HBAs to

work as a single virtual interface.

VOLUME MANAGER An OS specific volume manager to group multiple LUNs from

the same or different storage subsystems into virtual volume

groups.

FILE SYSTEM OS specific file systems created on top of the virtual volumes

provided by the volume manager.

DATABASE The database files contained on the file system(s).

Each of the layers above can be configured in many different ways using lots of

variations. Therefore not all of those layers are controlled by FlexFrame.

FlexFrame Architecture Storage Systems

42 Administration and Operation

2.7.2.2 Scope of the FlexFrame SAN Integration

This section describes the scope of the FlexFrame SAN integration. The decisions were

made based on a couple of customer projects.

In order to broaden the set of ''usable'' SAN components in the FlexFrame environment

the scope of the project was not to integrate the full stack of hardware and software

layers described in 2.7.2.1, but rather take care of the configuration as off a certain level.

Thus, installation and configuration of storage systems, SAN switches/fabrics, volume

managers and even driver software is by no means automated, but left to the

responsibility of the user. Still, some predefined sets of the software stack components

have been tested and explicitly went through a Quality Assurance process.

As a baseline the SAN fabric is ''out of scope'' for all administrative tools within the

FlexFrame environment. Configuration of the storage system, such as configuration of

disks into RAID groups, provisioning of LUNs, zoning on FC switches is completely left to

the responsibility of the customer. However, some rules do apply to the outcome of a set-

up, such as ''all servers in one FlexFrame group (NOT pool) have to be able to access

the LUNs which are relevant for the SAP databases in question at the same time''. This

rule and further ones are described in detail later in this document. Since FlexFrame

4.2A, this special rule is obsoleted if dynamic LUN masking is used.

On the Application Nodes, a set of adapters with pre-defined drivers is supported, in a

way that those set-ups are fully described in the documents and went through a quality

assurance cycle. Specific software components mentioned later are required in order to

support certain features (such as multipathing). Some software packages (esp. for Linux),

which are generally available, may explicitly be excluded from usage in a FlexFrame

environment, but this is usually due to SAP's strict rules concerning the use of none-GPL

drivers in Linux systems with their software on-top.

FlexFrame comes with the utilities to optionally use a set of volume managers; however

customer specific software can easily be hook into the environment, if necessary.

The SAN integration is limited to the use for database files and logs only. Meaning

everything else will still be placed on NAS (NFS) storage. This explicitly includes the

database software, SAP applications and the OS. Even the database files do not

necessarily have to be placed onto a SAN based storage. One might choose to place

large, productive databases with IO requirements onto the SAN based storage while

smaller databases will be on NAS storage. The choice is based on SID level. One SID

might be on SAN another SID on NAS.

The following pictures illustrate the components which can be placed onto NAS and SAN

storage.

Storage Systems FlexFrame Architecture

Administration and Operation 43

FlexFrame Components on NAS

FlexFrame Components on SAN

2.7.2.3 Rules and Restrictions

In order to make things work correctly with FlexFrame some rules upon each layer must

be configured. This chapter lists some rules and restrictions which must be observed.

1. All Application Nodes within one pool group must be configured equally. This

includes the number and type of HBAs. They must also use the same OS image.

2. All Application Nodes within one pool group must have access to the same set of

LUNs of each involved storage subsystem. Each Application Node of a pool group

must be able to access the same LUNs without reconfiguring the access settings of

the storage subsystem. Since FlexFrame 4.2A, this rule is obsoleted if dynamic LUN

masking is used. With dynamic LUN masking, the access settings on the storage

subsystems are dynamically modified using StorMan Software, so that each

Application Node gets access to the LUNs needed by the database component of a

SAP system when this database instance is started on the Application Node.

3. Host based mirroring can be used on volume manager layer to mirror one or multiple

LUNs from one storage subsystem to another (e.g. for a two-site-concept). Copy

mechanisms of the storage subsystems can also be used. If the failover to a different

storage subsystem requires a change in the WWN and host LUN addressing of the

fail-over LUN, this change must also be done in the FlexFrame LDAP database

using the appropriate FlexFrame script..

4. A volume group has an n:1 relationship to a SID. There may be multiple volume

groups for one SID, but a volume group must not contain volumes for more than one

SID.

The appropriate volume groups are switched from one Application Node to

the next if a database is moved.

5. Access to all file systems of a SID must be tested on each Application Node of its

pool group before it can be monitored and used (―watch‖).

6. A database which was installed using SAN based database files in one pool group

cannot be started in a different pool group. If this is required the database must be

installed on NAS.

FlexFrame Basic AdministrationAccessing a FlexFrame Landscape (Remote Administration)

44 Administration and Operation

3 FlexFrame Basic Administration

3.1 Accessing a FlexFrame Landscape (Remote Administration)

A FlexFrame landscape can be accessed through Secure Shell (ssh, scp) connections

to the Control Node. Any other remote administration tools like rsh or telnet have been

disabled for reasons of security.

3.2 Powering up the FlexFrame Landscape

If the complete FlexFrame landscape was powered off, the following power-on sequence

is recommended:

Powering off the FlexFrame Landscape FlexFrame Basic Administration

Administration and Operation 45

Before an Application Node can be booted, the NTP server must be set up correctly. This

can be verified by running the following command:

control1:~ # ntpq -p

remote refid st t when poll reach delay offset jitter

=================================================================

*LOCAL(0) LOCAL(0) 5 l 45 64 377 0.000 0.000 0.004

control2-se control1-se 7 u 248 1024 377 2.287 2.587 0.815

This command needs to be repeated until an asterisk (*) is displayed at the beginning of

one of the data lines. This character indicates that the NTP server is now ready. If you do

not wait and continue with booting, the Application Nodes may work with a different time

than the Control Nodes and may (among other possible side effects) create files which

may have wrong time stamp information.

If this sequence is not used and all servers are powered on at the same time, the

Application Nodes will try to boot while the Control Nodes are not ready to

receive the Application Node's boot request. If this is the case, manual

intervention is required to re-initiate the boot process of the Application Nodes.

3.3 Powering off the FlexFrame Landscape

If you need to power-off the complete FlexFrame landscape (e.g. to move it to a different

location) we recommend following the steps as outlined below:

FlexFrame Basic Administration Powering off the FlexFrame Landscape

46 Administration and Operation

Before shutting down all SAP and DB services, check if users, batch jobs, print

jobs and RFC connections to other SAP systems have finished working.

To shutdown the SAP and DB services, use the following command for each pool:

control1:~ # stop_all_sapservices <pool_name>

Before you can stop the NAS-System, you need to stop all processes on the Control

Nodes. Since those processes are under control of the Linux-HA cluster you have to

switch all resources from control node 2 to control node 1 and then stop all resources.

control1:~ # ssh control2 /sbin/rcheartbeat stop

control1:~ # /sbin/rcheartbeat stop

Keep in mind, stopping all resources will take some time.

Now stop the myAMC LogAgent on both Control Nodes.

control1:~ # ssh control2 /etc/init.d/myAMC.FA_LogAgent stop

control1:~ # /etc/init.d/myAMC.FA_LogAgent stop

And umount all NFS filesystems on both Control Nodes.

control1:~ # ssh control2 umount -a -t nfs

control1:~ # umount -a -t nfs

Now the NAS systems can be safely switched off without impacting the Control Center

shutdown later on.

To stop the NAS system, use the following command.

For a NetApp Filer:

control1:~ # rsh <filer_name> halt

There is no explicit power-off sequence for the switches. Assuming there are no

other devices connected to the switches, they may simply be plugged-off after all

components were powered down.

If you do not send an explicit halt command to the Filer, the backup-battery of

the Filer may be drained since the Filer assumes a power-loss and tries to

preserve the contents of its NVRAM. If the Filer is powered-off for too long the

result can be a loss of NVRAM data and a long waiting period during next

startup of the Filer.

For an EMC Celerra (where cel-co is the control LAN address of the control station of

Celerra):

Powering off the FlexFrame Landscape FlexFrame Basic Administration

Administration and Operation 47

control1:~ # ssh <cel-co> -l root

cel-co: # init 0

FlexFrame Basic Administration Reactivating ANs after Power Shutdown by FA Agents

48 Administration and Operation

3.4 Reactivating ANs after Power Shutdown by FA Agents

FlexFrame Autonomy places an Application Node out of service if it is confronted with a

problem it cannot solve. The reactions, messages and alarms which take place in this

case are described in the ―FA Agents - Installation and Administration‖ manual in the

context of the switchover scenarios. It is the responsibility of the administrator to analyze

why the node could not be used any longer. We recommend analyzing the FA log and

work files as these may be able to supply valuable information. A node which is to start

operating as a Spare Node after switchover must be validated using suitable test

scenarios.

It may happen in some cases that the "placing out of service" of an Application Node fails

(e.g. when the IPMI-interface has broken down) and so it cannot be decided if the

Application Node is down.

In these cases - as default - the services that were running on the server are not switched

over to another node.

Therefore there is an additional high availability FA Agents parameter

"IgnoreShutdownFailure" with impact on the failover reaction for services.

Setting this parameter to value "true"

assigns the FA Agents to deactivate the network interfaces of the server

if successful the FA Agents automatically switch over services that were running on

the server to another node

the failed server can only be reactivated by manually reactivating its network

interfaces via the command ff_sw_ports.pl

(see the chapters "Switchover Control Parameters" and "Shutdown Configuration" in the

FA Agents Installation and Administration manual).

Default value "false" means, that a not successful server shutdown has no effect on the

server network interfaces and the services that were running on the server are not

switched over to another node.

Setting this parameter strengthens the service high availability, but the FlexFrame

administrator has to be aware of its impact in certain failure scenarios.

Recommendation is to set the parameter "IgnoreShutdownFailure" to "true" after

installation of the Flex-Frame system, if the FlexFrame administration wants to expand

the high availability of service and is willing to manually reactivate the network interfaces

of a server in such error cases.

Displaying the Current FlexFrame Configuration State FlexFrame Basic Administration

Administration and Operation 49

3.5 Displaying the Current FlexFrame Configuration State

To obtain a general overview of an active FlexFrame system, use the FA WebGUI.

In principle, the FA WebGUI can be used with every browser with SUN JAVA Plugin

V1.4.1 or higher which has access to the page on the Control Node. The WebGUI can

always be accessed when the Apache Tomcat service is running. This service is normally

started by the Linux-HA cluster.

The WebGUI is described in detail in chapter 3.5.1.

The login mask expects a user name and password for authentication purposes. You can

only use the WebGUI with a valid combination of user name and password. For details on

the configuration of the users, see the myAMC documentation for the WebGUI.

The FA WebGUI provides a presentation of all elements of a FlexFrame system. On the

left-hand side the pools, groups, nodes and the active SAP services on the individual

nodes are shown in a TreeView.

FlexFrame Basic Administration Displaying the Current FlexFrame Configuration State

50 Administration and Operation

The TreeView of the FA WebGUI can show either the physical view or the application-

related view of the active SAP systems and their instances or a mixed view.

The panels derived from the TreeView (Application Server Panel, Application System

Panel, ServiceView and MessageView) always show the objects in relation to the

selected hierarchical level in the TreeView.

The FA WebGUI is thus the central cockpit for displaying the static configuration of a

FlexFrame infrastructure, but as well it is the display for the active SAP systems and their

instances. Here, all user interactions such as startup and shutdown are shown directly.

All reactions initiate by the FA Agents are displayed as well.

If the FA messenger component has been configured and activated, all traps are stored

in a support database. This permits the temporal process for the traps to be displayed

very simply at pool, group or node level.

3.5.1 FlexFrame Web Portal

The FlexFrame Control Nodes provide a web portal with links to Web interfaces of

several FlexFrame components, i.e.

FA Autonomous Agents

ServerView Operations Manager

Cluster status

Supported hardware

Management Blades of BX cabinets

To access this portal, start firefox and enter the Control Node's IP address in the location

bar. If you are directly on the Control Node you want to configure, just call firefox from the

shell:

control1:~ # firefox localhost

You will see an overview of all Web interfaces installed on the Control Nodes.

Displaying the Current FlexFrame Configuration State FlexFrame Basic Administration

Administration and Operation 51

3.5.2 FA Autonomous Agents

Use this tool to manage the virtualized services in your FlexFrame environment.

For information on usage, please refer to the FA Autonomous Agents manuals.

You can access the FlexFrame Autonomous Agents WebGUI directly from the

active Control Node by entering the following URL:

http://localhost:8080/FAwebgui

This only works if the Jakarta Tomcat web server is running. If it is not running,

check the Linux-HA cluster configuration.

3.5.2.1 State of Pools

The active configured pools are displayed on the FA WebGUI. The nodes or systems

belonging to a pool are displayed in the FA WebGUI TreeView. Each node element in the

tree which represents a pool is identified by the prefixed keyword Pool.

FlexFrame Basic Administration Displaying the Current FlexFrame Configuration State

52 Administration and Operation

3.5.2.2 State of Application Nodes

Each Application Node with a running FA Application Agent is shown in the FA WebGUI.

It is shown in the Nodes TreeView with its host name (Linux), and also in the Application

Server Panel. In the Application Server Panel the data displayed depend on the

hierarchical level selected in the TreeView.

3.5.2.3 State of SAP Systems

The active SAP system IDs can be displayed very easily in the Systems TreeView of the

FA WebGUI and also in the SAP System Panel. The SAP System Panel is shown in

parallel to the Application Server Panel. In the SAP System Panel, the data displayed

depend on the hierarchical level selected in the TreeView.

3.5.2.4 State of SID Instances

The active instances of a SID can be displayed very simply in the InstancesView of the

FA WebGUI by clicking on a SID in the SAP System Panel.

For each view, information is provided in tabular form specifying the service's current

pool, group, node, priority and status.

3.5.3 Networks

The network of FlexFrame is its backbone. Here are some tips to get an overview of the

current situation on the various networks:

To double-check the network addresses, their names and pool assignment you can use the getent command:

control1:~ # getent networks

loopback 127.0.0.0

control 192.168.20.0

storage_pool1 192.168.10.0

server_pool1 192.168.3.0

client_pool1 10.1.1.0

storage_pool2 192.168.11.0

server_pool2 192.168.4.0

client_pool2 10.1.2.0

The loopback network is local for each host and always has the IP address 127.0.0.0.

The control network is the Control LAN network segment for the complete FlexFrame

landscape.

In the example we have configured two pools called pool1 and pool2. For each pool

there are the three dedicated and distinct segments storage, server and client. The

building rule of the network name is <segment>_<pool_name>.

Displaying the Current FlexFrame Configuration State FlexFrame Basic Administration

Administration and Operation 53

On the Control Nodes you can see the relation of each pool specific segment to its interface using the netstat -r command like this:

control1:~ # netstat -r

Kernel IP routing table

Destination Gateway Genmask Flags MSS Window irtt Iface

server_pool2 * 255.255.255.0 U 0 0 0 vlan42

control * 255.255.255.0 U 0 0 0 bond0

server_pool1 * 255.255.255.0 U 0 0 0 vlan32

storage_pool2 * 255.255.255.0 U 0 0 0 vlan41

client_pool1 * 255.255.255.0 U 0 0 0 vlan30

storage_pool1 * 255.255.255.0 U 0 0 0 vlan31

client_pool2 * 255.255.255.0 U 0 0 0 vlan40

default gw216p1 0.0.0.0 UG 0 0 0 vlan30

Here you can quickly see that the Server LAN segment of pool2 (server_pool2) is using

the VLAN ID 42 on interface vlan42. Note that the VLAN ID of the control Control LAN

segment cannot be seen because that vlan is native on bond0.

3.5.4 ServerView Operations Manager

ServerView Operations Manager allows you to query information of PRIMERGY servers

in the FlexFrame environment.

This tool is not needed for administration of FlexFrame but is required by Fujitsu support.

If you would like to monitor the hardware status of PRIMERGY servers in your

FlexFrame™ environment, click on ServerView and on the next page, click on the

button labeled start:

FlexFrame Basic Administration Displaying the Current FlexFrame Configuration State

54 Administration and Operation

Please note that SSL connections are not supported.

Beginning with FlexFrame 4.0, the ServerView database for configuration, traps

and server lists are no longer shared across both Control Nodes. Each Control

Node runs its own instance with its own configuration.

Displaying the Current FlexFrame Configuration State FlexFrame Basic Administration

Administration and Operation 55

An overview of all configured servers (initially only the local host) will be shown on the

main screen:

To add more servers, run the Server Browser by either clicking on Administration /

Server Browser on the top navigation bar or right clicking anywhere in the list and

selecting New Server from the context menu.

Be aware that in some cases the ServerView Operations Manager needs a valid user ID

in order to authenticate itself. For example, the Server Browser will only recognize an

ESXi Server if ServerView Operations Manager can authenticate itself using a valid user

ID from the user/password list which can be updated using the Users/Passwords

dialog in the Administration menu.

FlexFrame Basic Administration Displaying the Current FlexFrame Configuration State

56 Administration and Operation

The Server Browser window will be opened:

The easiest way to add servers is to scan a subnet for each pool, for example the

―Server‖ subnet for Application Nodes and the ―Control‖ subnet for management blades

and ESXi hosts.

Displaying the Current FlexFrame Configuration State FlexFrame Basic Administration

Administration and Operation 57

To scan a subnet, enter the first three bytes of the network address in the Subnet input

field on the bottom left of the screen and click on Start Browsing:

The browsing process is finished when the button Stop Browsing changes its caption to

Start Browsing.

FlexFrame Basic Administration Displaying the Current FlexFrame Configuration State

58 Administration and Operation

To add all manageable servers, right click anywhere on the list and click on Select Manageables.

Displaying the Current FlexFrame Configuration State FlexFrame Basic Administration

Administration and Operation 59

After clicking on Select Manageables, all servers with known type will be selected:

To finally add those selected servers, click on Apply in the upper right corner of the

Server Browser window.

After closing the Server Browser, the main window containing the server list will show the

added servers. Further servers can be added by repeating the recent steps using a

different subnet address.

FlexFrame Basic Administration Displaying the Current FlexFrame Configuration State

60 Administration and Operation

To view the state of a single server, click on the blue server name. To view the state of a

Server Blade, click on the Management Blade / Blade Center, then select the right blade

and click on ServerView on the bottom navigation bar.

Events such as SNMP Traps can be viewed by navigating to

Event Management on the top navigation bar. This will open ServerView Alarm

Service, which will not be described in detail. To monitor SNMP Traps, we

recommend using FA WebGUI / myAMC Messenger.

Administration and Operation 61

3.5.5 Cluster Status

A short overview of the cluster status is available from the FlexFrame Web Portal.

3.5.6 Supported Hardware

Clicking on the ‗Supported Hardware‘ link in the FlexFrame Web Interface displays the

currently supported hardware and the features used by FlexFrame. The list includes

hardware that is supported only by an approved special release request.

If JavaScript is enabled, clicking on the table headers displays information about the

listed content.

3.5.7 Management Blades

Clicking on the links in the management blade section of the FlexFrame Web Interface

opens https connections to the respectice management blades of BX cabinets.

Note that the management blade section may be missing if the FlexFrame landscape

contains no BX cabinets.

FlexFrame Backup with Tape Library

62 Administration and Operation

3.6 FlexFrame Backup with Tape Library

3.6.1 NetWorker

A dedicated backup server will be used for maintaining NetWorker. As a backup tool

NetWorker is used including the database module NetWorker Module PLUS for

Oracle.

In the case of a NetApp Filer as NAS device the database module NetWorker Module

PLUS for Oracle is used. This concept is based on snapshots and uses the NDMP

protocol for transferring data from the NetApp Filer directly to the tape library as shown in

the example below.

Configuration Example for Oracle on NetApp Filer:

In the graphic the * means: NetWorker Module PLUS for Oracle for NDMP NetApp

Detailed information on FlexFrame for SAP Backup with NetWorker such as

Implementation and configuration of the NetWorker Backup Solution for Oracle (at

great length)

Slide sets for different target groups like marketing and sales

White papers will be provided by Fujitsu Storage Consulting and is available at

https://partners.ts.fujitsu.com/com/products/infrastruc-solutions/FlexFrame, see Best

Practices / Description Papers.

FlexFrame Backup with Tape Library

Administration and Operation 63

3.6.2 ARCserve

Detailed information on a backup solution with CA ARCserve is available at

https://partners.ts.fujitsu.com/com/products/storage/software/brightstor/arcserve_backup/

arc_tech_slides .

Pools and Groups Adding a Pool

64 Administration and Operation

4 Pools and Groups

4.1 Adding a Pool

A new pool can be added using ff_pool_adm.pl. Some parameters have to be defined

on the command line. They are used to configure switch VLANs and ports, to create the

NAS system volume folder structures, to create LDAP pool subtree and to configure

Control Nodes.

An ADMINPOOL (see chapter 6) is created, if <pool name> is adminpool or

force_admin is given.

If the pool shall be equipped with Application Nodes of type ―BX600‖ (blade

server in a BX600 blade server cabinet with switch blades of type 1GbE-10/6-Q

(SB9a)), then the following restriction concerning the VLAN-ID has to be consi-

dered:

Switch blades of type 1GbE-10/6-Q only allow the VLAN ID range from 1 to

3965!

Only one ADMINPOOL is allowed in a FlexFrame system.

Adding a pool changes the exports file on all given NAS systems. Temporary

exports on these NAS systems will be gone after running ff_pool_adm.pl.

Be sure not to have temporary exports.

Adding a Pool Pools and Groups

Administration and Operation 65

Creating a new pool establishes a "ClientLAN to CorporateLAN" (CLAN) switch

configuration. Two global parameters are used to determine the type of this

CLAN.

The settings are originally given by the Management Tool (MT, see the manual

Management Tool, chapter ―Network‖ Object) and may be changed by

ff_swgroup_adm.pl --op parameters (SWGADM) (see this manual

chapter 9).

The parameters are One ClientLAN per SWP (true/false) at MT or

clanPortPerVlan (yes/no) at SWGADM

true/yes Use an independent switch port pair for each pool Client LAN.

These ports will be configured as native access ports.

false/no All pool ClientLANs use the same switch port pair. Both ports are

configured as Cisco trunk ports, with tagged VLANs.

Use 1Gbit SWPs for ClientLANs at MT or

useTxToClan at SWGADM

true/yes use copper twisted pair ports as CLAN ports.

false/no use fiber optic (SFP) ports as CLAN ports if available else use

copper ports.

So if the original global setting does not match your current wish for wiring the

new pool, you can change it temporarily with the two corresponding

ff_swgroup_adm.pl commands.

Synopsis

ff_pool_adm.pl

--op add --name <pool name>

--storage <vlan id>,<network ip>,<netmask>

--server <vlan id>,<network ip>,<netmask>

--client <vlan id>,<network ip>,<netmask>

--dns <domain name>[,<dns server ip>[,<dns server ip>]

-- defrouter <default router ip>]

[--dns_search_list <list of domains>]

[--sapdata <nas name>[,<volume path>]]

[--saplog <nas name>[,<volume path>]]

[--volff <nas name>[,<volume path>]]

[--defrouter <default router ip>]

Pools and Groups Adding a Pool

66 Administration and Operation

[--switchgrp <id>[,<id>]]

[--force_admin]

Options

--op add

Adds a pool and displays information about processing steps, errors and warnings.

--name <pool name>

Name of new pool (has to be unique withinFlexFrame). A pool name may consist of

2 up to 13 characters (letters, numbers and dashes). It must start with a letter and

must not end with a dash.. The pool name adminpool always creates an

ADMINPOOL.

--storage <vlan id>,<network ip>,<netmask>

Pool specific storage network segment. The option is followed by a comma

separated list of the VLAN ID, the network IP and the netmask of the network IP

address.

--server <vlan id>,<network ip>,<netmask>

Pool specific server network segment. The option is followed by a comma separated

list of the VLAN ID, the network IP and the netmask of the network IP address.

--client <vlan id>,<network ip>,<netmask>

Pool specific client network segment. The option is followed by a comma separated

list of the VLAN ID, the network IP and the netmask of network IP.

--dns <domain name>,[<dns server ip>]

DNS domain name and servers to be used for this pool. More than one DNS server

IP address may be given. Keep in mind to use the default router option if server IP

addresses are given. A DNS option may look like this:

my.domain.com,192.168.251.17,192.168.251.18

--dns_search_list <list of domains>

List of DNS domain names to be used for this pool. At most six domain names are

allowed. For technical reasons the first DNS domain name of this list always must be

the DNS domain name given in option dns.

Adding a Pool Pools and Groups

Administration and Operation 67

--sapdata <nas name>[,<volume path>]

Optional NAS name and volume path the pool should use for sapdata. A missing

volume path is auto-filled with the default (/vol/sapdata/<pool_name>). e.g.

filer1,/vol/sapdata/pool1. The entire option defaults to common NAS name

with default path. <nas name> is the NAS system's node name for this pool

(without -st suffix).

--saplog <nas name>[,<volume path>]

Optional NAS name and volume path the pool should use for saplog. A missing

volume path is auto filled with the default (/vol/saplog/<pool_name>). e.g.

filer1,/vol/saplog/pool1. The entire option defaults to common NAS name

with default path. <nas name> is the NAS system's node name for this pool

(without –st suffix).

--volff <nas name>[,<volume path>]

Optional NAS name and volume path the pool should use for volFF. A missing volume path is auto filled with the default (/vol/volFF/pool-<pool_name>).

e.g. filer1,/vol/volFF/pool-pool1. The entire option defaults to common

NAS name with default path. <nas name> is the NAS system's node name for this

pool (without -st suffix).

--defrouter <default router ip>

The default router is a gateway to route IP data to other, non-pool local networks. All

IP data that can not be addressed to a local network will be sent to the default router

to be forwarded to the destination network. The option parameter is an IP address of

this default router. Use a default router IP address matching the client pool network,

because otherwise it will not be accessible by Application Nodes.

Pools and Groups Adding a Pool

68 Administration and Operation

--switchgrp <id>[,<id>]

The switch group ID(s) the Client LAN to corporate LAN ports should be configured.

If not given, the client VLAN is assigned to the existing trunk ports or a new port pair

at first both switch groups. Not more than two switch group IDs are accepted.

--force_admin

This pool (with any pool name) should be the ADMINPOOL

Example

cn1:~ # ff_pool_adm.pl --op add --name pool4

--storage 30,192.168.30.0,255.255.255.0 --server 31,192.168.31.0,255.255.255.0

--client 32,192.168.32.0,255.255.255.0 --sapdata filer --saplog filer

--volff filer --dns my.domain.com --defrouter 192.168.32.254

update LDAP

...

update switch 1/1 configuration

Notice: Update will take about 1 minute.

vlan: storage-30 has been created

restart cluster service ldap_srv1

Notice: restart will take up to 1 minute.

stop and wait until service is offline

start and wait until service is online

restart cluster service ldap_srv2

Notice: restart will take up to 1 minute.

stop and wait until service is offline

start and wait until service is online

restart cluster service netboot_srv

Notice: restart will take up to 2 minutes.

stop and wait until service is offline

start and wait until service is online

If not reported any warnings or errors all precautions are done

and the pool was successfully created.

Use ff_poolgroup_adm.pl to define the host groups of this pool

to be able to add application nodes.

See /tmp/pool-pool4/ff_pool_adm.errlog for complete error and warning log.

Removing a Pool Pools and Groups

Administration and Operation 69

4.2 Removing a Pool

A pool can be removed using ff_pool_adm.pl. Some parameters have to be defined

on the command line. Switch VLANs will be removed and the affected ports reconfigured.

The LDAP pool subtree will be removed and Control Node configurations rewritten.

The pool created with the attribute "CN Pool = true" may not be removed by an

administration command, since this pool is used to mount file systems from the NAS

device.

A pool may not be removed if any Application Node or SID is defined. The first

pool may not be removed due to system requirements. Removing a pool

changes the exports on all NAS systems used by this pool. Use list or list-

all to get the storage configuration of pool to be removed. Temporary exports

(not written to the exports file /vol0/etc/exports) on these NAS systems will

be gone after running ff_pool_adm.pl. Be sure not to have temporary

exports.

Synopsis

ff_pool_adm.pl --op rem --name <pool name>

Options

--op rem

Removes a pool and displays only errors and warnings.

--name <pool name>

Name of pool to be removed. Use ff_pool_adm.pl --op list-all to get a list

of currently configured pools (see 4.4).

Example

cn1:~ # ff_pool_adm.pl --op rem --name pool4

update LDAP

...

update switch 1/1 configuration

Notice: Update will take about 1 minute.

restart cluster service ldap_srv1

Notice: restart will take up to 1 minute.

stop and wait until service is offline

start and wait until service is online

restart cluster service ldap_srv2

Notice: restart will take up to 1 minute.

Pools and Groups Listing Pool Details

70 Administration and Operation

stop and wait until service is offline

start and wait until service is online

restart cluster service netboot_srv

Notice: restart will take up to 2 minutes.

stop and wait until service is offline

start and wait until service is online

If not reported any warnings or errors the pool was successfully removed.

Keep in mind, the volumes and their data were not harmed. It's on you to remove

them.

See /tmp/pool-pool4/ff_pool_adm.errlog for complete error and warning log.

4.3 Listing Pool Details

To list the configurations details of a pool like used networks, pool groups, SIDs and

Application Nodes, the maintenance tool ff_pool_adm.pl can be used. The pool name

has to be defined on the command line.

Synopsis

ff_pool_adm.pl --op list --name <pool name>

[--list <part>[,<part>]]

Options

--op list

Displays pool configuration details

--name <pool name>

Name of pool to be listed. Use ff_pool_adm.pl --op list-all to get a list of

currently configured pools (see 4.4).

--list <part>[,<part>]

To reduce the output to the interesting parts, use this option. The parameters to this

option are the display sections. Add them as a comma separated list. The default

sections are: network,storage,dns,group,sid,an,cn,NAS system. You may also use a two character abbreviation instead the full section name like ne for network.

Listing Pool Details Pools and Groups

Administration and Operation 71

Examples

cn1:/opt/FlexFrame/bin # ff_pool_adm.pl --op list --name p1

Pool configuration details of pool p1

Networks

Client-LAN

Network: 10.10.10.0 Netmask: 255.255.255.0 VLAN ID: 100

Server-LAN

Network: 192.168.10.0 Netmask: 255.255.255.0 VLAN ID: 110

Storage-LAN

Network: 192.168.20.0 Netmask: 255.255.255.0 VLAN ID: 120

Def.Router: 10.10.10.254

Storage Volumes

sapdata fas01-p1-st:/vol/sapdata/p1

saplog fas01-p1-st:/vol/saplog/p1

volFF fas01-p1-st:/vol/volFF/pool-p1

volFF shared fas01-p1-st:/vol/volFF

DNS data

Domain Name: my.domain.com

Pool Groups

Linux

OS: SUSE Linux SLES-9.X86_64

SIDs and their instances

D01

SAP Version: SAP-6.20 DB Version: Oracle-9

Instances

Type db

ID 0 Server-LAN: dbd01-se 192.168.10.110

Type ci

ID 9 Client-LAN: cid01 10.10.10.111

Server-LAN: cid01-se 192.168.10.111

Type app

ID 10 Client-LAN: app10d01 10.10.10.112

Server-LAN: app10d01-se 192.168.10.112

ID 11 Client-LAN: app11d01 10.10.10.113

Server-LAN: app11d01-se 192.168.10.113

P01

SAP Version: SAP-6.20 DB Version: Oracle-9

Instances

Type db

ID 0 Server-LAN: dbp01-se 192.168.10.100

Type ci

Pools and Groups Listing Pool Details

72 Administration and Operation

ID 0 Client-LAN: cip01 10.10.10.101

Server-LAN: cip01-se 192.168.10.101

Type app

ID 1 Client-LAN: app01p01 10.10.10.102

Server-LAN: app01p01-se 192.168.10.102

ID 3 Client-LAN: app03p01 10.10.10.103

Server-LAN: app03p01-se 192.168.10.103

ID 4 Client-LAN: app04p01 10.10.10.104

Server-LAN: app04p01-se 192.168.10.104

ID 5 Client-LAN: app05p01 10.10.10.105

Server-LAN: app05p01-se 192.168.10.105

Q01

SAP Version: SAP-6.20 DB Version: Oracle-9

Instances

Type db

ID 0 Server-LAN: dbq01-se 192.168.10.106

Type ci

ID 6 Client-LAN: ciq01 10.10.10.107

Server-LAN: ciq01-se 192.168.10.107

Type app

ID 7 Client-LAN: app07q01 10.10.10.108

Server-LAN: app07q01-se 192.168.10.108

ID 8 Client-LAN: app08q01 10.10.10.109

Server-LAN: app08q01-se 192.168.10.109

Application Nodes

blade01

Type: BX600 Cabinet ID: 1 Slot/Partition ID: 1

OS: SUSE Linux SLES-9.X86_64

Group: Linux

Client-LAN blade01 10.10.10.23

Server-LAN blade01-se 192.168.10.23

Storage-LAN blade01-st 192.168.20.23

blade02

Type: BX600 Cabinet ID: 1 Slot/Partition ID: 2

OS: SUSE Linux SLES-9.X86_64

Group: Linux

Client-LAN blade02 10.10.10.24

Server-LAN blade02-se 192.168.10.24

Storage-LAN blade02-st 192.168.20.24

blade03

Type: BX600 Cabinet ID: 1 Slot/Partition ID: 3

OS: SUSE Linux SLES-9.X86_64

Group: Linux

Client-LAN blade03 10.10.10.25

Listing Pool Details Pools and Groups

Administration and Operation 73

Server-LAN blade03-se 192.168.10.25

Storage-LAN blade03-st 192.168.20.25

rx801

Type: RX800

OS: SUSE Linux SLES-9.X86_64

Group: Linux

Client-LAN rx801 10.10.10.2

Server-LAN rx801-se 192.168.10.2

Storage-LAN rx801-st 192.168.20.2

Control Nodes

cn1

Client-LAN cn1-p1 10.10.10.10

Server-LAN cn1-p1-se 192.168.10.10

Storage-LAN cn1-p1-st 192.168.20.10

cn2

Client-LAN cn2-p1 10.10.10.11

Server-LAN cn2-p1-se 192.168.10.11

Storage-LAN cn2-p1-st 192.168.20.11

NAS Nodes

fas01-p1

Storage-LAN fas01-p1-st 192.168.20.14

A sample with a reduced output:

cn1:/opt/FlexFrame/bin # ff_pool_adm.pl --op list --name p1 --list ne,gr

Pool configuration details of pool p1

Networks

Client-LAN

Network: 10.10.10.0 Netmask: 255.255.255.0 VLAN ID: 100

Server-LAN

Network: 192.168.10.0 Netmask: 255.255.255.0 VLAN ID: 110

Storage-LAN

Network: 192.168.20.0 Netmask: 255.255.255.0 VLAN ID: 120

Def.Router: 10.10.10.254

Pool Groups

Linux

OS: SUSE Linux SLES-9.X86_64

Pools and Groups Listing All Pools

74 Administration and Operation

4.4 Listing All Pools

To display an overview of all pools with their used networks, pool groups, SIDs and Control Node and NAS system interfaces, the maintenance tool ff_pool_adm.pl can

be used. No arguments except the operation mode have to be defined on the command

line.

Synopsis

ff_pool_adm.pl --op list-all [--list <part>[,<part>]]

Options

--op list-all

Displays the configuration details of all configured pools.

--list <part>[,<part>]

To reduce output to interesting parts use this option. The parameters to this option

are the display sections. Add them as a comma separated list. The default sections

are: network,storage,group,sid,cn,NAS system. You may also use a two character

abbreviation instead the full section name like ne for network.

Examples

cn1:/opt/FlexFrame/bin # ff_pool_adm.pl --op list-all

Pool configurations

p1

Pool Networks

Client-LAN

Network: 10.10.10.0 Netmask: 255.255.255.0 VLAN ID: 100

Server-LAN

Network: 192.168.10.0 Netmask: 255.255.255.0 VLAN ID: 110

Storage-LAN

Network: 192.168.20.0 Netmask: 255.255.255.0 VLAN ID: 120

Pool Storage Volumes

sapdata fas01-p1-st:/vol/sapdata/p1

saplog fas01-p1-st:/vol/saplog/p1

volFF fas01-p1-st:/vol/volFF/pool-p1

volFF shared fas01-p1-st:/vol/volFF

Pool Groups

Linux

OS: SUSE Linux SLES-9.X86_64

Listing All Pools Pools and Groups

Administration and Operation 75

Pool SIDs

D01

SAP Version: SAP-6.20 DB Version: Oracle-9

P01

SAP Version: SAP-6.20 DB Version: Oracle-9

Q01

SAP Version: SAP-6.20 DB Version: Oracle-9

Pool Control Node Interfaces

cn1

Client-LAN cn1-p1 10.10.10.10

Server-LAN cn1-p1-se 192.168.10.10

Storage-LAN cn1-p1-st 192.168.20.10

cn2

Client-LAN cn2-p1 10.10.10.11

Server-LAN cn2-p1-se 192.168.10.11

Storage-LAN cn2-p1-st 192.168.20.11

Pool NAS Node Interfaces

fas01-p1

Storage-LAN fas01-p1-st 192.168.20.14

A sample with a reduced output on a single pool configuration:

cn1:/opt/FlexFrame/bin # ff_pool_adm.pl --op list-all --list sid,group

Pool configurations

p1

Pool Groups

Linux

OS: SUSE Linux SLES-9.X86_64

Pool SIDs

D01

SAP Version: SAP-6.20 DB Version: Oracle-9

P01

SAP Version: SAP-6.20 DB Version: Oracle-9

Q01

Pools and Groups Changing Pool DNS Domain

76 Administration and Operation

4.5 Changing Pool DNS Domain

Changing a pools DNS configuration can be done using the maintenance tool ff_pool_adm.pl. The DNS configuration is written to

/FlexFrame/volFF/pool-<pool name>/pooldata/config/etc/resolv.conf .

The DNS configuration is newly configured, therefore all domain names and ser-

vers to be used for this pool have to be given in the options. Any modification in resolv.conf, done by the customer, is lost.

Synopsis

ff_pool_adm.pl --op dns --name <pool name>

--dns <dns domain>[,<dns server ip>

[,<dns server ip>] ]

[--dns_search_list <list of domains>]

Options

--op dns

Adds or changes DNS configuration of a pool.

--name <pool name>

Name of pool the DNS configuration has to be changed. Use ff_pool_adm.pl –op list-all to get a list of currently configured pools

(see 4.4).

--dns <dns domain>[,<dns server ip>]

DNS domain name and servers to be used for this pool.

A dns_server_ip is only allowed if a default router of this pool is already specified.

--dns_search_list <list of domains>

List of DNS domain names to be used for this pool. At most six domain names are

allowed. For technical reasons the first DNS domain name of this list always must be

the DNS domain name given in option dns.

Example

cn1:~ # ff_pool_adm.pl --op dns --name pool3

--dns pool3.wdf.fsc.net

DNS domain successfully changed at LDAP and AN images.

Changing Pool Default Router Pools and Groups

Administration and Operation 77

4.6 Changing Pool Default Router

Add or change a pool's default router with opcode defrouter, remove a pool's default

router with opcode rem-defrouter. The default router may be necessary to reach

external DNS servers. Within all cases a default router has to have an IP address

matching the client IP network (. The default router is used to transfer network traffic

between the local IP networks and external IP networks.

Synopsis

ff_pool_adm.pl --op defrouter --name <pool name>

--defrouter <default router ip>

ff_pool_adm.pl --op rem-defrouter --name <pool name>

Options

--op defrouter

Adds or changes default router configuration of a pool.

--op rem-defrouter

removes default router configuration of a pool.

--name <pool name>

Name of pool the default router configuration has to be changed and removed

respectively.

Use ff_pool_adm.pl –op list-all to get a list of currently configured pools

(see 4.4).

--defrouter <default router ip>

The default router to be used for this pool to communicate with other, non local

networks. The IP address has to match to client pool network.

Example

cn1:~ # ff_pool_adm.pl --op defrouter --name pool3 --defrouter 192.168.100.200

Defaultrouter successfully changed at LDAP and AN images.

4.7 Adding a Group to a Pool

A pool may have more than one group. To add a group to a pool, the maintenance tool

/opt/FlexFrame/bin/ff_poolgroup_adm.pl can be used. To associate a group to

a pool, some parameters have to be defined on the command line. Groups are used with

the FlexFrame Autonomous Agents.

Pools and Groups Removing Pool Group

78 Administration and Operation

In the ADMINPOOL only a group named SPARE is allowed. This group acts exclusively

as a container for all types of spare nodes. The group is added to a pool in the LDAP

database.

Synopsis

ff_poolgroup_adm.pl --op add --pool <pool name>

--group <group name>

Options

--op add

Adds a group to a pool.

--pool <pool name>

Name of pool the group should be added to.

--group <group name>

The name of the group to be added.

4.8 Removing Pool Group

To remove a group of a pool, use the maintenance tool /opt/FlexFrame/bin/ff_

poolgroup_adm.pl. You have to define the pool and the group name on the command

line.

The group is removed from a pool in the LDAP database.

Synopsis

ff_poolgroup_adm.pl --op rem --pool <pool name>

--group <group name>

Options

--op rem

Removes a group from a pool.

--pool <pool name>

Name of the pool the group should be removed from.

--group <group name>

The name of the group to be removed.

Changing Group Assignment of Application Nodes Pools and Groups

Administration and Operation 79

4.9 Changing Group Assignment of Application Nodes

Change the assigned pool group of an Application Node with

/opt/FlexFrame/bin/ff_an_adm.pl. Command line arguments are the Application

Node name for which the pool group should be changed and the name of the new pool

group. The command changes the pool group of Application Node in the LDAP database.

The configuration of FA Agents currently has to be changed manually.

Synopsis

ff_an_adm.pl --op group --name <node name> --group <group name>

Options

--op group

Changes pool group of Application Node.

--name <node name>

Name of the Application Node the pool group has to be changed.

--group <group name>

The name of the pool group the Application Node should be assigned to. Use

ff_pool_adm.pl --op list-all to get available pool groups (see 4.4).

4.10 Changing Group and Pool Assignment of Application Nodes

There is currently no maintenance tool to do this. The recommended way is to remove an

Application Node and add it with the new pool and group name.

4.11 Hosts Database

It may become necessary to have additional entries in the hosts database. Those entries

may be required by 3rd-party products installed on customized Application Node images.

The hosts database is stored in LDAP. To maintain the additional host entries use the command ff_hosts.sh.

You cannot remove names or addresses which are essential to the FlexFrame

landscape. Each pool has its own hosts database.

Therefore you have to maintain each pool individually.

Pools and Groups Hosts Database

80 Administration and Operation

4.11.1 Script: ff_hosts.sh

This tool allows the administrator to list, add and delete host names or aliases to the

LDAP database. Note: When adding host names with an IP address that matches one of

the pool's network segments or the Control LAN segment, the list of IP addresses for that

segment gets extended by the IP address of this host name to prevent automatic

allocation of the same IP address by other FlexFrame tools.

Only one option of -l, -a or -r can be used at one time.

Synopsis

ff_hosts.sh [-d] -p <pool name> [{-l|-a <ip> -n <host name>|-r

<host name>}]

Options

-d.

This option will log debug information

-l

List all the hosts entries for the pool as provided by option -p.

-p <pool name>

Pool where the host name should be added to.

-a <ip>

Add the IP address <ip>. Has to be used together with option -n. If an entry with

the IP address <ip> already exists, the name provided will be added as an alias.

-n <host name>

Host name <host name> will be added to the list.

-r <host name>

Deletes the host name or alias of name. The host name cannot be deleted if it has

additional aliases. Remove the aliases first.

Examples

The following example will list all additional hosts entries for pool poolname created

using this tool:

cn1:~ # ff_hosts.sh -l -p poolname

Hosts Database Pools and Groups

Administration and Operation 81

The following example will add a host newhost with the IP address 1.2.3.4 to the pool

poolname:

cn1:~ # ff_hosts.sh -p poolname -a 1.2.3.4 -n newhost

The following example will remove the hosts entry for newhost:.

cn1:~ # ff_hosts.sh -p poolname -r newhost

User and Groups AdministrationCreate, Modify, Delete, or List User(s) for Application Nodes

82 Administration and Operation

5 User and Groups Administration

5.1 Create, Modify, Delete, or List User(s) for Application Nodes

Synopsis

/opt/FlexFrame/bin/ff_user_adm.pl

--op add –-user <user name> --pool <pool name>

[–-group {<group name> | <group id>},...]

[–-uid <uid number>]

[--home <home directory>] [–-pass <password>]

[--shell <login shell>] [-–gecos <comment>]

[--shadowmin <number of days>] [--shadowmax <number of days>]

[--shadowwarn <number of days>]

[--shadowinactive <number of days>]

[--shadowexpire <number of days]

/opt/FlexFrame/bin/ff_user_adm.pl

--op mod –-user <user name> --pool <pool name>

[–-pass <password>] --shell <login shell> [-–gecos <comment>]

[--shadowmin <number of days>] [--shadowmax <number of days>]

[--shadowwarn <number of days>]

[--shadowinactive <number of days>]

[--shadowexpire <number of days>]

[--home <home directory>]

[--group {<primary group name> | <primary group id>}]

[--uid <uid number>]

/opt/FlexFrame/bin/ff_user_adm.pl

--op rem –-user <user name> --pool <pool name>

/opt/FlexFrame/bin/ff_user_adm.pl

--op list –-user <user name> --pool <pool name>

/opt/FlexFrame/bin/ff_user_adm.pl

--op list-all --pool <pool name>

Options

--op add –-user <user name> --pool <pool name>

Creates a new user in a given pool.

Create, Modify, Delete, or List User(s) for Application NodesUser and Groups Administration

Administration and Operation 83

--op mod –-user <user name> --pool <pool name>

Modifies a user in a given pool.

--op rem –-user <user name> --pool <pool name>

Deletes a user in a given pool.

--op list –-user <user name> --pool <pool name>

Displays information of a user in a given pool.

--op list-all --pool <pool name>

Displays information of all users in a given pool.

--group <group name> | <group id>,...

A comma separated list of group names/ids the user belongs to. The first group is

taken as primary group. Default: gid number 1 (=other).

--group <primary group name> | <primary group id>

The affiliation to the primary group can be changed with this option.

--uid <uid number>

A specific uid number, if desired.

Default: One higher than the highest used uid number.

SAP recommends usage of uid numbers >= 3600.

In connection with the mod option the existence of the uid is not checked.

--home <home_directory>

Home directory of the user. If this option is not used the home directory is set to

/FlexFrame/pooldata/home/<user> and created if it does not exists.

The following rules are valid if the option is used:

<home_directory>::= <name-1>/<name-2> | <name-3>

Using ‗<name-1>/<name-2>‘ or ‗<name3>‘ creates a directory in

/FlexFrame/pooldata/<appropriate_dir> if the appropriate directory

does not exist and sets the home directory to

‗/FlexFrame/pooldata/<appropriate_dir> ‘.

<home_directory>::= /home_sap/<name-1> | / <name- 2>/ <name- 3>

Using ‗/home_sap/<name-1>‘ sets the home directory to

‗/home_sap/<name-1>‘ and creates the directory if it does not exist. Using

‗/<name-2>/<name-3>‘ sets the home directory to‘/<name-2>/<name-3>‘

but does not create the appropriate directory

--pass <password>

User and Groups AdministrationCreating, Modifying, Deleting or Listing Group(s) for Application Nodes

84 Administration and Operation

Login password for the user. Default: password.

--shell <login shell>

Login shell of the user. Default: /bin/csh. In connection with the mod option the

existence of the login shell is not checked.

--gecos <comment>

Some comment used for the user's gecos attribute. Default: Normal user. This

replaces the previous used option comment.

--shadowmin <number of days>

Min. number of days until the password can be changed.

--shadowmax <number of days>

Max. number of days until the password have to be changed.

--shadowwarn <number of days>

Number of days you get a warning message before the password expires.

--shadowinactive <number of days>

Number of days of inactivity that is allowed for this account.

--shadowexpire <number of days>

Number of days after which the account expires.

5.2 Creating, Modifying, Deleting or Listing Group(s) for Application Nodes

This command enables you to extend the operating system group entries stored in the

LDAP database. The modification or deletion of group entries is restricted to those entries

which are created with ff_group_adm.pl.

Synopsis

ff_group_adm.pl --op add –-name <osgroup name> --pool <pool name>

[–-gid <group id>] [--member <member, ...>]

[--gmember <member, ...>] [-–text <description>]

[--force] [--fname <filename>]

ff_group_adm.pl --op mod –-name <osgroup name> --pool <pool name>

[–-gid <group id>] [--member <member, ...>]

[--gmember <member, ...>] [--force]

[--fname <filename>]

Creating, Modifying, Deleting or Listing Group(s) for Application NodesUser and Groups Administration

Administration and Operation 85

ff_group_adm.pl --op rem –-name <osgroup name> --pool <pool name>

[--member <member, ...>] [--gmember <member, ...>]

[--fname <filename>]

ff_group_adm.pl --op list –-name <osgroup name> --pool <pool name>

ff_group_adm.pl --op list-all [–-name <osgroup name>]

[--pool <pool name>]

ff_group_adm.pl --help

User and Groups AdministrationCreating, Modifying, Deleting or Listing Group(s) for Application Nodes

86 Administration and Operation

Options

--op add

Creates a new group in a given pool.

--op mod

Modifies a group in a given pool.

--op rem

Deletes a group in a given pool.

--op list

Displays information of group entries in a given pool.

--op list-all

Displays information of all group entries in a given pool or in all pools.

--name <osgroup name, ...>

A comma separated list of group names.

--pool <pool name>

Name of the pool the group should belong to.

--gid <gid number>

A certain GID number, if desired. If a list of groups is specified, it is the start GUID

number for the first group. For the next group names the value is always incremted

by one. If the increment means a conflict with an already existing GUID the script

looks for the next free GUID. You can modify the GID with operation mod, but you are responsible for the effects.

It is also not possible to use a list of groups if you want to change a certain GID.

--member <member, ...>

A comma separated list of user names which should belong to this group. No check

is done if the user really exists.

--gmember <member, ...>

A comma separated list of user names which is inserted into the group members list of a group. This option requires a modified flexframe.schema file. Group

members usually used with a DB2 installation.

--text <description>

A user specific description of the group entry.

Creating, Modifying, Deleting or Listing Service(s) for Application NodesUser and Groups Administration

Administration and Operation 87

--force

The add operation continues with the next group if any add operation before

terminates unexpectedly.

--fname <filename>

Name of the file to store the used LDIF statements for this request.

--help

Shows the usage of the command.

5.3 Creating, Modifying, Deleting or Listing Service(s) for Application Nodes

This command enables you to extend the service entries stored in LDAP database. The

modification or deletion of service entries is restricted to those entries which are created

with ff_services_adm.pl.

Synopsis

ff_services_adm.pl --op add –-name <service name, ...>

--pool <pool name> [–-port <port number>]

[--prot {udp|tcp}] [-–text <description>]

[--force] [--fname <filename>]

ff_services_adm.pl --op mod –-name <service name>

--pool <pool name> [–-port <port number>]

[--prot {udp|tcp}] [--force]

[--fname <filename>]

ff_services_adm.pl --op rem –-name <service name, ...>

--pool <pool name>

ff_services_adm.pl --op list –-name <service name, ...>

--pool <pool name>

ff_services_adm.pl --op list-all [–-name <service name>]

[--pool <pool name>]

ff_services_adm.pl --help

Options

--op add

Creates new services in a given pool.

User and Groups AdministrationCreating, Modifying, Deleting or Listing Service(s) for Application Nodes

88 Administration and Operation

--op mod

Modifies a service in a given pool.

--op rem

Deletes services in a given pool.

--op list

Displays information of service entries in a given pool.

--op list-all

Displays information of all service entries in a given pool or in all pools.

--name <service name>,...

A comma separated list of service names. In case of operation mod you cannot use a list of services.

--pool <pool name>

Name of the pool the services should belong to.

--port <port number>

A certain port number, if desired. If a list of services is specified it is the start port

number for the first service. For the next service names the value is always

incremted by one.

--prot {udp|tcp}

Specifies the used protocol (default tcp).

--text <description>

A user specific description of the service entry.

--force

The add operation continues with the next service if any add operation before

terminates unexpectedly.

--fname <filename>

Name of the file to store the used LDIF statements for this request.

--help

Shows the usage of the command.

Administration and Operation 89

6 Pool-independent Spare Nodes

Before FlexFrame version 4.2 a group specific failover concept for Application Nodes was

used, i.e. if a node in a certain pool group failed, the FA Agents searched another

Application Node in this group with enough free capacity to take over the service of the

failed node. As of FlexFrame version 4.2 the new concept of pool-independent spare

nodes was introduced:

If no adequate spare node is found in the group of the failed node, the FA Agents can

search a spare node in the special pool ADMINPOOL, which mostly is called adminpool

(except in cases, where this name has already been reserved). The ADMINPOOL is a

reservoir of spare nodes for all pools of a FlexFrame landscape, and it must not serve as

a normal production pool.

A spare node in the ADMINPOOL is ready for takeover by the FA Agents only if

it has been booted at least once!

To move a spare node from the ADMINPOOL to its target pool and target group, the option --op move of ff_an_adm.pl has been implemented. Furthermore StorMan has

been integrated into the Control Node image to support the switch of SAN configurations

for spare nodes. The following is an overview of the details of this concept.

6.1 Creation of Spare Nodes in the ADMINPOOL

To create spare nodes, you have to perform the following steps using the Management

Tool or the corresponding administration scripts (see parentheses):

1. Create the ADMINPOOL (ff_pool_adm.pl –-op add; see section 4.1)

2. Create the pool group SPARE within this pool (ff_poolgroup_adm.pl –-op add;

see section 4.7)

3. Add the necessary spare nodes to this group (ff_an_adm.pl –-op add; see

section 7.2)

4. Create the os images with ff_new_an.sh and boot them.

6.2 Moving of a Spare Node

To move a spare node from the ADMINPOOL to the desired target pool and target group, use ff_an_adm.pl –op move (see section 7.5).

Pool-independent Spare Nodes Listfunction for Spare Nodes

90 Administration and Operation

6.3 Listfunction for Spare Nodes

With this function you can select spare nodes from group SPARE in the ADMINPOOL,

which satisfy several conditions, such as OS type HBA type, or 10Gbit.

Synopsis

ff_an_adm.pl --op list-spare [--os-type {LINUX}]

[--hba-type <hba type>] [--10GB]

Options

--op list-spare

Lists the spare node.

--os-type {LINUX}

Os type the selected node must have.

--hba-type <hba type>

HBA type the selected node must have.

--10GB

Specifies whether the node supports 10Gbit facility.

6.4 Handling Pool-independent Spare Nodes with FA Agents

If no adequate spare node is found in the group of the failed node, the FA agents can

search a spare node in the ADMINPOOL. For more information see the ―FA Agents -

Installation and Administration― manual.

Administration and Operation 91

7 Application Nodes Administration

Manual Application Node administration would be very complex and error-prone. The

script /opt/FlexFrame/bin/ff_an_adm.pl does the major changes and supports

adding, changing, removing and listing of Application Nodes. Below, each action is des-

cribed in detail.

In this document, you often will find console output, configuration data and

installation examples which are based on earlier FlexFrame versions. Please

keep in mind that these are examples and may look slightly different on the new

operating systems introduced in FlexFrame 5.1A.

7.1 Listing Application Nodes

7.1.1 Displaying Information on a Specific Application Node

Synopsis

ff_an_adm.pl --op list --name <node name> [--cmdline]

Options

--op list

Lists the configuration details of an Application Node.

--name <node name>

The name of the Application Node to be listed.

--cmdline

The command line that used to create the Application Node is displayed at the end

of the node listing. This may be used as a preparation step to rename the AN.

The output is structured in sections: hardware, software, network, pool and group and

LAN interface connections or virtual machine.

Hardware

This section contains information about system type, shutdown facility with IP

address and host name, mac addresses and on blade ser-vers the chassis and slot

ID.

Software

This section contains information on OS type, vendor and version and the root

image path.

Application Nodes Administration Listing Application Nodes

92 Administration and Operation

Network

This section lists the VLAN ID, IP address and host name of all configured networks,

sorted by LAN segments.

Pool and Group

This section lists the names of the assigned pool and group.

LAN Interface Connections

In case link aggregates are configured, this section identifies the aggregate and its

ports. Each used switch port is shown with the switch group ID, switch ID and port ID

(cabinet ID, switch blade ID and port on blade servers) for the common LANs

(Storage, Server and Client LAN) and the Control LAN (if used).

Listing Application Nodes Application Nodes Administration

Administration and Operation 93

Examples

The command displays detailed information on the selected Application Node. The output

differs between blade servers and all others.

Example of a PRIMERGY blade server output:

cn1:~ # ff_an_adm.pl --op list --name bx924

Configuration details of node bx924

Hardware

System: BX924S2

10GBit: yes

Shut.Facil.: Mgmt Blade bx91-co (172.20.13.40)

iRMC Facil.: iRMC bx924-co (172.20.13.41)

MAC Addr.: 00:23:8b:97:ca:02 00:23:8b:97:ca:03

Cabinet/Slot: 1 / 1

Software

OS: SuSE / Linux / SLES-11.x86_64 (Vendor / Type / Version)

OS Path: fil1na-pool2-st:/vol/volFF/os/Linux/FSC_5.1A00-000.SLES-11.x86_64

Network

VlanID Host IP Hostname

Storage LAN: 2021 172.20.21.51 bx924-st

Server LAN: 2022 172.20.22.51 bx924-se

Client LAN: 2020 172.20.20.51 bx924-1

Pool and Group

Pool: pool2

Group: group1

LAN Interface Connections

Cabinet SwBlade Port

data NIC-1: 3 1 1

data NIC-2: 3 2 1

Application Nodes Administration Listing Application Nodes

94 Administration and Operation

Example of a PRIMERGY rack server output:

cn1:~ # ff_an_adm.pl --op list --name rx36-1

Configuration details of node rx36-1

Hardware

System: RX300S6

10GBit: no

Shut.Facil.: IPMI rx36-1-co (172.20.13.23)

MAC Addr.: 00:c0:9f:dc:3a:da 00:c0:9f:dc:3a:db

Software

OS: SuSE / Linux / SLES-11.x86_64 (Vendor / Type / Version)

OS Path: fil1na-st:/vol/volFF/os/Linux/FSC_5.1A00-000.SLES-11.x86_64

Network

VlanID Host IP Hostname

Storage LAN: 2011 172.20.11.1 rx36-1-st

Server LAN: 2012 172.20.12.1 rx36-1-se

Client LAN: 2010 172.20.10.1 rx36-1

Pool and Group

Pool: pool1

Group: group1

LAN Interface Connections

SwGroup Switch Port

data NIC-1: 1 2 5

data NIC-2: 1 1 5

IPMI NIC-1: 1 2 24

Listing Application Nodes Application Nodes Administration

Administration and Operation 95

7.1.2 Displaying Information of all Application Nodes

Synopsis

ff_an_adm.pl --op list-all [--pool <pool name>]

Options

--op list-all

Lists all configured Application Nodes.

--pool <pool name>

The name of the pool of which the Application Nodes have to be listed.

Example

cn1:/opt/FlexFrame/bin # ff_an_adm.pl --op list-all

Nodes sorted by pool, group and name

Pool pool1

Pool Group group1

bx924

System: BX924S2

MAC Addr.: 00:23:8b:97:ca:02 00:23:8b:97:ca:03

Cabinet/Slot: 1 / 1

OS: SuSE / Linux / SLES-11.x86_64 (Vendor / Type / Version)

OS Path: fil1na-pool2-st:/vol/volFF/os/Linux/FSC_5.1A00-000.SLES-11.x86_64

VlanID Host IP Hostname

Storage LAN: 2021 172.20.21.51 bx924-st

Server LAN: 2022 172.20.22.51 bx924-se

Client LAN: 2020 172.20.20.51 bx924-1

Pool pool2

Pool Group group2

bx36-1

System: RX300S65

MAC Addr.: 00:c0:9f:dc:3a:da 00:c0:9f:dc:3a:db

OS: SuSE / Linux / SLES-11.x86_64 (Vendor / Type / Version)

OS Path: fil1na-st:/vol/volFF/os/Linux/FSC_5.1A00-000.SLES-11.x86_64

VlanID Host IP Hostname

Storage LAN: 2011 172.20.11.1 rx365-1-st

Server LAN: 2012 172.20.12.1 rx365-1-se

Client LAN: 2010 172.20.10.1 rx365-1

Application Nodes Administration Adding Application Nodes

96 Administration and Operation

The output of list-all is less detailed than the list output. It is used to get an

overview. It shows the Application Nodes sorted by pool and group in alphabetical order.

For each node the system type, the cabinet and slot ID (if node is a blade server), the OS

type, vendor, version, the root image path, the main IP addresses, host names and the

MAC addresses are listed.

7.2 Adding Application Nodes

This section describes how to provide the required information for adding a new AN to an

existing FlexFrame environment. See also section 11.5.1.

You have to define some parameters at the command line. They are used to configure

switch ports, to create the boot information and the OS image.

Synopsis

ff_an_adm.pl --op add --type <system type> --name <node name>

--pool <pool name> --group <group name>

--swgroup <switch group id>

--mgmtswgroup <switch group id>

--mac <mac addresses> ospath <path to os image> <--

new-image>

[--host <ip host number> [,<cntl-LAN ip host number>,

<2nd cntl-LAN ip hostnumber>]]

[--slot <cabinet/slot>]

[--hba <list of hba names][--hba-type <hba type>]

[--sw <multipath software name>] [--10gbit]

[--port switch:port,switch:port,switch:port,

[switch:port]]

[--esx <esxi node name>]

[--vcpus <number of virtual cpus>]

[--vmem <virtual machine memory size]

[--force]

Adding an Application Node changes the exports file on the common volFF NAS

system. Temporary exports on this NAS system will be gone after running

ff_new_an.sh. or ff_an_adm.pl with option –new_image, Be sure not to

have temporary exports.

Adding Application Nodes Application Nodes Administration

Administration and Operation 97

Options

--op add

Adds an Application Node and displays some information about processing steps.

--type <system type>

Specifies the product name and type. Call ff_an_adm.pl without any parameter to

get a list of supported system types. A system type of ESXVM denotes a virtual

machine on an ESXi host, and a virtual machine with the same name is implicitly

created on the denoted ESXi host

--name <node name>

The name of the Application Node. This name has to be unique in the FlexFrame

system. All interface names are based on this node name. The name may consist of

2 up to 13 characters (letters, numbers and dashes). It must start with a letter and

must not end with a dash.

--pool <pool name>

The name of the pool this node should belong to. See usage (call ff_an_adm.pl

without any parameter) to get a list of currently configured pools.

--group <group name>

The name of the pool group this node is a member of. A group must consist of

Application Nodes of the same OS image version and should be of the same capacity (CPU, Memory etc.). Use command ff_pool_adm.pl with op mode list

or list-all to get the pool groups.

--swgroup <switch group id>

Defines the switch group the Application Nodes data NIC 1 and 2 are connected to.

This information is necessary to assign and configure switch ports. Call

ff_an_adm.pl without any parameter to get a list of currently configured switch

group IDs.

--mac <mac addresses>

Add here both MAC addresses of the data NICs used for booting. Use the colon

separated hex notation for each MAC address. Concatenate them with a comma.

The MAC address syntax is a six colon separated hex values, eg. 00:e0:00:c5:19:41. For an Application Node with type ESXVM, an appropriate

MAC address is generated by the script.

Application Nodes Administration Adding Application Nodes

98 Administration and Operation

--ospath <path to os image>

Defines the OS image to be used. Add the relative path to

/FlexFrame/volFF/os/ as seen from the Control Node. See usage (call

ff_an_adm.pl without any parameter) to get a list of available OS paths.

--host <ip host number>[,<cntl-LAN ip host number>,

2nd cntl-LAN ip host number>]

Host part to be used to build IP addresses for the three networks. If necessary host

part(s) to be used to build IP addresses for the management network can be added,

separated by commas. If this option is omitted, the script uses free host numbers to

calculate the IP addresses.

--slot <cabinet_id/slot>

With PRIMERGY server blades use this option to define the cabinet and slot ID of

the server blade. New cabinets have to be defined with the

/opt/FlexFrame/bin/ff_bx_cabinet_adm.pl command.

For models that occupy more than one slot in width (e.g.BX630 S2 quad),

the part of the server blade that occupies the slot with the highest slot

number is called the master. The master is usually the rightmost slot and its

slot ID has to be chosen as the master slot ID.

--hba <list of hba names>

Specifies a list of symbolic SAN HBA names. The HBA name may consist of

alphanumerical characters and dashes. Separate HBA names with a comma.

--hba-type <hba type>

Specifies the type of the HBAs on this node. This option can be important for sear-

ching an adequate spare node from the ADMINPOOL (see chapter 6). It is

necessary to know what the meaning is: With this option all relevant information of

HBAs, which have influence on the OS image, must be coded. This can be, for

example, the speed of a HBA. The administrator must know what information is

important for the spare node selection. He can code this information, as he wants it,

he is free in choosing the value of this option, but he must do it obeying a norm,

which is unique for the whole FlexFrame.

--sw <multipath software name>

Specifies the name of the SAN multipath software to be used. See usage for a list of

known and accepted software names.

--10gbit

Specifies that the node is used with 10 Gigabit data NICs. The specification can be

omitted if the nodes system type only allows usage with 10 Gigabit data NICs.

Adding Application Nodes Application Nodes Administration

Administration and Operation 99

--mgmtswgroup <switch group id>

Defines the switch group the Application Nodes management interface (IPMI) is

connected to. If omitted the effective switch group is computed as follows: If the

switch group specified with --swgroup is no NEXUS switch group, that switch group

is used; else the switch group the NEXUS switch management interfaces are

connected is used. Call ff_an_adm.pl --help to get a list of currently configured

switch group IDs.

--new-image

Creates a new os image (need not call ff_new_an.sh !).

--port switch:port,switch:port,switch:port,[switch:port]

Defines the switch ports to be used. First two tuples are for data NIC 1 and 2 and

allocated in the switch group defined with --swgroup. The following tuples are for

mgmt NICs and allocated in the effective switch group for management interfaces. If

--10gbit is specified, the first two ports must be 10 Gigabit capable ports.

--esx <esxi node name>

When adding a node with type ESXVM, this parameter is mandatory and specifies

the name of the ESXi host where the virtual machine must be created. An ESXi host

with this name must already exist in FlexFrame and must be configured for

FlexFrame usage. For details refer to chapter 7.8 "Administrating ESX servers and

Virtual Machines".

--vcpus <number of virtual cpus>

Specifies the number of virtual CPUs when creating a virtual machine. The default is

2 CPUs.

--vmem <virtual machine memory size in MB>

Defines the memory size in MB when creating a virtual machine. The default is 8

GB. The available resources of the ESXi host must be taken into account when

choosing a value for this parameter. For details refer to section 7.8.9 ―Virtual

Machine Properties and ESXi Resources―.

--force

Specifies that the memory usage of the ESXi host may be overcommitted when

creating a virtual machine. Default is to deny creation of a virtual machine on a host if the total vmem of the virtual machines on this host, including the new one, exceeds

the memory size of the host

Application Nodes Administration Adding Application Nodes

100 Administration and Operation

Examples

Output for a blade server:

cn1:/opt/FlexFrame/bin # ff_an_adm.pl --op add --type BX924S1

--name bx1-6 --pool pool1 --group bx900

--ospath Linux/FSC_5.1A00-000.SLES-11.x86_64 --host 1 --slot 1/6

--mac 00:C0:9F:99:E6:CC,00:C0:9F:99:E6:CD

update swblade 1/1 configuration

Notice: Update will take about 1 minute.

update swblade 1/2 configuration

Notice: Update will take about 1 minute.

If not reported any error all precautions are done to create

application nodes os image. To do this call:

ff_new_an.sh -n bx1-6

Creating and customizing an image may take some minutes.

Don't get anxious.

Output for a non-blade server:

cn1:~ # ff_an_adm.pl --op add --name rx300-1 --type RX300S6 --pool pool2 --group

group1 --swgroup 1 --ospath Linux/FSC_5.1A00-000.SLES-11.x86_64 --mac

00:15:17:2d:ab:a8,00:15:17:2d:ac:02

update switch 1/1 configuration

Notice: Update will take about 1 minute.

Connect your systems LAN interfaces to named switch ports:

SwitchGroup / Switch / Port LAN Interface

1 / 1 / 6 data NIC-1

1 / 2 / 6 data NIC-2

1 / 1 / 24 IPMI NIC-1

If not reported any error all precautions are done to create

application nodes os image. To do this call:

ff_new_an.sh -n rx300-1

Creating and customizing an image may take some minutes.

Don't get anxious.

The script first checks all arguments and aborts with error messages in case of errors.

Then it fetches free IP addresses and switch ports. The switch ports are reconfigured to

match requirements, the LDAP data is created and a netboot file is written. The netboot

file is used by ff_new_an.sh to create Application Node images and extend the NAS

system's exports list. At the end you get a cabling advice and instructions how to call ff_new_an.sh script to finish the Application Node creation.

Removing Application Nodes Application Nodes Administration

Administration and Operation 101

7.3 Removing Application Nodes

You only have to specify the node name to be removed at the command line. All switch

ports will be unconfigured and the boot information and OS image are deleted. For an

Application Node of type ESXVM, the virtual machine is also destroyed.

Synopsis

ff_an_adm.pl --op rem --name <node name>

Options

--op rem

Removes an application node.

--name <node name>

The name of the Application Node to be removed. Use operation mode list-all

to get all configured Application Nodes (see 7.1.2).

Example

cn1:/opt/FlexFrame/bin # ff_an_adm.pl --op rem --name rx300-1

7.4 Renaming Application Nodes

Changing of node names may be necessary for various reasons. Removing and adding

the node may result in changes of network cabling while renaming does not. With

renaming a node an alternate os image path may be specified. Renaming is not

supported for virtual machine Application Nodes.

Synopsis

ff_an_adm.pl --op rename --name <node name>

--newname <new node name>

[--ospath <path to os image>] [--remove-image]

Removing an Application Node results in direct deletion of its image, removal of

its LDAP entries as well as disabling the respective switch ports. Please make

sure you really want to remove the Application Node (AN) when calling the

script, the script does not ask for further confirmation.

Removing an Application Node changes the exports file on the common volFF

NAS system. Temporary exports on this NAS system will be gone after running

ff_an_adm.pl. Please make sure not to have temporary exports.

Application Nodes Administration Moving Application Nodes between Pools

102 Administration and Operation

Options

--op rename

Changes the node name of the specified Application Node.

--name <node name>

Current name of node to be changed.

--newname <new node name>

New name of the node.

--ospath <path to os image>

Os image path the renamed node should use.

--remove-image

Removes the old unused image of node.

Example

cn1:~ # ff_an_adm.pl --op rename --name node4 --newname node5

update LDAP

................

If not reported any error all precautions are done to create

a new application node os image. To do this call:

ff_new_an.sh -n node5

Creating and customizing an image may take some minutes.

Don't get anxious.

7.5 Moving Application Nodes between Pools

Moving of Application Nodes from one pool to another pool may often be necessary, spe-cially for pool-independent spare nodes (see chapter 6). The opcode move serves to

satisfy this requirement. While performing the move action, the node gets new Vlans in its

new target pool, but changes of network cabling are not necessary. This operation is not

supported for virtual machine Application Nodes.

Moving Application Nodes between Pools Application Nodes Administration

Administration and Operation 103

Synopsis

ff_an_adm.pl --op move --name <node name> [--to-pool <pool name>]

[--to-group <group name>] [--newname <new node name>]

[--to-host <ip host number>[,<cntl-LAN ip host

number>,<2nd cntl-LAN ip host number>]]

[--to-ospath <path to os image>]

[--failed-host <node name>] [--new-image]

[--clean-swap]

Options

--op move

Moves the Application Node given by --name.

--name <node name>

Current name of the node to be moved.

--to-pool <pool name>

Name of target pool.

--to-group <group name>

Name of target group.

--newname <new node name>

New name of the node in its new group and pool. The name may consist of 2 up to

13 characters (letters, numbers and dashes). It must start with a letter and must not

end with a dash.

--to-host <ip host number>[,<cntl-LAN ip host number>, <2nd cntl-

LAN ip host number>]

Host part of Ip address in the target pool.

--to-ospath <path to os image>

Os image path the moved node should use.

--new-image

Creates a new os image (need not call ff_new_an.sh !).

--clean-swap

Clean local disks of node on first boot of moved node. This parameter has only

meaning together with --new-image.

Application Nodes Administration Application Nodes and SAN

104 Administration and Operation

If you specify --failed-host, all other options except –name, --new-image and –

clean-swap are ignored. In this case all necessary information for the moved node in its

target pool is derived from the failed host.

7.6 Application Nodes and SAN

For adding or changing the entire list of SAN HBA (Host Bus Adapter) names, use operation mode hba-change. The names are symbolical names and may consist of

alphanumerical characters, dashes and underscores. The names have to be separated

by comma. For redundancy at least two names for two HBAs are needed.

To specify the HBA type, use the option --hba-type.

This option can be important for searching an adequate spare node from the

ADMINPOOL (see chapter 6). It is necessary to know what the meaning is: With this

option all relevant information of HBAs, which have influence on the OS image, must be

coded. This can be, for example, the speed of a HBA. The administrator must know what

information is important for the spare node selection. He can code this information, as he

wants it, he is free in choosing the value of this option, but he must do it obeying a norm,

which is unique for the whole FlexFrame.

ff_an_adm.pl --op hba-change --name <node name> --hba <hba list>

--hba-type <hba type>

To remove the entire list of HBA names use operation mode hba-rem. It removes the list

from nodes LDAP data. To remove only a single HBA, use operation mode hba-change.

ff_an_adm.pl --op hba-rem --name <node name>

To define the name of the available SAN multipath software, use operation mode

sansw-change. See usage for a list of known software names.

ff_an_adm.pl --op sansw-change --name <node name>

--sw <multipath software>

To remove the SAN multipath software name use operation mode sansw-rem. It

removes the name from nodes LDAP data.

ff_an_adm.pl --op sansw-rem --name <node name>

7.7 Administrating Blade Server Cabinets

Some network settings have to be changed to add or remove a blade server cabinet. The

script /opt/FlexFrame/bin/ff_bx_cabinet_adm.pl will simplify the administration

by doing LDAP changes automatically and preparing configurations to be done manually.

Administrating Blade Server Cabinets Application Nodes Administration

Administration and Operation 105

The script supports adding, moving, removing and listing of blade server cabinets as well

as modifying switch blade attributes. Each action is described in detail below.

7.7.1 Listing Blade Server Cabinets

7.7.1.1 Displaying Information on a Specific Blade Server Cabinet

Synopsis

ff_bx_cabinet_adm.pl --op list --name <cabinet name>

Options

--op list

Lists the configuration details of a blade server cabinet.

--name <cabinet name>

The name of the blade server cabinet to be listed.

The output is structured in sections: hardware, software, network, assigned pool and

group, switch ports.

Output example

PRIMERGY Cabinet 1 (cab1)

System Type: BX900S1

Management Blade

Hostname / IP Address: cab1-co 195.40.224.75

Integrated LAN Switch Ports: SwitchGroup SwitchID PortID

1 1 8

1 2 8

Switch Blade

SwitchID Type Switch name Hostname IP Address

1 10GbE-18/8-F cab1-swb1 cab1-swb1-co 195.40.224.76

2 10GbE-18/8-F cab1-swb2 cab1-swb2-co 195.40.224.77

Switch Blade Port Integrated LAN Switch Port

Switch Blade ID PortID <--> SwitchGroup SwitchID PortID

1 19 2 1 1/14

1 20 2 2 1/13

2 19 2 2 1/14

2 20 2 1 1/13

As seen from the sample above, the cabinet ID and name, the cabinet system type, the

management blade and the switch blades are listed.

Application Nodes Administration Administrating Blade Server Cabinets

106 Administration and Operation

For the management blade the host name, the IP address and both FlexFrame integrated

LAN switch ports are displayed.

The switch blade information shows the switch and host name, the IP address and the

switch blade port to FlexFrame integrated LAN switch port connections, structured by

switch blade ID.

7.7.1.2 Displaying Information on all Configured Blade Server Cabinets

Synopsis

ff_bx_cabinet_adm.pl --op list-all

Option

--op list-all

Lists all configured blade server cabinets.

Output example

PRIMERGY Cabinets

1 (cab1) BX600S3

Management Blade: cab1-co / 195.40.224.75

Switch Group ID: 1

Server Blades (by slot id)

1 (blade1) BX924S1

Pool / Group: pool1 / PROD

2 (blade2) BX924S1

Pool / Group: pool1 / PROD

3 (blade3) BX924S1

Pool / Group: pool1 / PROD

4 (blade4) BX924S1

Pool / Group: pool2 / DEV

5 (blade5) BX924S1

Pool / Group: pool2 / DEV

For each cabinet the ID, the cabinet name, the management host name and IP address

and the server blades are displayed.

Each server blade is shown with its slot ID and name, the system type and the pool and

group it belongs to.

Administrating Blade Server Cabinets Application Nodes Administration

Administration and Operation 107

7.7.2 Adding Blade Server Cabinets

This section describes how to provide the required information for adding a new blade

server cabinet to an existing FlexFrame environment. You have to define some

parameters at the command line. They are used to configure switch ports and to create

the switch blade configurations.

Synopsis

ff_bx_cabinet_adm.pl --op add --type <system type>

--name <cabinet name>

--swgroup <switch group id>

[--swblades <switch blade type>]

[--swblogin <login name>]

[--swbpwd <password>]

[--host <ip host parts>] [--10gbit]

[--mgmtswgroup <switch group id>]

[--uplinkportcnt <nic count>]

[--port <sw:port>[,...]]

Options

--op add

Adds a blade server cabinet.

--type <system type>

PRIMERGY blade system type e.g. BX600S3, BX900S1.

Call ff_bx_cabinet_adm.pl without any parameter to get a list of supported

system types.

--name <cabinet name>

Name of the subsystem (cabinet). It is used to generate a name for the management

blade and has to be unique within FlexFrame. It may consist of 2 up to 13 characters

(letters, numbers, dashes). It must start with a letter and must not end with a dash.

--swgroup <switch group id>

Switch group number the cabinet has to be connected to (physically). Call

ff_bx_cabinet_adm.pl without any parameter to get a list of currently configured

switch group IDs.

Application Nodes Administration Administrating Blade Server Cabinets

108 Administration and Operation

--mgmtswgroup <switch group id>

Defines the switch group the cabinets management blade interfaces are connected

to. If omitted the effective switch group is computed as follows: If the switch group

specified with --swgroup is no NEXUS switch group that switch group is used; else

the switch group the NEXUS switch management interfaces are connected is used.

Call ff_bx_cabinet_adm.pl --help to get a list of currently configured switch

group IDs

--swblades <switch blade type>

The type of switch blades. For valid types see the usage. For the default switch

blades this option may be omitted.

--swblogin <login_name>

Name used to login into the switch blades. If this option is omitted, the login name of

the switch group is used.

--swbpwd <password>

Password used to login into the switch blades. If this option is omitted, the login

password of the switch group is used.

--host <ip host parts>

Host parts to be used to build IP addresses for the control LAN networks. If this

option is omitted the script uses free host numbers to calculate the IP addresses.

Order of comma separated host numbers: first host number for both management

blades and then one for each switch blade.

--10gbit

Use 10 Gigabit switch blade ports as uplink. The specification can be omitted if only

10 Gigabit switch blade ports are available

--uplinkportcnt <nic count>

Set the NIC count of switch blades uplink channel to given value. Defaults to 2.

--port <sw:port>[,...]

Specifies the ports that are to be used on the connected switch group(s).

‗sw‘ is the switch id within the switch group. ‗port‘ is the port id on that switch.

Multiple ports are specified by a comma separated list. The first and second

specification are for the management blades, followed by specifications for the first

switch blades. Specifications for the second switch blade constitute the rest of the

list. If the list contains an empty specification, the port is calculated as if the --port

option was omitted (e.g. --port ,,1:13,2:13,1:14,2:14 to specify ports only

for the switch blades).

Administrating Blade Server Cabinets Application Nodes Administration

Administration and Operation 109

Output example

At the end of the output, the command displays further instructions.

Configure the ManagementBlade with the control lan settings:

control lan IP address: 195.40.224.75

control lan name: cab1-co

to interoperate correctly with the FA Agents.

Interconnect the ManagementBlades and SwitchBlades with the switches noted below:

SwitchGroup SwitchID Port Mgmt/SwitchBlade

1 1 8 ManagementBlade 2

1 1 11 SwitchBlade 1 Port 12

1 1 12 SwitchBlade 2 Port 12

1 2 8 ManagementBlade 1

1 2 11 SwitchBlade 1 Port 11

1 2 12 SwitchBlade 2 Port 11

Uploads of initial SwitchBlade configurations have to be done manually.

See the document(s)

'Quick Start Hardware FlexFrame for SAP – PRIMERGY SWB 1GbE-10/6-Q'

in the doc/hwinfo section of the Service CD for details.

The files to be uploaded are named:

SwitchBlade Blade Type File Path

1 1GbE-10/6-Q /tftpboot/swblade-2-1.config

2 1GbE-10/6-Q /tftpboot/swblade-2-2.config

Look at "/opt/FlexFrame/network/wiring-BX600-cab1.txt" to get a copy of this

message.

Set up the management blade initially with name and IP address listed by the output as

seen above. Use console redirection of management blade to connect to the console of

the switch blades, and upload configuration as described by FlexFrame Installation

Guide.

Finally, plug in the network cables according to the wiring plan given by the command

output.

7.7.3 Removing Blade Server Cabinets

You only have to specify the ID or the name of the cabinet that is to be removed. All

FlexFrame integrated LAN switch ports will be unconfigured. Removing a blade server

cabinet requires removing of all of its server blades.

Application Nodes Administration Administrating Blade Server Cabinets

110 Administration and Operation

Synopsis

ff_bx_cabinet_adm.pl --op rem --id <cabinet id> | --name <cabinet

name>

Options

--op rem

Removes a blade server cabinet.

Administrating Blade Server Cabinets Application Nodes Administration

Administration and Operation 111

--id <cabinet id>

Specifies the cabinet ID of the cabinet to be removed. Use the

list-all option to get the ID (see section 7.7.3).

--name <cabinet name>

Specifies the name of the cabinet.

Output examples

If there are any server blades configured for this cabinet, an error message is displayed.

ERROR: there are server blades configured for this cabinet.

To remove the cabinet, remove application nodes (server blades) first.

Use command ff_an_adm.pl to do this.

Use the list operation mode to list the configured server blades. You have to remove

them before you can remove the cabinet they are in.

If no server blades are configured for this cabinet, the command displays a summary at

the end.

If not reported any warnings or errors the cabinet was removed

from LDAP and integrated LAN switches.

The cabinet has been removed successfully from LDAP and the FlexFrame integrated

LAN switch ports used by the cabinet have been reconfigured to default.

7.7.4 Changing Switch Blade Type

Within service cases it may be necessary to change the type of a switch blade due to a

defective part replacement. Only switching blades can be used for a type change.

To change the switch blade type the cabinet, the switch blade id and the switch blade

type have to be specified.

Synopsis

ff_bx_cabinet_adm.pl --op swb-change --id <cabinet id>

--swbid <switch blade id>

--swbtype <switch blade type>

Application Nodes Administration Administrating Blade Server Cabinets

112 Administration and Operation

Options

--op swb-change

Selects the operation mode. Change the type of a switch blade.

--id <cabinet id>

Specifies the cabinet ID of the cabinet. Use the list-all option to get the ID.

--swbid <switch blade id>

Specifies the switch blade ID. The ID is the slot number of the selected switch blade.

--swbtype <switch blade type>

Defines the new type of the switch blade. See usage for the currently supported

types.

Output example

Switch type of switch blade 2 was successfully changed from "1GbE-10/6-Q" to "1GbE-

10/2+4-C" at LDAP database.

The switch blade type was changed at LDAP database. To get the initial configuration use operation mode swb-config of this program. It will display instructions how to

upload the configurations too.

7.7.5 Changing Switch Blade Name

On adding a new cabinet, the name of the switch blade is like the cabinet name with a

slot extension. In some cases the names of the switch blades have to be changed to

match naming conventions.

Synopsis

ff_bx_cabinet_adm.pl --op swb-name --id <cabinet id>

--swbid <switch blade id>

--swbname <switch blade name>

Options

--op swb-name

Selects the operation mode. Change the name of a switch blade.

--id <cabinet id>

Specifies the cabinet ID of the cabinet. Use the list-all option to get the ID.

Administrating Blade Server Cabinets Application Nodes Administration

Administration and Operation 113

--swbid <switch blade id>

Specifies the ID of the switch blade. The ID is the slot number of the selected switch

blade.

--swbname <switch blade name>

Defines the new name of the switch blade.

Output example

If not reported any warnings or errors the hostname was successfully changed at

switch blade, hosts files and LDAP.

As noted by the program the name of the switch will be changed at /etc/hosts of both

control nodes, the LDAP database and at least the hostname and, if possible, the SNMP

sysname at the selected switch blade itself.

7.7.6 Changing Switch Blade Password

If on adding a cabinet the switch blade password could not be derived from the switch

group and was not specified in the command, a default password is assigned. This is not

secure.

Synopsis

ff_bx_cabinet_adm.pl --op swb-passwd --id <cabinet id>

--swbid <switch blade id>

--swbpwd <password>

Options

--op swb-passwd

Selects the operation mode. Change the login password of a switch blade.

--id <cabinet id>

Specifies the subsystem (cabinet) ID of the cabinet. Use the list-all option to get the

ID.

--swbid <switch blade id>

Specifies the switch blade ID. The ID is the slot number of the selected switch blade.

--swbpwd <password>

Defines the new login and enable password of the switch blade.

Application Nodes Administration Administrating Blade Server Cabinets

114 Administration and Operation

Output example

If not reported any warnings or errors the password was successfully changed at

switch blade and LDAP.

As noted by the program the password will be changed at the LDAP database and the

selected switch blade. At the switch blade the login password and the enable password

are changed and have to be the same.

7.7.7 Getting Switch Blade Configuration

In case of a service issue it may be necessary to get a switch blade configuration, which

has to be uploaded manually. Use this command to get the actual necessary

configuration according to LDAP for a specific switch blade.

Synopsis

ff_bx_cabinet_adm.pl --op swb-config --id <cabinet id>

--swbid <switch blade id>

Options

--op swb-config

Selects the operation mode. Create the switch blade configuration.

--id <cabinet id>

Specifies the cabinet ID of the cabinet. Use the list-all option to get the ID.

--swbid <switch blade id>

Specifies the ID of the switch blade. The ID is the slot number of the selected switch

blade.

Output example

If not reported any warnings or errors the switch configuration was successfully

created and stored into /tftpboot/swblade2.config.

To upload the initial switch blade configuration see detailed description at

"FlexFrame(TM) Installation Guide" chapter "SwitchBlade Configuration".

This hint is additionally stored at:

/var/log/FlexFrame/tmp/swb-config_bx_cabinet/todo.txt

The configuration file is stored directly to /tftpboot. Upload of configuration is

described using TFTP which uses /tftpboot as top level directory. A detailed

instruction can be found in the „Installation Guide―.

Administrating Blade Server Cabinets Application Nodes Administration

Administration and Operation 115

7.7.8 Change Switch Blade Uplink

More than the two default ports can be used for the uplink to switch group. To change the

count of ports used by the uplink link aggregates, including switch port configuration, use

the operation mode swb-uplink of ff_bx_cabinet_adm.pl. On switch group and

switch blades the appropriate link aggregates will be expanded by new ports until count

of given ports is reached.

Synopsis

ff_bx_cabinet_adm.pl --op swb-uplink --id <cabinet id>

--uplinkportcnt <nic count>

Options

--op swb-uplink

Change the count of NICs of uplink link aggregate for each switch blade of cabinet.

--id <cabinet id>

Defines the cabinet to change.

--uplinkportcnt <nic count>

Set the NIC count of switch blades uplink channel to given value. Defaults to 2

Example

cn1:~ # ff_bx_cabinet_adm.pl --op swb-uplink --id 1 –-uplinkportcnt 6

update LDAP

...

update swblade 1/1 configuration

Notice: Update will take about 1 minute.

..........

update swblade 1/2 configuration

Notice: Update will take about 1 minute.

..........

Interconnect additional NICs between SwitchBlades and the switches of

SwitchGroup 1 as noted below:

SwitchBladeID/Port SwitchID/Port

1 / 13 <==> 1 / 9

1 / 14 <==> 2 / 9

1 / 15 <==> 1 / 11

1 / 16 <==> 2 / 11

2 / 13 <==> 1 / 14

2 / 14 <==> 2 / 14

Application Nodes Administration Administrating Blade Server Cabinets

116 Administration and Operation

2 / 15 <==> 1 / 16

2 / 16 <==> 2 / 15

Unless any errors are reported follow instructions above to extend

switch blade channels to switch group.

Look at "/opt/FlexFrame/network/wiring-channel-extend-BX600-1.txt"

to get a copy of this message.

7.7.9 Move a Blade Cabinet to Another Switch Group

On growing installations sometimes the load of a switch group is split onto two groups.

While doing this sometimes entire blade cabinets have to be moved to another switch group. To simplify this action use the operation mode move of

ff_bx_cabinet_adm.pl.

Synopsis

ff_bx_cabinet_adm.pl --op move --id <cabinet id>

--swgroup <switch group id>

[--mgmtswgroup <switch group id>]

[--port <sw:port>[,...]]

Options

--op move

Change the switch group of a cabinet, including switch blade but not server blade

connections.

--id <cabinet id>

Defines the cabinet to be moved.

--swgroup <switch group id>

The switch group id the cabinet should be moved to.

--mgmtswgroup <switch group id>

Defines the switch group the cabinet‘s management blade interfaces should be

moved to. If omitted the effective switch group is computed as follows: If the switch group specified with --swgroup is no NEXUS switch group that switch group is

used; else the switch group the NEXUS switch management interfaces are

connected to is used.

Administrating ESX Servers and Virtual Machines Application Nodes Administration

Administration and Operation 117

--port <sw:port>[,...]

Specifies the ports that are to be used on the connected switch group(s).

'sw' is the switch id within the switch group. 'port' is the port id on that switch.

Multiple ports are specified by a comma separated list. The first and second

specification are for the management blades, followed by specifications for the first

switch blades. Specifications for the second switch blade constitute the rest of the

list. If the list contains an empty specification, the port is calculated as if the -port

option was omitted (e.g. -port ,,1:13,2:13,1:14,2:14 to specify ports only for the

switch blades).

Example

cn1:~ # ff_bx_cabinet_adm.pl --op move --id 1 –-swgroup 2

update LDAP

...

update swblade 1/1 configuration

Notice: Update will take about 1 minute.

..........

update swblade 1/2 configuration

Notice: Update will take about 1 minute.

..........

Interconnect NICs between MgmtBlades,SwitchBlades and the switches of

the new SwitchGroup 2 as noted below:

SwGrpID/SwitchID/Port Mgmt/SwitchBlade

2 / 1 / 1 <==> master ManagementBlade

2 / 1 / 2 <==> SwitchBlade 1 Port 11

2 / 1 / 3 <==> SwitchBlade 2 Port 11

2 / 2 / 1 <==> slave ManagementBlade

2 / 2 / 2 <==> SwitchBlade 1 Port 12

2 / 2 / 3 <==> SwitchBlade 2 Port 12

Unless any errors are reported follow instructions above to move

cabinet to new switch group.

Look at "/opt/FlexFrame/network/wiring-cabinet-move-BX600-1.txt"

to get a copy of this message.

Application Nodes Administration Administrating ESX Servers and Virtual Machines

118 Administration and Operation

7.8 Administrating ESX Servers and Virtual Machines

Instead of being used directly as Application Nodes, PRIMERGY servers may also be

used in FlexFrame as ESXi servers. An ESXi server can host a number of virtual

machines that are used as Application Nodes.

VMware ESXi and VMware ESX are "bare-metal" hypervisors that form the foundation of

VMware vSphere. The VMware vSphere 4 product line supports three different types of

hypervisors - ESX classic, ESXi installable and ESXi embedded. In FlexFrame only ESXi

installable and ESXi embedded are supported. However, the terms "ESX" and "ESXi" are

both used in the FlexFrame documentation and code to denote the VMware hypervisor

and always mean "ESXi" unless explicitly stated otherwise.

ESXi servers and virtual machines used as Application Nodes can be predefined with the

FlexFrame Management Tool and put into operation during the FlexFrame installation

process, or added to a FlexFrame configuration later with FlexFrame administrative tools

as described in section 7.8.1 below. The FlexFrame administrative tools also offer

functions to display or adjust the configuration as shown in detail in the following sections.

Besides the FlexFrame tools which focus mainly on the FlexFrame specific aspects of

ESXi servers and virtual machines, there are several ways to access vSphere

components, such as the vSphere Client or the vSphere Command-Line Interface (vCLI).

These tools can be used in addition to the FlexFrame tools, but certain actions must be

avoided to insure proper operation of the FlexFrame system. A short overview of this

topic is given in section 7.8.10.

7.8.1 Getting Started with ESX Servers and VMs

This section gives an overview of the steps needed to add an ESX server and virtual

machines used as Application Nodes on the ESX server to a FlexFrame system when

using FlexFrame administrative tools.

1. Add the ESX server to the FlexFrame configuration.

cn1:~ # ff_esx_adm.pl --op add --name <node name> --type <system type> ...

Depending on system type, more options are needed when entering this

command.

See section 7.8.3 for details.

2. Install/Boot the ESXi server.

The necessary actions depend on the used hypervisor type. For embedded ESXi,

the server is already installed on an internal USB device and the only action is to

boot it from this device. For ESXi installable, install the ESXi server from its CD or

DVD (an ISO-Image of this software is also contained on the FlexFrame Service

Administrating ESX Servers and Virtual Machines Application Nodes Administration

Administration and Operation 119

DVD) and then boot it from local hard disk. Refer to the "Install and boot ESXi"

section of the manual "Installation of a FlexFrame Environment" for details and to the

section "Setup the <server type> for an ESXi Server installation" in the appropriate

FlexFrame for SAP - HW Characteristics Quickguide.

3. Do a minimal configuration of the ESXi server using its Direct Console User

Interface. This includes setting a password and selecting the network adapters for

the ESXi server's network connection. Refer to the "ESXi Preconfiguration on the

Server Console" section of the manual "Installation of a FlexFrame Environment" for

details.

4. Complete the ESX server configuration.

cn1:~ # ff_esx_adm.pl --op complete-config --name <node name>

--user <user name> --pass <password>

See section 7.8.4 below for details.

Application Nodes Administration Administrating ESX Servers and Virtual Machines

120 Administration and Operation

5. No explicit creation of virtual machines is needed.

Instead of that, simply use the well-known ff_an_adm.pl --op add command

with a system type of ESXVM to specify that the "hardware" of the new Application

Node is a virtual machine on an ESXi host. In addition to its usual functions such as

reserving IP addresses or creating LDAP data, the script then creates a virtual

machine with the same name as the Application Node on the ESXi host specified on

the command line. The memory size and number of CPUs for the virtual machine

are set to default values, but can also be specified on the command line (options --vmem and --vcpus). To select appropriate values for these parameters, the

available resources of the ESXi host must be taken into account, as well as the

intended use of the Application Node. See section 7.8.9 ‖Virtual Machine Properties

and ESXi Resources‖ on page 134 for details on this topic.

cn1:~ # ff_an_adm.pl --op add --name <node name> --type ESXVM

--pool <pool name> --group <group name>

--ospath <path to os image> --esx <node name> ...

Proceed as you would do it for any other Application Node.

Create a new personalized boot image for the Application Node (if not implicitly done

using the --new-image operand of ff_an_adm.pl):

cn1:~ # ff_new_an.sh -n <node name>

Then boot the node by calling:

cn1:~ # ff_ponpoff.pl --op power-on --name <node name>

7.8.2 ESX related global FlexFrame parameters

Besides the ESX servers and virtual machines, some FlexFrame global parameters must

be taken into account when planning a FlexFrame system using VMware functionality.

They are usually set with the FlexFrame Management Tool, but can be also set with

administrative tools in some special cases. As an exception, the Control LAN default

router that can be set for usage by the ESXi hosts cannot be set with the Management

Tool up to now.

7.8.2.1 System Code for ESX Servers and VMs

The FlexFrame system code for ESX servers and virtual machines is a numeric value

between 0 and 63 that is used to generate MAC addresses for FlexFrame virtual

machines and to build names of some ESX related resources such as Port Groups and

Datastores. Its purpose is to differentiate these objects belonging to different FlexFrame

systems used in a common environment, for example if using a common vCenter Server

or a not completely separated ethernet network.

Administrating ESX Servers and Virtual Machines Application Nodes Administration

Administration and Operation 121

The FlexFrame system code is usually set with the FlexFrame Management Tool and

must not be changed for a system with already configured ESX servers and virtual

machines.

The ability to set the System Code using the tool ff_esx_adm.pl is mainly provided for

cases when the FlexFrame system has been upgraded from an older version without

VMware support or if the initial FlexFrame setup has been done without VMware usage.

Use a different system code for each FlexFrame system in your environment.

Do not change the system code for an existing FlexFrame system with

configured ESX servers and virtual machines.

Synopsis

ff_esx_adm.pl --op set-syscode --syscode <system code>

Options

--op set-syscode

Sets the FlexFrame system code for ESX servers and virtual machines.

--syscode <system code>

Numeric value between 0 and 63 to be used as FlexFrame system code

7.8.2.2 vCenter Server

More complex administrative actions and functionalities, as well as the centralized view

and monitoring of the VMware Infrastructure are outside the scope of FlexFrame. For

these functions, a vCenter Server can be used.

If FlexFrame ESX servers are administrated by a vCenter Server, the name and IP

address of the vCenter Server have to be made known to FlexFrame. This is usually

done with the FlexFrame Management Tool, but can also be done using the tool ff_esx_adm.pl, for example in cases when the FlexFrame system has been upgraded

from an older version without VMware support or if the initial FlexFrame setup has been

done without VMware usage. You can also remove the previously set vCenter Server

information, e.g. if the information set with the FlexFrame Management Tool is not

correct, and you can set the authentication data for the vCenter Server in FlexFrame, so

that the FlexFrame tools can access this server.

Application Nodes Administration Administrating ESX Servers and Virtual Machines

122 Administration and Operation

Synopsis

ff_esx_adm.pl --op vcenter-add --vc-name <vcenter name>

[--vc-ip <vcenter ip>] [--vc-in-cntllan]

ff_esx_adm.pl --op vcenter-rem

ff_esx_adm.pl --op vcenter-auth --user <user name>

--pass <password>

Options

--op vcenter-add

Add vCenter Server hostname and IP address to FlexFrame.

--op vcenter-rem

Remove vCenter Server information from FlexFrame.

--op vcenter-auth

Set authentication data for vCenter Server in FlexFrame.

--vc-name <vcenter name>

Defines the name of the vCenter Server with --op vcenter-add.

--vc-ip <vcenter-ip>

Defines the IP address of the vCenter Server with --op vcenter-add. It can be

an address in the FlexFrame Control LAN or an address outside FlexFrame, that is

reachable from the Control Nodes.

If the --vc-ip option is omitted, and --vc-in-cntllan is specified, a free

address from Control LAN is selected.

--vc-in-cntllan

Specifies that the vCenter IP address is in the Control LAN. If no address is given,

advises the script to select a free address from Control LAN.

--user <user name>

Specifies the user name for accessing the vCenter Server.

--pass <password>

Specifies the password for accessing the vCenter Server.

Administrating ESX Servers and Virtual Machines Application Nodes Administration

Administration and Operation 123

If using vCenter Server in FlexFrame, the names and IP addresses of the ESX servers

must be resolvable on a DNS server that is accessible for the vCenter Server.

The FlexFrame ESX servers must be added manually to vCenter Server using their

hostnames as they are known in FlexFrame. Add the ESX servers to a Datacenter called

FlexFrame or, if more than one FlexFrame system is administrated with the same

vCenter Server, use a distinct Datacenter called FlexFrame-<i> for each FlexFrame

system, where i is the FlexFrame system code explained in section 7.8.2.1.

7.8.2.3 Control LAN Default Router for ESX Servers

If FlexFrame ESX servers are administrated by a vCenter Server, it may also be

necessary to use a Control LAN router that is set on the ESX servers as default router.

This can be done using the tool ff_esx_adm.pl with operation mode defrouter.

With this operation mode, the address of a default router for the Control LAN can be set

or removed in the FlexFrame configuration, including the DHCP configuration. The ESX

servers get this address via DHCP together with the IP address for ESXi management.

Synopsis

ff_esx_adm.pl --op defrouter --defrouter <default router ip | 0>

Options

--op defrouter

Add or remove the address of the Control LAN default router to the FlexFrame

configuration

--defrouter <default router ip | 0>

Defines the IP address of the Control LAN default router. Specify a Control LAN

address to add a default router for the ESXi hosts or to change the current setting, or

use the value 0 to remove it.

Application Nodes Administration Administrating ESX Servers and Virtual Machines

124 Administration and Operation

7.8.3 Adding ESX Servers

This section describes how to add an ESXi server to an existing FlexFrame environment.

Synopsis

ff_esx_adm.pl --op add --type <system type> --name <node name>

--swgroup <switch group_ id> --mac <mac addresses>

[--host <ip host number>[,<IPMI ip host number>

[,<2nd IPMI ip host number>]]]

[--slot <cabinet_id/slot>] [--10gbit]

[--mgmtswgroup <switch group id>]

[--port switch:port,switch:port,switch:port

[,switch:port]]

Options

--op add

Adds an ESX server node and displays some information about processing steps.

--type <system type>

Specifies the product name and type. Call ff_esx_adm.pl --help to get a list of

supported system types

--name <node name>

The name of the ESX server node. This name has to be unique for the entire

FlexFrame system. All interface names are based on this node name. The name has

to be lowercase and consists of 2 up to 13 characters (letters, digits and dashes). It

must start with a letter and must not end with a dash.

--swgroup <switch group id>

Defines the switch group the ESX server node is connected to. This information is

necessary to assign and configure switch ports. Call ff_esx_adm.pl --help to

get a list of currently configured switch group IDs.

For blade servers, the --swgroup option may be omitted, as the switch group is

already defined with the blade cabinet.

Administrating ESX Servers and Virtual Machines Application Nodes Administration

Administration and Operation 125

--mac <mac addresses>

Add here the MAC addresses of the server's data NICs.They are used to configure

DHCP. Use the colon separated hex notation for each MAC address. Concatenate

the two MAC addresses with a comma. A MAC address syntax example is

00:e0:00:c5:19:41.

For blade servers, the --mac option may be omitted. The MAC addresses are then

fetched via SNMP from the management blade.

--host <ip host number>[,<IPMI ip host number>

[,2nd IPMI ip host number>]]

Host part to be used to build the IP addresses for ESXi management and,

depending on system type, for IPMI, all of them in the Control LAN.

If the --host option is omitted, the script uses free host numbers to calculate the IP

addresses.

--slot <cabinet_id/slot>

With PRIMERGY server blades use this option to define the cabinet and slot ID of

the server blade. New cabinets have to be defined with the

ff_bx_cabinet_adm.pl command.

--mgmtswgroup <switch group id>

Defines the switch group the ESX servers management interface (IPMI) is

connected to. If omitted the effective switch group is computed as follows: If the

switch group given with --swgroup is no NEXUS switch group that switch group is

used else the switch group the NEXUS switch management interfaces are

connected is used. Call ff_esx_adm.pl --help to get a list of currently

configured switch group IDs.

--10gbit

Specifies that the node is used with 10 Gigabit data NICs. The specification can be

omitted if the nodes system_type only allows usage with 10 Gigabit data NICs.

--port switch:port,switch:port,switch:port,[switch:port]

Defines the switch ports to be used. First two tuples are for data NIC 1 and 2 and allocated in the switch group defined with --swgroup. The following tuples are for

mgmt NICs and allocated in the effective switch group for management interfaces. If

--10gbit is specified, the first two ports must be 10 Gigabit capable ports

If --port is not specified, the script assigns free ports according to an internal

algorithm.

Application Nodes Administration Administrating ESX Servers and Virtual Machines

126 Administration and Operation

Example

cn1:~ # ff_esx_adm.pl --op add --name rx300s5-esx1 --type RX300S5 --

swgroup 2

--mac 00:19:99:48:e2:aa,00:19:99:48:e2:ab

update LDAP

.........

update switch 2/1 configuration

Notice: Update will take about 1 minute.

restart cluster service dhcpd

stopping dhcpd (timeout=20) . dhcpd:done . done.

starting dhcpd (timeout=20) .. dhcpd:done . done.

Connect your systems LAN interfaces to named switch ports:

LAN Interface SwGroup / Switch / Port

data NIC-1 2 / 2 / 2

data NIC-2 2 / 1 / 2

IPMI NIC-1 2 / 2 / 24

The script first checks all arguments and aborts with an error message in case of errors. It

then fetches free IP addresses and switch ports or checks whether the given ones can be

used for the node. LDAP data are created and the switch ports are reconfigured to match

the requirements. The DHCP service on the control nodes is reconfigured to assign the

ESXi management IP address in the Control LAN based on the given MAC addresses.

At the end, you get a cabling advice if the node must be connected to a switch group.

How to continue

The next step is to do the cabling according to the advice if applicable.

Then proceed with installing/booting ESXi and preconfiguring the ESXi server using the

Direct Console User Interface (see section "Getting Started with ESX Servers and VMs"

on page 118 and the sections "Install and boot ESXi" and "ESXi Preconfiguration on the

Server Console" of the manual "Installation of a FlexFrame Environment").

Finally, complete the ESX server configuration as shown in the next section.

7.8.4 Completing ESX Server Configuration

This section describes how to complete the ESXi server configuration.

Synopsis

ff_esx_adm.pl --op complete-config [--name <node name>]

--user <user name> --pass <password>

Administrating ESX Servers and Virtual Machines Application Nodes Administration

Administration and Operation 127

Options

--op complete-config

Completes the ESX server configuration and stores the authentication data needed

to access the ESX server in FlexFrame for subsequent access of the server.

--name <node name>

Name of the ESX server node. If the name is omitted, the operation is done for all

defined ESX server nodes. This possibility is intended for usage during the

installation process.

--user <user name>

Specifies the user name for accessing the ESX server.

--pass <password>

Specifies the password for accessing the ESX server.

Example

cn1:~ # ff_esx_adm.pl --op complete-config --name rx300s5-esx1 --user

root

--pass password

The script accesses the ESXi server using the given authentication data and completes

the ESXi server configuration to match the FlexFrame needs, e.g. the FlexFrame Control

Nodes are set as NTP servers and as trap targets for SNMP. If it is already known for

which pools the ESXi server must be prepared, the corresponding Port Groups and

Datastores are created on the ESXi server (see section "ESX Servers and Pools" on

page 131 for details).

Finally, the script stores the authentication data needed to access the ESXi server to

enable the subsequent access of the server by FlexFrame scripts.

7.8.5 Removing ESX Servers

To remove an ESX server node from the FlexFrame configuration, the only needed

parameter is the node name. The node is removed from the LDAP database, the hosts

file and the DHCP configuration. The switch ports used by this node are also

unconfigured. An ESX server cannot be removed if there are still virtual machine

Application Nodes that use this ESX server in the FlexFrame configuration.

Please make sure that you really want to remove the ESX server node, as the

script does not ask for confirmation.

Application Nodes Administration Administrating ESX Servers and Virtual Machines

128 Administration and Operation

Synopsis

ff_esx_adm.pl --op rem --name <node name>

Options

--op rem

Removes an ESX server node from the FlexFrame configuration.

--name <node_name>

Name of the ESX server node.

Example

cn1:~ # ff_esx_adm.pl --op rem --name rx300s5-esx1

7.8.6 Displaying Information about ESX Servers and VMs

Using the script ff_esx_adm.pl, one can get an overview about all ESX servers, or

detailed information about a specific ESX server including its virtual machines.

With ff_esx_vm.pl, it is possible to get an overview of all FlexFrame virtual machines,

irrespective of the involved ESX server.

Synopsis

ff_esx_adm.pl --op list-all

ff_esx_adm.pl --op list --name <node name> [--cmdline]

Options

--op list-all

Displays an overview of all ESX server nodes and also shows ESX related global

FlexFrame parameters, such as the vCenter Server name and address or the

FlexFrame system code or ESXi servers and virtual machines.

--op list

Displays detailed information about a specific ESX server, including its virtual

machines.

--name <node name>

Name of the ESX server node.

Administrating ESX Servers and Virtual Machines Application Nodes Administration

Administration and Operation 129

--cmdline

Used with --op list, the command line that can be used to recreate the ESX

server node is displayed at the end of the node listing.

Synopsis

ff_esx_vm.pl --op list-all

Options

--op list-all

Displays an overview of all FlexFrame virtual machines, irrespective of the ESX

host.

Examples

Overview of all ESX servers and global information:

cn1:~ # ff_esx_adm.pl --op list-all

Global information

vCenter Server Hostname: vcenter-co

vCenter Server IP: 10.10.13.94

FlexFrame Systemcode: 1

Control LAN Default Router: 10.10.13.99

ESX Nodes sorted by name

bx31-esx

Node Type: BX620S4

Cabinet/Slot ID: 3/1

ESX mgmt IP/Hostname: 10.10.13.18 / bx31-esx

Mac Addr.: 00:1b:24:2d:ab:03 00:1b:24:2d:ab:04

bx33-esx

Node Type: BX620S4

Cabinet/Slot ID: 3/3

ESX mgmt IP/Hostname: 10.10.13.93 / bx33-esx

Mac Addr.: 00:1b:24:2d:a0:01 00:1b:24:2d:a0:02

rx300s5-esx1

Node Type: RX300S5

ESX mgmt IP/Hostname: 10.10.13.16 / rx300s5-esx1

Mac Addr.: 00:19:99:48:e2:aa 00:19:99:48:e2:ab

Application Nodes Administration Administrating ESX Servers and Virtual Machines

130 Administration and Operation

Detailed information about a specific ESX server:

cn1:~ # ff_esx_adm.pl --op list --name rx300s5-esx1

Configuration details of ESX node rx300s5-esx1

Hardware

System: RX300S5

10GBit: No

Shut.Facil.: IPMI rx300s5-esx1-co (10.10.13.27)

Mac Addr.: 00:19:99:48:e2:aa 00:19:99:48:e2:ab

ESX management interface (Control LAN)

Host IP: 10.10.13.16

Hostname: rx300s5-esx1

LAN Interface Connections

LAN Interface SwGroup / Switch / Port

data NIC-1 2 / 2 / 2

data NIC-2 2 / 1 / 2

IPMI NIC-1 2 / 2 / 24

ESX resources

Product Name: VMware ESXi 5.0.0 build-474610

Memory Size: 8179 MB

CPU Cores: 8

License Name: VMware vSphere 5 Enterprise Plus

List of Virtual Machine - Application Nodes:

Name Pool Group State CPUs Memory

lin-vm13 pool1 p1-vms Off 2 2048

lin-vm21 pool2 p2-vms Off 2 2048

Overview of all FlexFrame virtual machines:

cn1:~ # ff_esx_vm.pl --op list-all

List of Virtual Machine - Application Nodes registered in LDAP

VM-AN Name Pool Group ESX Host

lin-vm11 pool1 p1-vms bx31-esx

lin-vm13 pool1 p1-vms rx300s5-esx1

lin-vm2 pool1 p1-vms bx33-esx

lin-vm21 pool2 p2-vms rx300s5-esx1

lin-vm23 pool2 p2-vms bx33-esx

lin-vm39 pool3 p3-vms bx33-esx

Collecting information from ESX Hosts:

Administrating ESX Servers and Virtual Machines Application Nodes Administration

Administration and Operation 131

rx300s5-esx1 ... done

bx31-esx ... done

bx33-esx ... done

List of FlexFrame Virtual Machines found on available ESX servers

VM Name ESX Host State

lin-vm11 bx31-esx On

lin-vm13 rx300s5-esx1 Off

lin-vm2 bx33-esx On

lin-vm21 rx300s5-esx1 Off

lin-vm23 bx33-esx On

lin-vm39 bx33-esx Off

7.8.7 ESX Servers and Pools

An ESX server can host virtual machine Application Nodes from different pools. For each

pool the server needs some specific resources, such as a virtual machine Port Group for

each of the pool networks and two datastores (config and software) in the pool specific

volFF volume of the pool.

Usually, these resources are implicitly prepared when the first virtual machine Application

Node of a pool is created on an ESX server, so that there is no need to do an explicit pool

preparation of the ESX server.

However, if a FlexFrame virtual machine is moved to another ESX server by mechanisms

outside FlexFrame (see also section 7.8.10), these resources must be prepared in

advance. This can be done by using the add-pool operation of the ff_esx_adm.pl

script.

Synopsis

ff_esx_adm.pl --op add-pool --name <node name> --pool <pool name>

Options

--op add-pool

Prepares an ESX server for usage by virtual machine Application Nodes from given

pool.

--name <node name>

Name of the ESX server node.

--pool <pool name>

Name of the pool for which the ESX server must be prepared.

Application Nodes Administration Administrating ESX Servers and Virtual Machines

132 Administration and Operation

Example

cn1:~ # ff_esx_adm.pl --op add-pool --name rx300s5-esx1 --pool pool3

7.8.8 Special Functions for Virtual Machines

A virtual machine designated for use as a FlexFrame Application Node is usually created

when the Application Node is added to the FlexFrame configuration with ff_an_adm.pl

--op add. Likewise, the virtual machine is destroyed when the associated Application

Node is removed using ff_an_adm.pl --op rem.

Under special circumstances, it is necessary to create FlexFrame virtual machines for

Application Nodes already defined in LDAP, for example during the FlexFrame

installation process. For special purposes, it may also be helpful to destroy a FlexFrame

virtual machine while preserving its LDAP entry, or if the LDAP entry is missing, e.g. as

an effect of an Application Node removal that failed to destroy the virtual machine.

The script ff_esx_vm.pl provides these functionalities, as well as some other special

functions acting on virtual machines. This script is not intended for usual administrative

interactions with virtual machine Application Nodes. Use ff_an_adm.pl to administrate

these Application Nodes too in almost the same manner as other Application Nodes.

Synopsis

ff_esx_vm.pl --op create [--name <node name>]

[--esx <node name>]

[--vcpus <number of virtual cpus>]

[--vmem <virtual machine memory size]

[--force]

ff_esx_vm.pl --op destroy --name <nodename> [--force]

ff_esx_vm.pl --op move --name <node name>

--to-esx <node name>

ff_esx_vm.pl --op refresh-ldap [--name <node name>]

ff_esx_vm.pl --op list --name <node name>

ff_esx_vm.pl --op list-all

Administrating ESX Servers and Virtual Machines Application Nodes Administration

Administration and Operation 133

Options

--op create

Creates virtual machines that are already defined in the FlexFrame configuration

database (LDAP). If called without option --name, a virtual machine is created for

each Application Node with system type ESXVM found in LDAP. Otherwise, only the

Application Node with the specified name is considered. The virtual machines get

the same names as the Application Nodes. The target ESXi server, the number of

virtual CPUs and the memory size of the virtual machine are taken from LDAP by

default, but can be overridden using the options --esx, --vcpus and --vmem.

--op destroy

Destroys a FlexFrame virtual machine while preserving its LDAP entry or if the

LDAP entry is missing. If the LDAP entry is missing, the operation is started only if

the option --force is specified.

--op move

Moves a powered off FlexFrame virtual machine to another ESXi server.

--op refresh-ldap

Adjusts the LDAP information for a specific or for all virtual machine Application

Node(s) after changes done with mechanisms outside FlexFrame, e.g. when a

virtual machine has been moved to another ESXi host by VMware HA. For more

information about using other mechanisms on FlexFrame virtual machines see

section 7.8.10 below.

--op list

Displays information about a specific FlexFrame virtual machine.

--op list-all

Displays information about all FlexFrame virtual machines. See also section 7.8.6.

--name <node name>

The name is used to identify a specific node. If specified, an Application Node with

this name and system type ESXVM must exist in LDAP, except for the case when

the --force option is specified with --op destroy.

--esx <node name>

Specifies the node name of the ESXi host where the virtual machine must be

created. By default, the value from LDAP is used. When no <node name> is given,

the --esx option is used to select all VMs for creation for which the specified ESXi

host is preset in LDAP.

Application Nodes Administration Administrating ESX Servers and Virtual Machines

134 Administration and Operation

--vcpus <number of virtual cpus>

Defines the number of virtual CPUs when creating a virtual machine. By default, the

value from LDAP is used. The argument must be a number between 1 and 8.

Moreover, the available resources of the ESXi host must be taken into account. See

also section 7.8.9 below.

--vmem <virtual machine memory size>

Defines the memory size in MB when creating a virtual machine. By default, the

value from LDAP is used. The memory size must be a number between 256 and

261120 (=255GB). Moreover, the available resources of the ESXi host must be

taken into account. See also section 7.8.9 below.

--force

Specifies that the memory usage of the ESXi host may be overcommitted when

creating virtual machines. Default is to deny creation of virtual machines on a host if the total vmem of the virtual machines on this host, including the new ones, exceeds

the memory size of the host. With operation mode destroy, --force allows to

destroy a FlexFrame virtual machine even when no corresponding LDAP entry

exists.

--to-esx <node name>

This option is used with --op move to specify the name of the target ESXi server.

7.8.9 Virtual Machine Properties and ESXi Resources

In a vSphere environment, ESX/ESXi hosts provide resources for virtual machines.

Resources include CPU, memory, power, storage and network resources. In this section,

we focus on CPU and memory resources, as well as on storage requirements.

As a part of the virtual hardware properties, the memory size and the number of virtual

CPUs are defined when creating a virtual machine. In a FlexFrame environment, values

for these parameters are either predefined with the FlexFrame Management Tool for

each virtual machine Application Node and can be later overridden when creating the

virtual machine using the ff_esx_vm.pl script, or are defined when creating the

Application Node including its virtual machine with ff_an_adm.pl.

When choosing values for these parameters, a careful planning must be done that takes

into account which machines will be powered on on a given ESX server at the same time

and for what kind of SAP workload they are planned to be used.

VMware techniques allow to configure more memory for the virtual machines on an ESX

server than the available physical memory of the host. This feature is called memory

overcommit and basically assumes that a VM does not use all assigned memory and

therefore it can be shared with other VMs.

Administrating ESX Servers and Virtual Machines Application Nodes Administration

Administration and Operation 135

For SAP systems it is strongly recommended not to overcommit memory usage, because

SAP allocates memory permanently and does not release it again. Furthermore, a

minimum of 4GB should be used, and for Unicode systems, a minimum of 8GB.

You should also be aware that if the memory size defined for a virtual machine is not

sufficient for the workload, excessive Linux swapping in the virtual machine may occur

which has a negative impact on performance and leads to an increased I/O load on the

filer and network which may even disturb the correct operation of the system.

A virtual machine can be configured with up to eight virtual CPUs, but it cannot be

powered on if it has more CPUs than the number of logical processors of the host - that is

the number of physical processor cores if hyperthreading is disabled or two times the

number of physical processor cores if hyperthreading is enabled. Similarly as with

memory, VMware techniques allow to overcommit processor usage. When a host runs

multiple virtual machines that require more than the available CPU resources, the host

time-slices the physical processors across all virtual machines so that each virtual

machine runs as if it has its specified number of virtual processors. The CPU virtualization

adds varying overhead with performance implications.

SAP has successfully run performance tests in vSphere virtual machines (utilizing all

available virtual CPUs to 100%) which overcommitted the host system up to 200%. The

performance degradation inside the virtual machines was linear reciprocal to the

overcommitment. You may exceed the 200% overcommitment, but keep in mind, that the

performance of virtual machines in such a scenario is not guaranteed. In case of

performance problems, SAP can demand you to shutdown or pause other running virtual

machines to check if the overcommitment caused problems.

When creating virtual machines using FlexFrame scripts, the parameters given for the

memory size and the number of virtual CPUs (--vmem and --vcpus) are roughly

checked for plausible values and it is checked that the memory usage of the designated

ESX server is not overcommitted, but no further checks and resource evaluations are

carried out. This task is in the responsibility of the specialists that make the configuration

and sizing planning for the FlexFrame system and assumes a good knowledge of the

virtualization techniques and the requirements of the used SAP systems.

A FlexFrame virtual machine also needs some space on the pool specific volFF of the

pool to which the Application Node belongs. The most significant resources located here

are the file representing the virtual hard disk of the virtual machine (vmdk) and the file

used by the VMkernel to swap out the virtual machine‘s physical memory in certain

shortage cases (vswp). The size of the vmdk file is defined by the size of the virtual hard

disk, which is 36 GB for FlexFrame virtual machines. The size of the vswp file is

correlated to the memory size of the virtual machine. If there is not sufficient space

available, the creation of the virtual machine may fail or the virtual machine fails to start.

These space requirements must also be taken into account during the configuration and

sizing planning mentioned above.

Application Nodes Administration Administrating ESX Servers and Virtual Machines

136 Administration and Operation

7.8.10 Using vSphere Functions for FlexFrame Objects

While the FlexFrame tools focus mainly on the FlexFrame specific aspects of ESXi

servers and virtual machines, there are other ways to access vSphere components, such

as the vSphere Client or the vSphere Command-Line Interface (vCLI). Moreover, for

more complex administrative actions and functionalities, as well as the centralized view

and monitoring of the VMware Infrastructure, a vCenter Server can be used

These tools and functions are outside the scope of FlexFrame, and for information on

their usage please refer to the appropriate VMware documentation.

However, when using them for ESX servers and virtual machines known in FlexFrame, it

must be taken care not to disturb the FlexFrame functionality. Some special requirements

are:

1. Virtual machines for FlexFrame usage must be created using FlexFrame tools

2. Do not rename FlexFrame virtual machines

3. Do not destroy FlexFrame virtual machines while the associated Application Node

still exists. Removing the Application Node usually also destroys the virtual machine.

4. Do not change the virtual machine properties and devices. As an exception, you may

change the number of CPUs and the memory size of an powered off virtual machine,

but take into account the resources of the involved ESX server. After such changes,

please call ff_esx_vm.pl --op refresh-ldap to adjust the corresponding

LDAP settings, or use the script ff_ponpoff.pl to power on the virtual machine,

as this does an implicit refresh of the LDAP settings.

5. Do not move a FlexFrame virtual machine to an ESX server that is not part of the

same FlexFrame system. To move a powered off virtual machine to another ESX

server of the FlexFrame system, you can use ff_esx_vm.pl --op move.

Otherwise, you must also take care that the target ESX server is prepared for the

pool the virtual machine Application Node belongs to. After a move done with non-

FlexFrame tools, please call ff_esx_vm.pl --op refresh-ldap to adjust the

corresponding LDAP setting as soon as possible.

6. Do not suspend/resume FlexFrame virtual machines. This may result in a service

being executed twice when a service formerly running on the suspended virtual

machine has been moved to a spare node and then the other VM is resumed. To

enforce this requirement, the FlexFrame scripts create an alarm on vCenter server (if

known in FlexFrame, including authentication data) that powers off a virtual machine

when it is resumed.

For virtual machines created by FlexFrame on ESXi 5.0 hosts, a special attribute is

set which denies the suspend action on these virtual machines.

Script for Power on/off/Reboot of a Computing Node in FF4SApplication Nodes Administration

Administration and Operation 137

7.9 Script for Power on/off/Reboot of a Computing Node in FF4S

7.9.1 Synopsis

ff_ponpoff.pl --op power-on | power-off | reboot

--name <node name | @all>

--op power-on power-on the node(s)

power-off power-off the node(s)

reboot reboot the node(s)

--node <node name> execute the above actions for a single node <node name>

@all execute the above actions for all nodes of a FlexFrame

The script takes the shutdown configuration files from /opt/myAMC/vFF/vFF_<pool

name>/config/myAMC_FA_SD_Sec.xml to get shutdown-users and –passwords, and

in case of server blades, gets the address of the management blade. The content of

these configuration files is supplied during the installation process of a FlexFrame, or it

can be actualized late by the FA Agents (see also FlexFrame Installation Guide V5.1A

Power Shutdown Configuration). It can happen that a certain node isn't in the appropriate

configuration file: You can write it manually into it, or answer the procedure's prompts for

user and password.

Storage Systems Administration NAS Systems Configuration (EMC and NetApp)

138 Administration and Operation

8 Storage Systems Administration

8.1 NAS Systems Configuration (EMC and NetApp)

8.1.1 Adding a New NAS

To add a NAS to FlexFrame environment the NAS has to be configured, the network

prepared and some data has to be stored at LDAP database. The NAS configuration has

to be done manually. But all necessary data and locations are displayed by this program.

The network and LDAP preparation is done directly by ff_nas_adm.pl.

Synopsis

ff_nas_adm.pl --op add --name <node name> --type <sytem type>

--swgroup <switch group id> [--host <ip host part>]

[--ports <port count>] [--blades <blade count>]

[--partner <node name>]

[--cstations <control station count>]

[--shcmd <shell command>]

[--10gbit]

Options

--op add

Adds a NAS and displays some information about processing steps.

--name <node name>

Defines the node name of the NAS. The name may consist of 2 upt to 13 characters

(letters, digits and dashes). It must start with a letter and must not end with a dash.

--type <systemtype>

Defines the type of the NAS. See usage for a list of known NAS types.

--swgroup <switch group id>

Defines the switch group the NAS should be added to. See usage for a list of

configured switch group IDs.

--host <ip host part>

Defines the host part to be used to build IP addresses for the Control or Storage

LAN networks. If this option is omitted the script uses a free host number to calculate

the IP address.

NAS Systems Configuration (EMC and NetApp) Storage Systems Administration

Administration and Operation 139

--ports <port count>

Defines the count of ports to be used with pool Storage LAN networks. If this option

is omitted the script uses the default of two ports. The maximum number of ports per

channel is 8 (dependant on hardware capabilities)..

--blades <blade count>

Blade/data mover count of an EMC Celerra NAS. If this option is omitted the script

uses the default of one blade.

--partner <node name>

Defines the cluster partner in a NetApp filer cluster or the remote partner for an EMC

Celerra.

--cstations <control station count>

Count of control stations of an EMC Celerra NAS. If this option is omitted the script

uses the default of one control station.

--shcmd <shell command>

Shell command (absolute path) used to configure NAS.

If this option is omitted the script uses the default (/usr/bin/ssh).

--10gbit

NAS is a 10Gbit system. This option is only usable for NetApp Filers.

Example for a NetApp Filer

cn1:/opt/FlexFrame/bin # ff_nas_adm.pl --op add --name filer2 --type FAS3000

--swgroup 1

update LDAP

.....

update switch 1/1 configuration

Notice: Update will take about 1 minute.

................................+

Some manual interventions are necessary to integrate the NAS

into FlexFrame environment.

The following list of actions has to be performed in order to integrate the NAS

into your FlexFrame landscape. Since your exact configuration may vary, these steps

have to be performed manually. However, the VIF must be named 'storage'.

The /etc/rc, /etc/exports and /etc/hosts.equiv have to be edited

at volume vol0.

These lines have to be added to or changed at /etc/rc:

Storage Systems Administration NAS Systems Configuration (EMC and NetApp)

140 Administration and Operation

hostname filer2

vif create multi storage <-- add your NICs here, eg. e0a e0b for 1GB/sec ports;

e1 e2 for 10GB/sec ports.

vlan create storage 13

ifconfig storage-13 192.168.20.4 netmask 255.255.255.0 broadcast 192.168.20.255

mtusize 1500 -wins up

options dns.enable off

options nis.enable off

savecore

These lines have to be added to or changed at /etc/exports:

/vol/vol0 -sec=sys,rw=192.168.20.2:192.168.20.1,anon=0

These lines have to be added to /etc/hosts.equiv:

192.168.20.2 root

192.168.20.1 root

As the switch ports are already configured the correct wiring between NAS and

switch ports has to be done. See below a list of cable connections.

Connect your NAS LAN interfaces to named switch ports:

SwitchGroup / Switch / Port LAN Interface

1 / 2 / 3 filer2 (FAS3000): port "data NIC-1"

1 / 1 / 3 filer2 (FAS3000): port "data NIC-2"

Finally execute command "mount /FlexFrame/filer2/vol0"

on both Control Nodes to mount filers vol0 at Control Nodes. This is necessary for

further automated configuration of filer.

The complete instruction above is listed at file /tmp/filer2-add/todo

NAS Systems Configuration (EMC and NetApp) Storage Systems Administration

Administration and Operation 141

Example for an EMC Celerra:

cn1:~ # ff_nas_adm.pl --op add --name emc01 --type NS-TYPE --swgroup 1 --blades 2

update LDAP

...........

update switch 1/1 configuration

Notice: Update will take about 1 minute.

.........................................................................+

Some manual interventions are necessary to integrate the NAS

into FlexFrame environment.

The following list of actions has to be performed in order to integrate

the NAS into your FlexFrame landscape. Since your exact configuration

may vary, these steps have to be performed manually.

As the switch ports are already configured the correct wiring between

NAS and switch ports has to be done. See below a list of cable

connections.

Connect your NAS LAN interfaces to named switch ports:

SwitchGroup / Switch / Port LAN Interface

1 / 2 / 8 emc01 (NS-TYPE): server_2: port "data NIC-1"

1 / 1 / 8 emc01 (NS-TYPE): server_2: port "data NIC-2"

1 / 2 / 9 emc01 (NS-TYPE): server_3: port "data NIC-1"

1 / 1 / 9 emc01 (NS-TYPE): server_3: port "data NIC-2"

1 / 2 / 20 emc01 (NS-TYPE): port "mgmt NIC-1"

Mounting of any Celerra filesystems to Control Nodes is not

necessary for further configuration.

The complete instruction above is listed at file /tmp/emc01-add/todo

8.1.2 Removing a NAS

To remove a NAS from FlexFrame landscape it must not be used by any pool.

ff_nas_adm.pl --op list will display the current configuration of the NAS.

The switch port configuration will be removed from switches and LDAP database.

Synopsis

ff_nas_adm.pl --op rem --name <node name>

Storage Systems Administration NAS Systems Configuration (EMC and NetApp)

142 Administration and Operation

Options

--op rem

Removes a NAS from FlexFrame landscape about the NAS configuration and

displays some information about processing steps.

--name <node name>

Defines the node name of the NAS.

Example

cn1:/opt/FlexFrame/bin # ff_nas_adm.pl --op rem --name filer2

update switch 1/1 configuration

Notice: Update will take about 1 minute.

.............................+

update LDAP

.....

NAS successfully removed from network and LDAP.

8.1.3 Configuring SNMP Traps for NetApp Filers

The NetApp Filers within FlexFrame should send their messages (SNMP traps) to the

Control Nodes (resp. myAMC.Messenger). This has to be configured on the Filers side.

Therefore logon to the Filer(s) with telnet <filer> and continue with the following

steps:

filer> snmp traphost add 192.168.20.1

filer> snmp traphost add 192.168.20.2

Use IP addresses of the Control LAN segment.

filer> snmp community add ro public

public may be replaced by the community you have specified in the

Management Tool.

For further information how to configure thresholds on Filer specific traps, please see the

NetApp Filer documentation.

NAS Systems Configuration (EMC and NetApp) Storage Systems Administration

Administration and Operation 143

8.1.4 Displaying all configured NAS

To get an overview of all configured NAS within the FlexFrame environment use the operation mode list-all of the program ff_nas_adm.pl. It displays IP addresses

and names, type, switch ports and the link aggregation id, separated by NAS.

Synopsis

ff_nas_adm.pl --op list-all

Option

--op list-all

Displays all configured NAS.

Example

cn1:/opt/FlexFrame/bin # ff_nas_adm.pl --op list-all

NAS configurations

filer

Control Lan

192.168.20.3 filer-co

Type: FAS3000

Shell: /usr/bin/ssh -l flexframe

Switch Link Aggregation

Port Count: 2

Link Aggr.ID: 5

Storage LAN switch ports

1 / 1 / 13 SwGroup / Switch / Port

1 / 2 / 13 SwGroup / Switch / Port

Control LAN switch ports

1 / 2 / 15 SwGroup / Switch / Port

Pools

pool1 192.168.2.131 filer-pool1-st master

pool2 10.3.1.3 filer-pool2-st master

pool5 192.168.40.3 filer-pool5 master

usr 192.168.5.3 filer-usr-st master

Storage Systems Administration NAS Systems Configuration (EMC and NetApp)

144 Administration and Operation

8.1.5 Displaying NAS Configuration

To display the detailed configuration of a NAS as known by LDAP database, use the

command ff_nas_adm.pl with operation mode list.

Synopsis

ff_nas_adm.pl --op list --name <node name>

Options

--op list

Displays the configuration of a switch port.

--name <node name>

Defines the node name of a NAS.

Example for a NetApp Filer:

cn1:/opt/FlexFrame/bin # ff_nas_adm.pl --op list --name filer

NAS configurations

filer

Control Lan

192.168.20.3 filer-co

Type: FAS3000

Shell: /usr/bin/ssh -l flexframe

Switch Link Aggregation

Port Count: 2

Link Aggr.ID: 5

Storage LAN switch ports

1 / 1 / 13 SwGroup / Switch / Port

1 / 2 / 13 SwGroup / Switch / Port

Control LAN switch ports

1 / 2 / 15 SwGroup / Switch / Port

Pools

pool1 192.168.2.131 filer-pool1-st master

pool2 10.3.1.3 filer-pool2-st master

pool5 192.168.40.3 filer-pool5 master

usr 192.168.5.3 filer-usr-st master

NAS Systems Configuration (EMC and NetApp) Storage Systems Administration

Administration and Operation 145

Example for an EMC Celerra:

cn1:~ # ff_nas_adm.pl --op list –name cel

NAS configurations

cel

Control Lan

192.168.100.204 cel-co

192.168.100.205 celcs1-co ControlStation1 priv.

192.168.100.206 celcs2-co ControlStation2 priv.

Type: NS-Type

Shell: /usr/bin/ssh

Switch Link Aggregation

DataMover Port Count Link Aggr.ID

2 2 5

Storage LAN switch ports

DataMover SwGroup / Switch / Port

Control LAN switch ports

1 / 1 / 20 SwGroup / Switch / Port

Pools

DataMover Pool IP Name

2 p1 192.168.11.204 cel-p1-st

8.1.6 Adding a Pool to a NAS

To be able to use an existing NAS for a pool, the network connection to the NAS has to

be enhanced. On the NAS, a new virtual interface has to be created and on the switch ports the new VLAN has to be configured. All these steps are done with ff_nas_adm.pl

using the operation mode add-pool, but the program will not change any exports.

For EMC NAS the switch ports to any data mover of an EMC Celerra are allowed for all

VLANs the NAS is using. So any data mover is able to take over the role of any other

data mover.

Synopsis

ff_nas_adm.pl --op add-pool --name <node name> --pool <pool name>

[--role {master|slave}] [--host <ip host part>]

[--blade <blade id>]

Options

--op add-pool

Adds the given pool to named NAS.

Storage Systems Administration NAS Systems Configuration (EMC and NetApp)

146 Administration and Operation

--name <node name>

Defines the node name of NAS.

--pool <pool name>

Defines the pool name the NAS has to be support. See usage for a list of configured

pools.

--role {master|slave}

Defines the role of NAS within given pool. This information is used by the ff_setup_sid_folder program. If this option is omitted the script uses the

default role master.

--host <ip host part>

Defines the host part to be used to build IP addresses for the Control or Storage

LAN networks. If this option is omitted, the script uses a free host number to

calculate the IP address.

--blade <blade id>

Blade/data mover Id of an EMC Celerra NAS. The Id is the numerical value of the

blade/ data mover (eg. server_2 means Id 2). If this option is omitted the script

uses the default of Id 2.

Example

cn1:/opt/FlexFrame/bin # ff_nas_adm.pl --op add-pool --name filer --pool pool4

update LDAP

....

update switch 1/1 configuration

Notice: Update will take about 1 minute.

...........+

vlan: storage-25 has been created

Pool pool4 successfully added to NAS, LDAP and network.

8.1.7 Removing a Pool from a NAS

The rem-pool mode is the opposite to the add-pool operation mode of

ff_nas_adm.pl. It removes the virtual interface to the pool and removes VLAN access

on NAS switch ports for the given pool. This action will not be permitted if any SID of the

pool uses this NAS or the NAS is the default NAS of the pool. Within the last case, the

pool interface of the NAS will be removed when removing the pool.

For EMC NAS the pool's VLAN will only be removed from EMC Celerra switch ports if no

longer any data mover is configured for this pool.

NAS Systems Configuration (EMC and NetApp) Storage Systems Administration

Administration and Operation 147

Synopsis

ff_nas_adm.pl --op rem-pool --name <node name> --pool <pool name>

--blade <blade id>

Options

--op rem-pool

Removes the given pool from the named NAS.

--name <node name>

Defines the node name of the NAS.

--pool <pool name>

Defines the pool name the NAS has to be support. See usage for a list of configured

pools.

--blade <blade id>

Blade/data mover Id of an EMC Celerra NAS. The Id is the numerical value of the

blade/ data mover, eg. server_2 means Id 2. If this option is omitted the script

uses the default of Id 2.

Example

cn1:/opt/FlexFrame/bin # ff_nas_adm.pl --op rem-pool --name filer --pool pool4

update switch 1/1 configuration

Notice: Update will take about 1 minute.

...........+

update LDAP

....

Pool pool4 successfully removed from NAS, LDAP and network.

8.1.8 Adding a Blade (Data Mover) to an EMC Celerra NAS

An EMC Celerra may consist of more than one data mover. To add one or more data

movers to an existing NAS use this command. It will add configuration data to LDAP and

assign and configure switch ports. It will display instructions how to interconnect the new

data movers with the FlexFrame switch group.

Synopsis

ff_nas_adm.pl --op add-blade --name <node name>

[--ports <port count>] [--blades <blade count>]

Storage Systems Administration NAS Systems Configuration (EMC and NetApp)

148 Administration and Operation

Options

--op add-blade

Adds blades/ data movers to NAS.

--name <node name>

Defines the node name of the NAS.

--ports <port count>

Count of ports to be used with pool storage lan networks. If this option is omitted the

program uses the default of 2.

--blades <blade count>

Number of blades/data movers to be added. If this option is omitted the program

uses the default of 1.

Example

cn1:~ # ff_nas_adm.pl --op add-blade --name cel

update LDAP

...

update switch 1/1 configuration

Notice: Update will take about 1 minute.

..................................+

As the switch ports are already configured the correct wiring between

DataMover and switch ports has to be done. See below a list of cable

connections.

Connect your DataMover LAN interfaces to named switch ports:

SwitchGroup / Switch / Port LAN Interface

1 / 2 / 11 cel (NS-Type): server_4: port "data NIC-1"

1 / 1 / 11 cel (NS-Type): server_4: port "data NIC-2"

The complete instruction above is listed at file /tmp/cel-add/todo

8.1.9 Removing a Blade (Data Mover) from an EMC Celerra NAS

To remove one data mover from an existing NAS use this command. It will remove con-

figuration data from LDAP and deactivate the corresponding switch ports.

NAS Systems Configuration (EMC and NetApp) Storage Systems Administration

Administration and Operation 149

Synopsis

ff_nas_adm.pl --op rem-blade --name <node name>

--blade <blade id>

Options

--op rem-blade

Removes blades / data movers from NAS.

--name <node name>

Defines the node name of the NAS.

--blade <blade id>

Blade/data mover Id to be removed. If this option is omitted the program uses the

default of 2.

8.1.10 Create NAS Cluster Partnership

Synopsis

ff_nas_adm.pl --op partner –-name <node name> --partner

<node name>

Options

--op partner

Define a cluster partnership for a NetApp filer cluster or an EMC Celerra.

--name <node name>

First partner in a NetApp filer cluster or an EMC Celerra cluster.

--partner <node name>

Second partner in a NetApp filer cluster or an EMC Celerra cluster.

8.1.11 Move a NAS to another Switch Group

A NAS may be moved from one switch group to another using operation mode move.

This includes moving from and to external (core) switches, distinguished by the switch

group id ‗0‘. During this process, the speed of the network connection may be changed

(even within one switch group by specifying the current switch group id and one of ‗—

1gbit‘ or ‗—10gbit‘).

Storage Systems Administration NAS Systems Configuration (EMC and NetApp)

150 Administration and Operation

Synopsis

ff_nas_adm.pl --op move --name <node name> --swgroup <switch group

id> [--mgmtswgroup <switch group id>]

[--1gbit|--10gbit}

Options

--op move

move a NAS to another switch group

--name <node name>

Defines the node name of the NAS.

--swgroup <switch group id>

Defines the switch group the data ports of the NAS should be moved to.

--mgmtswgroup <switch group id>

Defines the switch group for management ports. This is only relevant when moving

to a NEXUS5x00-VPC switch group. It defaults to the value defined with the switch

group.

--10gbit

move to 10 Gbit Connection .

--1gbit

move to 1Gbit Connection.

8.1.12 Switching a NetApp Filer between 1Gbit and 10Gbit

The named functionality, formerly available with operation mode conf 10gb, is now

supplied by operation mode move, above.

8.1.13 Changing NAS Command Shell

To change the command shell used to configure the NAS, the ff_nas_adm.pl

command supports the change_sh operation mode.

The given shell command replaces the currently used command or default configuring the

named NAS. To change the shell for more than one NAS the procedure has to be done

for each NAS.

NAS Systems Configuration (EMC and NetApp) Storage Systems Administration

Administration and Operation 151

Be sure there is no password request using this shell command to connect the

NAS.

Synopsis

ff_nas_adm.pl --op change-sh --name <node name>

--shcmd <shell command>

Options

--op change-sh

Changes the sh command for NAS configuration.

--name <node name>

Defines the node name of the NAS.

--shcmd <shell command>

Shell command (absolute path) used to configure NAS. If this option is omitted the

script uses the defaults /usr/bin/ssh for NetApp Filers and EMC Celerras.

Example

cn1:~ # ff_nas_adm.pl --op change-sh --name filer

–-shcmd '/usr/bin/ssh -l flexframe'

update LDAP

.

NAS shell command changed at LDAP.

Shell command changed to "/usr/bin/ssh -l flexframe".

8.1.14 Changing NAS LinkAggregate Ports

To change the count of NICs of the vif link aggregate the ff_nas_adm.pl commands

supports the change-ports operation mode. The given number is the count of NICs the

aggregate should contain. The difference between actual and given count of ports will be

allocated at switch group, proper configured and displayed at end of program run. The

additional wiring and configuration of NAS have to be done manually.

At EMC Celerra the link aggregates for all data movers/blades are expanded.

Storage Systems Administration NAS Systems Configuration (EMC and NetApp)

152 Administration and Operation

Synopsis

ff_nas_adm.pl --op change-ports --name <node name>

--ports <port count>

Options

--op change-ports

Change the count of NICs of link aggregate per Filer or data mover.

--name <node name>

Defines the node name of the NAS.

--ports <port count>

Count of NICs the link aggregate should contain.

Example

cn1:~ # ff_nas_adm.pl --op change-ports --name filer –-ports 4

update LDAP.

Expand StorageLAN vif with 2 NICs up to 4 NICs.

Do this before(!) connecting new NICs to switch ports.

The additional switch ports at Switch Group are already configured.

Interconnect additional NICs between NAS and the switches of

SwitchGroup 1 as noted below:

SwGrpID/SwitchID/Port Filer

1 / 1 / 20 <==> Filer data NIC-3

1 / 2 / 20 <==> Filer data NIC-4

Unless any errors are reported follow instructions above to extend

NAS link aggregate(s) to switch group.

Look at "/opt/FlexFrame/network/wiring-channel-extend-filer.txt"

to get a copy of this message.

8.1.15 NAS Disk Free

To get information on available space on NAS file systems, the program ff_nas_df.pl

may be used. It connects to all FlexFrame NAS and views the disk free (df) information.

To be comparable all size values are given as kilo bytes. The view may be restricted to a

single NAS.

NAS Systems Configuration (EMC and NetApp) Storage Systems Administration

Administration and Operation 153

Synopsis

ff_nas_df.pl [--name <node name>] [--mounted]

Options

--name <node name>

Name of NAS to be viewed. Omit this option to view all.

--mounted

Adds "Mounted on" column to view. Typically it is the same as the file system and

therefore omitted.

Example

cn1:~ # ff_nas_df.pl

cel server_2

Filesystem total KB used KB avail KB capacity

mxi_data 20166984 18441704 1725280 91%

mxi_logs 5040984 4098672 942312 81%

root_fs_2 114592 712 113880 1%

root_fs_common 13624 5256 8368 39%

bar_default_data 302511984 265544152 36967832 88%

bar_default_logs 70585984 22607672 47978312 32%

volFF_bar 161339984 120676000 40663984 75%

f940

Filesystem total KB used KB avail KB capacity

/vol/data_o10/ 188743680 174010928 14732752 92%

/vol/foo_default_data/ 308281344 273950812 34330532 89%

/vol/foo_default_logs/ 51380224 24319148 27061076 47%

/vol/linux_luns/ 16777216 872 16776344 0%

/vol/logs_o10/ 18874368 1459008 17415360 8%

/vol/oxi_data/ 20552092 16974148 3577944 83%

/vol/oxi_logs/ 2055212 424528 1630684 21%

/vol/sapdata/ 10276048 26080 10249968 0%

/vol/saplog/ 2055212 384920 1670292 19%

/vol/vol0/ 8388608 380648 8007960 5%

/vol/volFF/ 99614720 76749520 22865200 77%

/vol/volFF_foo/ 209631316 181518524 28112792 87%

Storage Systems Administration NAS Systems Configuration (EMC and NetApp)

154 Administration and Operation

8.1.16 Celerra SRDF-NAS High Availability

A short description of a Celerra SRDF-NAS cluster can be found in the chapter

„2.7.1.2.5―. For more in-depth information please read the EMC Celerra manuals.

Beside the base Celerra-SRDF installation and configuration, which is described in the

FlexFrame installation manual, specific FlexFrame SRDF-NAS configurations are needed

for the functionality of the FlexFrame Celerra SRDF-NAS high availability.

The administration of the Celerra SRDF-NAS high availability in FlexFrame is performed

with the ff_nas_ha.pl command.

8.1.16.1 Syntax

ff_nas_ha.pl --op init

ff_nas_ha.pl --op list

ff_nas_ha.pl --op check

ff_nas_ha.pl --op switchover

ff_nas_ha.pl --help

ff_nas_ha.pl –-version

The init operation mode will be started on one Control Node and initializes a local

parameter file on each Control Node. The initialization must be done as first step after a

reconfiguration of the Celerra SRDF-NAS configuration or, in some cases, of a Control

Node reconfiguration. Otherwise all other operations will fail. If you are unsteady whether the initialization was performed, perform it (again). The init operation mode creates

local files named /opt/FlexFrame/etc/ff_nas_ha.pPar on each Control Node and

on the Celerra control station of R2 side an SRDF-NAS switch lock script. The file

creation is protected with a write lock. Additionally the content of the parameter file is

protected with a md5 checksum. Old existent parameter files will be saved with a time

stamp suffix. For the copy on both Control Nodes temporary files will be used, which are

temporarily located in the /opt/FlexFrame/tmp directory.

Please be aware that the call of ff_nas_ha.pl --op init must be done

when both Celerras are in normal operation mode; this means that all data

movers have their designated roles (no standby is activated at this time). You

can check the state of the data movers by entering the command nas_server -info –all at the control station of each Celerra.

With the list operation mode you can interactively get a current SRDF-NAS state of

both Celerra. The first part of the displayed information is static and comes from the

parameter file. A second part of the information will be dynamically collected from both

Celerra control stations. The third part of the information is the current SRDF pair state.

All information together should be a base of the Celerra configuration for an administrator

and should serve for a manual switchover.

NAS Systems Configuration (EMC and NetApp) Storage Systems Administration

Administration and Operation 155

The check operation mode delivers the SRDF-NAS state in form of a numeric return

code. For this SRDF-NAS check some single inspections will be done. In one inspection

a test write into /FlexFrame/volFF will be executed. In some rare conditions, files

named ff_nas_ha.pl-<cn>-<pid>-testwrite can stay as garbage and will be

erased with the next call of ff_nas_ha.pl. If you find such test write files and if you are

sure that it is garbage, you can erase them also manually. The check operation mode is

explicitly called from the myAMC FrameAgents for automatic state detection. So these

FrameAgents can initiate an automatic switchover that is configured in the FrameAgents.

The switchover operation mode initiates an SRDF-NAS switchover of the central nfs

based file system /FlexFrame/volFF. An SRDF-NAS switchover switches Celerra R1

volumes to the R2 side. The switchover can manually initiated by calling the ff_nas_ha.pl command directly or, in the case of a configured automatic switchover,

through the FrameAgent. If another switchover is already active, a second parallel

switchover will be rejected with the corresponding return value. The lock mechanism for

the switchover is realized in form of a lock daemon, which will be started on the control

station on the R2 side and stopped at the end of the switchover. A signal SIGUSR1 will

stop the lock daemon and release the lock.

The help option delivers the common syntax of the ff_nas_ha.pl command.

The version option delivers the version of the ff_nas_ha.pl command.

8.1.16.2 Background Processes

Monitor alert process:

For a timeout check communication with the FrameAgent the so called monitor alerts are

used. These will be started at the ff_nas_ha.pl command start for each operation and

will stay alive in the background for one minute. So they can be detected by

FrameAgents. In special conditions a monitor alert for timeout extension must be created.

The monitor alert looks in the process list like:

/opt/FlexFrame/tmp/10163/monitor_alert SYMSRV:SRV_NAS

TIMERANGE:180

PHASE:START-NAS-LIST PID:10163

Switchover lock process:

For the switchover a lock mechanism will be started on the control station of the R2 side with /root/.ff_nasrdf_switch.pl. This daemon logs his activity in the control

station system logging file /var/log/messages.

8.1.16.3 Diagnosis

For diagnosis purposes ff_nas_ha.pl writes on each Control Node into the local

logging file /var/log/ff_nas_ha.log. This logging file will be compressed and saved

Storage Systems Administration NAS Systems Configuration (EMC and NetApp)

156 Administration and Operation

with a timestamp suffix in the case of a maximal size. Beside this, all switchovers will also be logged into the Control Node system logging file /var/log/messages.

8.1.16.4 Return Codes

For operation mode init

0 Parameter file is created.

10 Parameter file isn't created.

For operation mode list and check

0 Status of R1 side is ok and automatic switchover isn't necessary.

A manual switchover can be done.

10 Status of R2 side is ok and a manual switchback is allowed.

20 Status isn't ok and an automatic or manual switchover is possible.

30 Status isn't ok and no switchover nor manual switchback is allowed.

40 The parameter file is missing.

41 The parameter file is inconsistent.

For operation mode list only

50 Only static parameter values are printed out.

For operation mode switchover

0 A switchover is successfully executed.

10 A switchover terminated with error.

20 A switchover is started and isn't finished yet.

30 A switchover isn't started.

40 The parameter file is missing.

50 Another switchover or manual switchback is already active.

For all operation modes

90 OS internal error occurred.

91 Unexpected error occurred.

92 No Implementation error occured (Only for none implemented functionality!)

93 Wrong syntax is used.

94 This is a version request.

95 Operation isn't supported.

96 User has no privilege for execution.

NAS Systems Configuration (EMC and NetApp) Storage Systems Administration

Administration and Operation 157

8.1.16.5 Used File Ressources

In directory <control_node>:/opt/FlexFrame/etc

ff_nas_ha.pPar CN local parameter

ff_nas_ha.pPar.lck lock file for CN local parameter

ff_nas_ha.pPar.md5 md5 checksum for local parameter

ff_nas_ha.pPar.<date_time> backups of CN local parameters

In directory <control_node>:/opt/FlexFrame/tmp

ff_nas_ha.pPar_$$ temporary CN local parameter

ff_nas_ha.pPar_$$_rem temporary CN remote parameter

ff_nas_ha.pPar.md5_$$ temporary local md5 file

ff_nas_ha.pPar.md5_$$_rem temporary remote md5 file

$$/monitor_alert_$$ temporary monitor alert script

In directory <control_node>:/var/log

ff_nas_ha.log CN local logging file

ff_nas_ha.log.lck lock file for CN local logging file

messages CN local system logging file

In directory <control_node>:/FlexFrame/volFF

ff_nas_ha.pl-<cn>-$$-testwrite temporary test write file

In directory <r2_celerra_CS0>:/root

.ff_nasrdf_switch.pl daemon lock script on R2-CS0

.ff_nasrdf_switch.pl.lck lock file for switchover lock on R2-CS0

In directory <r2-celerra-CS0>:/var/log

messages R2-CS0 local system logging file

The used directories are located on a Control Node or the control station of the R2 side.

This is expressed by the place holder in angel brackets.

The $$ in the file ressources is used as a shortcut for a process ID.

Storage Systems Administration SAN Configuration in a FlexFrame Environment

158 Administration and Operation

8.2 SAN Configuration in a FlexFrame Environment

8.2.1 Setting Up the SAN Configuration

The SAN configuration for a FlexFrame landscape needs careful planning and design. A

team of specialists for storage systems, servers and SAN infrastructure must work out a

concept that will be a guideline for the complete setup.

Furthermore, it is required to consult the actual recommendations from the manufacturer

of the used storage and SAN components and from the involved partners as well as the

MatrixEP from Fujitsu. These documents can contain important information that must be

taken into account when planning the configuration and when configuring the individual

components.

8.2.1.1 Configuring Storage

This section does not intend to give a detailed description of how to configure SAN

storage for usage in a FlexFrame environment. This task will usually be done by a

storage systems expert in cooperation with a member of the FlexFrame implementation

team.

Furthermore, the SAN storage system is not necessarily an integral part of the FlexFrame

system; you can use LUNs from a storage system which is also used for other purposes

in your computing environment. Therefore, the LUN creation and administration is a task

outside the scope of FlexFrame.

Before starting with this task, a detailed storage concept should be worked out.

8.2.1.2 General Remark for the Use of Navisphere

When using a CLARiiON storage subsystem within a FlexFrame environment (SAN- or

NAS-attached) and the Control Node is used for administrating and/or configuring the

CLARiiON via Navisphere Web GUI it may happen that caused by a switchover of an

Application Node the Servlet Container tomcat is started and requires the port 8080 used

by the Navisphere client. In that case the java_vm and the web browser, e.g. the

Navisphere client are terminated. To avoid this situation use a non-FlexFrame server for

the Navisphere client.

8.2.1.3 Storage System Access

The FlexFrame software itself does not need an administrative access to the SAN

storage system. However, if an administrative access to the storage system from inside

the FlexFrame environment is desired, you can add it either via the Management Tool

(External Connectivity), or later with the ff_swport_adm.pl tool.

SAN Configuration in a FlexFrame Environment Storage Systems Administration

Administration and Operation 159

If using dynamic LUN masking to reconfigure the LUN access settings on the

SAN storage system, the StorMan software (or more precisely, the SMI-S

provider used by StorMan) needs access to the storage system. For details on

StorMan configuration in FlexFrame see chapter "Dynamic LUN Masking".

A special case is when the SAN storage system is used for NAS access too. If it is a

NetApp Filer, the administrative access for SAN administration is the same as the access

used for NAS administrative tasks and will be usually created as a part of the initial setup

via the Management Tool (Controlling, NAS Storage). In case of a Celerra gateway that

uses a CLARiiON/FibreCAT as backend, the Celerra control station needs access to the

Storage Processors (SPs) of the CLARiiON/FibreCAT; as the Celerra control station is in

the FlexFrame control LAN, the SPs must be connected to the control LAN too. For

details see ―Configuration of NAS Systems‖ in the ―Installation of a FlexFrame

Environment‖ manual.

Be aware that to be able to use the SAN functionality of the NetApp Filer, you have to

license the fcp service (license add <license_code> - may require a reboot, depending on

filer model), start the fcp service (fcp start) and bring up the target HBA(s) (fcp config

<adapter> up). If you are using a clustered NetApp Filer, it is recommended to use a cfmode setting of single_image.

8.2.1.4 LUN Creation

For usage by the FlexFrame SAP databases, several LUNs must be created on the

storage system.

The term LUN refers here to a storage device exported from a storage system

that can be seen as a disk device by a host. It may be a LUN from a FibreCAT

CX or NetApp system or a Symmetrix device.

The number and size of LUNs depends on the number and type of the SAP systems to

be installed on SAN storage, as well as on the fact if a host based volume management

software will be used. Based on these LUNs, all file systems needed for the database

data and log files will be created.

A host based volume management software might be used for different reasons, such as

more administrative flexibility, or the ability to create volumes with mirrors on different

storage systems (host-mirroring). The supported volume manager is the native volume

manager from Linux (LVM2) . For FlexFrame specific details on creating volumes and file

systems see section 8.2.4.

It has to be ensured that a LUN is used by only one SAP system.

Never use a volume group for more than one SAP system.

Storage Systems Administration SAN Configuration in a FlexFrame Environment

160 Administration and Operation

When using a host based volume management software, at least two LUNs -preferably more - should be used for each SAP system, one for sapdata

volumes and one for saplog volumes. Only in test environments with low

requirements concerning database performance, you can place saplog and

sapdata volumes on the same LUN.

A similar recommendation also applies to the usage of partition/slices of a LUN

in cases where no host based volume management software is used.

For productive systems, you should also place the LUNs for sapdata and

saplog volumes on physically separated disks, e.g. by using different RAID

groups of a FibreCAT CX, or different aggregates of a NetApp system.

Example

Creation of two LUNs on a NetApp system for usage with Linux, without special

performance considerations. Be aware that if performance is an issue, it may be

advisable to create the LUNs for sapdata in a separate volume that is also located in a

separate aggregate.

> vol create SAN aggr0 120g

Creation of volume 'SAN' with size 120g on containing aggregate

'aggr0' has completed.

> vol options SAN nosnap on

> qtree create /vol/SAN/Sol

> lun create -s 50g -t linux /vol/SAN/Sol/dataSA1

> lun create -s 7g -t linux /vol/SAN/Sol/logsSA1

> lun show

/vol/SAN/Sol/dataSA1 50g (53687091200) (r/w, online)

/vol/SAN/Sol/logsSA1 7g (7516192768) (r/w, online)

>

> lun show -v /vol/SAN/Sol/dataSA1

/vol/SAN/Sol/dataSA1 50g (53687091200) (r/w, online)

Serial#: hpDnSJ8y2Uj0

Share: none

Space Reservation: enabled

Multiprotocol Type: linux

> lun show -v /vol/SAN/Sol/logsSA1

/vol/SAN/Sol/logsSA1 7g (7516192768) (r/w, online)

Serial#: hpDnSJ8y1eiR

Share: none

Space Reservation: enabled

Multiprotocol Type: linux

SAN Configuration in a FlexFrame Environment Storage Systems Administration

Administration and Operation 161

For more information on how to create RAID groups, aggregates, volumes, qtrees and

LUNs on a NetApp storage system, please refer to the documentation from

http://now.netapp.com.

To create a LUN on a FibreCAT CX, you may use the Bind LUN command of the

Navisphere Manager GUI as shown in the following example:

For more information on how to manage storage and create LUNs on EMC storage

systems, please refer to the documentation from http://powerlink.emc.com.

8.2.1.5 Recording LUN Information

For each LUN to be used in a FlexFrame pool for the database of a SAP system, you can

create a VLUN section in the SAN configuration file as described in section 8.5.2. If using

dynamic LUN masking, the creation of a VLUN section is mandatory.

For a NetApp LUN, use the LUN path as arrayLUN and determine the LUNGUID from the

LUN's Serial# by means of the tool ff_san_info.sh on a FlexFrame Node.

For a FibreCAT CX, use the LUN ID as arrayLUN and the LUN's Unique ID as LUNGUID.

The following example shows how to find these information using the Navisphere

Manager GUI:

Storage Systems Administration SAN Configuration in a FlexFrame Environment

162 Administration and Operation

The hostLUN value will be determined and added later, when mapping the LUNs to hosts

(see section 8.2.3.4).

8.2.2 Configuring Application Nodes

All Application Nodes that will run database instances of SAP systems which are to be

configured for SAN usage must be equipped with Host Bus Adapters (HBAs). To ensure

that the correct type of Fibre Channel HBA is used, please consult the FlexFrame

Support Matrix. The supported types include Qlogic and Emulex HBAs for Linux nodes.

For all types of Application Node images, depending on the used combination of

OS/HBA/storage system, specific driver settings may be needed. These must be set

using a Maintenance Cycle as described in section 11.4 ―Maintenance of Application

Nodes - Software‖ on page 318. Be sure to select one of the Application Nodes that is

equipped with Fibre Channel HBAs.

For performance reasons as well as to protect against hardware failure, it is strongly

recommended to use more than one HBA or at least a dual ported HBA on each

Application Node. To manage the multiple paths between host and storage system, a

multipath software is needed.

The FlexFrame Linux Application Node Images already contain a ready-to-use installation

of the Linux native multipath software DM-MPIO.

SAN Configuration in a FlexFrame Environment Storage Systems Administration

Administration and Operation 163

Be aware that the DB-Instance of a SAP system configured for SAN usage will

be able to run only on Application Nodes with the correct image. Therefore

special care must be taken not to start this instance on an Application Node

outside the pool group (of course it is assumed that all nodes run on the same

image, as generally required by a correct FlexFrame installation).

8.2.3 Connecting the Storage to the Application Nodes

Before connecting your Application Nodes to the SAN Storage System, please ensure

that your configuration including server, HBA, Fibre Channel switch and storage system

is supported according to the FlexFrame Support Matrix. Also check that all used

components are correctly configured according to the actual recommendations from the

respective manufacturer or partner and from the MatrixEP.

Fibre Channel switches used to connect the Application Nodes with the Storage Systems

are not considered an integral part of the FlexFrame landscape, and no automatic

configuration of these switches is done. For more information on setup and configuration

of the Fibre Channel switches, refer to the documentation of your switch. For high

availability, a redundant fabric should be set up, so use two separate fabrics if possible.

It is not necessary to connect the LAN ports of the Fibre Channel switches to one of the

FlexFrame LANs. However, if you wish to administrate the switches from the FlexFrame

Control Node, you can create connections to the control LAN either via the Management

Tool (External Connectivity), or later with the ff_swport_adm.pl command.

It is also supposed that the storage system is already connected to the Fibre Channel

switches. If not yet done, follow the guidelines of your storage system's documentation.

To connect an Application Node to the storage system and the LUNs created on the

system for usage by the SAP systems that will run on this Application Node, several

preparations have to be done.

You should go through this procedure for one Application Node first, then create volumes

and file systems on this Application Node as described in section 8.2.4. Finally repeat it

for all other nodes. Optionally, some steps can be done in parallel for all nodes (e.g.

create zones for all Application Nodes at the same time).

During the preparation of an Application Node it must be ensured the FlexFrame

Autonomous Agents are deactivated for this node.

1. Connect the Fibre Channel ports of the Application Node to the Fibre Channel

switches; for redundancy purposes, connect one port to one switch and the other

port to the other switch.

2. Create zones on the Fibre Channel switch, so that the Application Node can see the

storage system(s). For some hints see section 8.2.3.1.

Storage Systems Administration SAN Configuration in a FlexFrame Environment

164 Administration and Operation

3. Check the visibility of the storage system on the Application Node. An example for

this action is shown in section 8.2.3.2.

4. Check the visibility of the Application Node's HBAs (host initiators) on the storage

system. On a FibreCAT CX, register the connection of the host initiators (see section

8.2.3.3).

5. On the storage system, implement the actions required to assign the LUNs that will

be accessed by this Application Node to this node (LUN masking/mapping). Some

details on how this can be done are shown in section 8.2.3.4.

6. Check the visibility of the LUNs on the Application Node (see section 8.2.3.5).

8.2.3.1 Creating Zones on the Fibre Channel Switches

Zones are created on Fibre Channel switches to protect servers and storage systems.

There is no specific requirement how zoning is to be done for a FlexFrame environment;

you may use hard or soft zoning, port based or WWN-based.

We recommend to use WWN-based zoning, and create zones that contain two WWPNs:

one from a host and one from the storage system.

If you want to configure Brocade Fibre Channel switches, you can use the tool

ff_fc_gen_sw.sh on the FlexFrame Control Node to provide you with some assistance.

Consult the man page of this tool for details.

To get the WWPNs of the Application Nodes, you can use the command

ff_san_info.sh –i on each Application Node.

Example on a Linux Application Node:

# ff_san_info.sh -i

HBA WWNN WWPN

host1 20:00:00:c0:9f:c6:45:22 21:00:00:c0:9f:c6:45:22

host2 20:00:00:c0:9f:c6:45:23 21:00:00:c0:9f:c6:45:23

For a NetApp system, the WWPNs can be found as shown in the following example:

> fcp config

0c: ONLINE <ADAPTER UP> Loop Fabric

host address 280000

portname 50:0a:09:81:86:28:13:0c nodename 50:0a:09:80:86:28:13:0c

mediatype ptp speed 2Gb

0d: ONLINE <ADAPTER UP> PTP Fabric

host address 280400

portname 50:0a:09:82:86:28:13:0c nodename 50:0a:09:80:86:28:13:0c

mediatype ptp speed 2Gb

SAN Configuration in a FlexFrame Environment Storage Systems Administration

Administration and Operation 165

You can find the WWNs of a FibreCAT CX in the Navisphere Manager GUI as shown in

the following screenshot:

8.2.3.2 Checking Visibility of the Storage System on the Application Node

If the correct zones have been created on the Fibre Channel switch, the Application Node

should be able to see the storage system. As no LUNs have been mapped yet, the

storage system will present a pseudo LUN 0 to the host. You can verify now that the

correct number of pathes between storage system and Application Node is available.

8.2.3.3 Registering Host Initiators with a CLARiiON/FibreCAT CX

Use Navisphere Manager's Connectivity Status window for the storage system

to register each HBA connection with the storage system:

In the Connectivity Status window, select the connection for the WWN of the

HBA, click Register to open the Register Initiator Record window.

Storage Systems Administration SAN Configuration in a FlexFrame Environment

166 Administration and Operation

Select the following:

● Initiator Type of CLARiiON Open

● ArrayCommPath

● Unit Serial Number of Array

● Failover Mode of 1

In the Host Information box, select New Host and enter the Application Node's

host name and IP address for the first HBA connection of the host. For all other

connections of the host, select Existing Host and choose the appropriate host

name in the drop-down box.

Click OK.

After a few seconds the registration status of the HBA connection should change

from No to Yes.

If you have a host with the Navisphere CLI package installed and LAN access to the SPs,

you can alternatively use the following command to register the HBA connections of an

Application Node as shown in the following example:

# /opt/Navisphere/bin/navicli -h 10.10.13.51 storagegroup -setpath

-hbauid 20:00:00:00:c9:49:42:d2:10:00:00:00:c9:49:42:d2 -sp a

-spport 0 -failovermode 1 -arraycommpath 1 -o -host tompw4

# /opt/Navisphere/bin/navicli -h 10.10.13.51 storagegroup -setpath

-hbauid 20:00:00:00:c9:49:42:d1:10:00:00:00:c9:49:42:d1 -sp a

-spport 1 -failovermode 1 -arraycommpath 1 -o -host tompw4

# /opt/Navisphere/bin/navicli -h 10.10.13.51 storagegroup –setpath

-hbauid 20:00:00:00:c9:49:42:d2:10:00:00:00:c9:49:42:d2 -sp b

-spport 0 -failovermode 1 -arraycommpath 1 -o -host tompw4

# /opt/Navisphere/bin/navicli -h 10.10.13.51 storagegroup -setpath

-hbauid 20:00:00:00:c9:49:42:d1:10:00:00:00:c9:49:42:d1 -sp b

-spport 1 -failovermode 1 -arraycommpath 1 -o -host tompw4

Note that the hbauid above is a concatenation of the WWNN and the WWPN of the

Application Node's HBA.

8.2.3.4 Mapping LUNs to the Application Nodes

If using dynamic LUN masking, this step must be done using the StorMan component.

For details see section "Using StorMan to Reconfigure SAN". The rest of this section

refers to a static SAN setup.

For a FibreCAT CX, this means that you have to connect the Application Node to the

storage group that contains all LUNs that will be used by the Application Node's pool

group. If the storage group does not exist yet, create it and add the LUNs to the storage

group. When adding a LUN to a storage group, a Host LUN Id is automatically assigned.

SAN Configuration in a FlexFrame Environment Storage Systems Administration

Administration and Operation 167

For ease of administration, you may decide to assign this number by yourself and choose

the same number as the LUN ID. The following example shows how you can do this:

You may also decide to use storage groups that contain only one Application Node, but in

this case you have to take special care to add the same LUNs to all storage groups that

contain hosts from the FlexFrame pool group, and that the LUNs also have the same host

LUN number in all groups. Therefore we do not recommend this scheme.

Example on a NetApp host:

> igroup add igroupsol 10000000c949415c 10000000c949415b

> lun map /vol/SAN/Sol/dataSA1 igroupsol

lun map: auto-assigned igroupsol=0

> lun map /vol/SAN/Sol/logsSA1 igroupsol

lun map: auto-assigned igroupsol=1

The number reported behind the igroup name is the hostLUN to be used in the VLUN

section of the SAN configuration file described in section "FlexFrame SAN Configuration

File".

If you decide to assign this number by yourself, you can add a Lun ID parameter to the lun map command.

You can display all existing LUN mappings on a NetApp storage system as shown in the

following example:

> lun show -m

LUN path Mapped to LUN ID Protocol

------------------------------------------------------------------

/vol/SAN/Sol/dataSA1 igroupsol 0 FCP

/vol/SAN/Sol/logsSA1 igroupsol 1 FCP

However, if you decide to assign the host LUN ID by yourself, it is advisable to arrange

for a host LUN ID of 0 for each storage group or igroup, to avoid some known problems.

Storage Systems Administration SAN Configuration in a FlexFrame Environment

168 Administration and Operation

8.2.3.5 Checking Visibility of the LUNs on the Application Node

Example on a Linux Application Node

Make the LUNs visible by doing a reboot. Optionally you can use the ff_qlascan.sh

script for a host using QLA 2xxx HBA, followed by the command multipath –v2. The

visibility can be checked at various levels, starting from driver, SCSI layer up to multipath

layer.

# cat /proc/scsi/qla2xxx/1

QLogic PCI to Fibre Channel Host Adapter for FCI/O-CARD2Gb/s:

Firmware version 3.03.18 IPX, Driver version 8.01.02-sles

...

SCSI LUN Information:

(Id:Lun) * - indicates lun is not registered with the OS.

( 0: 0): Total reqs 26, Pending reqs 0, flags 0x0, 0:0:81 00

( 0: 1): Total reqs 27, Pending reqs 0, flags 0x0, 0:0:81 00

...

# multipath -ll

...

3600601603e851800917bab561d38db11

[size=3 GB][features="1 queue_if_no_path"][hwhandler="1 emc"]

\_ round-robin 0 [prio=2][active]

\_ 1:0:2:9 sdad 65:208 [active][ready]

\_ 2:0:1:9 sdbn 68:16 [active][ready]

\_ round-robin 0 [enabled]

\_ 1:0:1:9 sdl 8:176 [active][ready]

\_ 2:0:0:9 sdav 66:240 [active][ready]

3600601603e8518003a9fa75d1d38db11

[size=3 GB][features="1 queue_if_no_path"][hwhandler="1 emc"]

\_ round-robin 0 [prio=2][active]

\_ 1:0:2:6 sdaa 65:160 [active][ready]

\_ 2:0:1:6 sdbk 67:224 [active][ready]

\_ round-robin 0 [enabled]

\_ 1:0:1:6 sdi 8:128 [active][ready]

\_ 2:0:0:6 sdas 66:192 [active][ready]

...

SAN Configuration in a FlexFrame Environment Storage Systems Administration

Administration and Operation 169

8.2.4 Creating Volumes and File Systems for a SAP System

If you have decided to place the sapdata and saplog data of a SAP System on SAN

storage, you have to create file systems on the LUNs that have been created and made

accessible to an Application Node.

Usually, these file systems are not created directly on the LUNs, but on volumes created

on top of these LUNs with the help of a host based volume management software (the

native volume manager of Linux is supported only), or on partitions/slices of the LUNs.

For more information on how to create volume groups, add physical volumes to the

volume group and create logical volumes inside a volume group, refer to the

documentation of the used volume management software.

The type of the used volume manager, the names of the volume groups as well as the

used volumes have to be documented in the SAN configuration file as described in

section 8.5.2. Here you can also add some options to be used when importing or

deporting a volume group (e.g. forceimport; waitforce=5). Note that each volume name is

recorded as file SystemMountpointSource in the MOUNTPATH section for the file

system that is created on top of this volume. Keep in mind that a volume group has to be

exclusively used for the database of one SAP system only. For some notes that must be

taken into account when creating a volume group with Linux native Volume Manager

(LVM2) for usage by FlexFrame see section 8.2.4.1.

The number and size of the needed file systems depends on the type of the SAP System

to be installed and is determined as a part of the detailed storage concept that has been

defined in advance. As an example for an Oracle based installation you may need 6

sapdata file systems - sapdata1 to sapdata6 and 8 saplog file systems: saparch, oraarch,

sapbackup, sapreorg, origlogA, origlogB, mirrlogA, mirrlogB. Depending on the storage

concept, it is also possible to define a lower or higher number of file systems. For more

information refer to the appropriate SAP Installation Guide. For each file system that will

be used by a SAP system, you will have to add a MOUNTPATH entry in the SAN

configuration file as described in section 8.5.2. Consult the SAP notes applicable for the

database and file system type to get the correct attributes to be used for the creation of

the file systems and the recommended mount options.

If you have entered the description of the volume groups and file systems for a SAP

system ID (SID) in your SAN configuration file, it is time to add these information to the

FlexFrame database (LDAP) and test their usability. The needed steps are detailed in

section 8.2.4.2.

8.2.4.1 Creating a Linux LVM2 Volume Group for FlexFrame Usage

The FlexFrame images for Linux Application Nodes contain a configuration file for LVM

(/etc/lvm/lvm.conf) that is prepared so that LVM volumes can be set up on

DM-MPIO multipath nodes.

Storage Systems Administration SAN Configuration in a FlexFrame Environment

170 Administration and Operation

Furthermore, if you want to use an LVM volume group with FlexFrame, it is necessary to

mark this volume group with an ―FF4S‖ tag as shown in the following example:

bx1:~ # pvcreate /dev/disk/by-name/3600601603e851800681c1bad1d38db11

Physical volume "/dev/disk/by-name/3600601603e851800681c1bad1d38db11"

successfully created

bx1:~ # vgcreate sapdata_s12 /dev/disk/by-name/3600601603e851800681c1bad1d38db11

Volume group "sapdata_s12" successfully created

bx1:~ # vgchange --addtag FF4S sapdata_s12

Volume group "sapdata_s12" successfully changed

8.2.4.2 Completing the Configuration and Testing Usability of SAN for an SID

After you have entered the description of the volume groups and file systems for a SAP system in the SAN configuration file, you should add these information to the Flex-

Frame database (LDAP) and test their usability. You can do this by performing the

following steps:

1. Add the SAP SID to the FlexFrame configuration using the script ff_sid_adm.pl

as described in the section ―Adding/Removing/Modifying SAP SIDs and Instances‖

on page 247. Alternatively, you can also use an SID that has been defined with the

Management Tool.

2. Tell the LDAP database that this SID is using SAN volumes for its database as

shown in the following example:

control1:~ # ff_sid_mnt_adm.pl –-op add –-pool pool2 –-sid S01

--san /tmp/tftpboot/config/sancfgS10

Alternatively you can do this by using the script ff_san_ldap_conf.pl as shown

in section 8.5.1.

3. Before you can install SAP with its database on the SAN volumes, some folders for

the SID have to be created in advance. This also creates links to the mount points for

the SAN volumes. To do so, run the script ff_setup_sid_folder.sh as shown in

the following example:

control1:~ # ff_setup_sid_folder.sh –p pool2 –s S01

After running this script, you should also check that in the database directory for this

SID(/oracle/S01 according to the example shown above) the required links to

/var/FlexFrame/SAN/... have been created and there are no links into the

sapdata/saplog volumes on the NAS system. If some of these links are still here

(this can happen if not all standard areas are needed), delete them, as otherwise a

mixed SAN/NAS installation could result from this setup, if not taking care during

SAP installation. This must be avoided, as mixing SAN and NAS for the database of

one SID is not supported.

SAN Configuration in a FlexFrame Environment Storage Systems Administration

Administration and Operation 171

4. Mount the file systems defined for this SID by using the script ff_san_mount.sh

on the Application Node where you have prepared the file systems, as shown in the

following example:

# ff_san_mount.sh pre sapdb S01 start

/FlexFrame/scripts/ff_san_mount.sh: SAN tool. (SID=S01, PHASE=pre,

ACTION=start)

SID S01 of pool pool1 is not configured for SRDF.

storattach of sid S01 is done!

Volume group "vg_data_S01" successfully imported

Volume group "vg_data_S01" successfully changed

Volume group "vg_logs_S01" successfully imported

Volume group "vg_logs_S01" successfully changed

fsck 1.38 (30-Jun-2005)

e2fsck 1.38 (30-Jun-2005)

/dev/vg_data_S01/sapdata2: clean, 17/1645088 files, 84441/3306496 blocks

/dev/vg_data_S01/sapdata2 mounted at /var/FlexFrame/SAN/oracle/S01/sapdata2

fsck 1.38 (30-Jun-2005)

e2fsck 1.38 (30-Jun-2005)

/dev/vg_data_S01/sapdata6: clean, 10/51200 files, 10578/204800 blocks

/dev/vg_data_S01/sapdata6 mounted at /var/FlexFrame/SAN/oracle/S01/sapdata6

5. Check that the mounted file systems have the correct ownership required by the

database installation and change it as appropriate.

6. Unmount the file systems on this Application Node using the ff_san_mount.sh

script as shown in the following example:

# ff_san_mount.sh post sapdb S01 stop

Repeat the steps 4 and 6 on the other Application Nodes of the Pool Group to test that

each one can access the file systems if needed.

Before you start the SAP installation, mount the file systems again on the Application

Node that has been selected for this task.

After the SAP installation is finished, unmount the file systems again using the ff_san_mount.sh script. For more details on using this script refer to its man page.

Please note in particular that this script is usually implicitly called during the start and stop

of a database instance and should not be used directly, except in situations where you

have assured that no other host is using the volumes and file systems configured for the

database instance. Not taking care of this may result in severe data corruption, up to

making the respective database unusable.

Storage Systems Administration Dynamic LUN Masking

172 Administration and Operation

8.3 Dynamic LUN Masking

8.3.1 Using StorMan to Reconfigure SAN

To do the dynamic LUN Masking for a SAN based database instance of a SAP

application, you can use StorMan, which has been integrated in FlexFrame 5.1A.

StorMan is a virtualization layer for the dynamic management of storage resources and

their dynamic assignment to servers. FlexFrame 5.1A is delivered with StorMan V2.1.

You can use the same StorMan commands, independently of the usage of EMC or

NetApp storage systems. For more information see the ―StorMan V2.0‖ manual on

http://manuals.ts.fujitsu.com/index.php?id=9662-9663-9672-13117.

If you have EMC storage systems, before initial setup of StorMan on your FlexFrame first

of all you have to install the SMI-S provider. In case of a NetApp Filer the installation of

the SMI-S provider is done automatically by the ONTAP software.

SMI-S (Storage Management Initiative Specification) is a standard, which forms the base

technology of StorMan. Defined by the SNIA (Storage Networking Industry Association) is

it a widespread standard in the storage world.

8.3.1.1 Installation of SMI-S Provider

To install the SMI-S provider you have to consider:

Get the SMI-S provider for CLARiiONs and Symmetrix from EMC-Powerlink

(https://powerlink.emc.com).

An account is needed to access the EMC-Powerlink!

Provider-Kit:

Home -> Support -> Downloads & Patches -> Downloads S-Z

Release Notes:

Home -> Support -> Documentation Library -> Software S -> SMI-S Provider

Have a look at the StorMan Support Matrix at

http://www.fujitsu.com/de/products/matrixep/index.html to see which versions of SMI-S

providers are admitted and which Hardware is supported.

It is recommended to install the SMI-S provider for a CLARiiON/FibreCAT CX on both (!)

Control Nodes of your FlexFrame landscape, provided that there exist LAN-connections

from your Control Nodes to the storage system. Otherwise, the SMI-S provider must be

installed on an external node with LAN connection to the storage system, and the Control

Nodes must have LAN connections to this node. If you have a Symmetrix, a fibre

channel connection to the storage system is necessary. Therefore an external node with

Dynamic LUN Masking Storage Systems Administration

Administration and Operation 173

fibre channel connection is needed for the SMI-S provider, and the Control Nodes must

have a LAN Connection to this node. For high availability reasons, two external nodes

should be used, with an SMI-S provider installed on each of them. For more information

about SMI-S provider look into StorMan Manual V2.0, Chap. Installation -> Software ->

'SMI-S Provider' and 'Installation and start of EMC SMI-S provider').

The following picture shows the case of SMI-S providers for a CLARiiON on both (!)

Control Nodes. The second CLARiiON is optional and for high availability reasons.

CN1 CN2

StorMan-erver

CLARiiON-1

SMI-S

LAN LAN

CLARiiON-2

optional 2nd CLARiiON

HA:

In case of a failure of the 1st SMI-S-Provider the FSC Cluster Software PCL switches over to the 2nd SMI-S-Provider

SMI-S StorMan-Server StorMan-Server

The next picture shows SMI-S providers (SMI-S-1/2) installed on external nodes with

FibreChannel connections to the storage system (Symmetrix) and LAN connections to

the Control Nodes. For high availability reasons, two external nodes should be used, with

the SMI-S provider installed on each of them. The second Symmetrix is optional and for

high availability reasons.

Storage Systems Administration Dynamic LUN Masking

174 Administration and Operation

CN1*

SMI-S-1

Server

Symmetrix-1

CN2

SMI-S-2

Server

Symmetrix-2

StorMan-Server

FC

LAN

LAN

FC

FC

FC

LAN

LAN

optional 2nd Symmetrix

HA:

The 2nd SMI-S-Provider will automatically be involved by the StorMan-Server in case of a failure of the 1st SMI-S-Provider.

* The FlexFrame Control Nodes do not

allow Fiber Channel attachments !

StorMan-Server

8.3.1.2 Installation of StorMan

You can decide, whether you want to start StorMan manually or integrate it into your

Linux HA cluster configuration. But, for high availability reasons it is strongly

recommended to do the Linux HA cluster integration if you want to use dynamic LUN

masking!

8.3.1.2.1 Starting StorMan Manually

1. Start the StorMan server by using /etc/init.d/storman start.

2. Normally the SMI-S provider is started implicitly by its installation process. If not, start

the SMI-S provider by using startproc opt/emc/ECIM/ECOM/bin/ECOM -d

3. Connect the SMI-S provider to the StorMan server:

storcfg cimom –add –name <name of host, where SMI-S provider is running> ...

Dynamic LUN Masking Storage Systems Administration

Administration and Operation 175

4. In case of an EMC CLARiiON with no FibreChannel connection to the host where the

SMI-S provider is running: Tell the SMI-S provider, where to discover the storage

configuration. Example:

storemc –add -name <name of host, where SMI-S provider is running> ...

-sp <list of storage processors of CLARiiON/FibreCAT>

-cxuser <administrator user id on storage processors>

-cxpwd <password for administrator userid> ...

5. Tell the SMI-S provider to discover the storage configuration:

storcfg cimom -discover -name <name of host, where SMI-S provider is running>

...

6. Tell the StorMan server the storage system, which you want:

storcfg system -discover -system <systemid> ...

You can get the system id with

storcfg system -show

8.3.1.2.2 Integrating StorMan into the Linux HA cluster Configuration

After you have performed the steps 1 – 6 above:

1. Stop the running StorMan server with /etc/init.d/storman stop and the SMI-S

provider using killproc /opt/emc/ECIM/ECOM/bin/ECOM

(see also StorMan manual).

2. Test it by the commands storcheck and

checkproc /opt/emc/ECIM/ECOM/bin/ECOM.

3. Add StorMan to the cluster configuration

● if CIM is not locally installed or in case of NetApp storage systems:

ff_ha_tool.sh –a storman

● if CIM is locally installed (only for EMC):

ff_ha_tool.sh –a storman

ff_ha_tool.sh –a cim

Adjust path names of SMI-S provider in FlexFrame/volFF/FlexFrame/SMAWstor/cim.conf. Don't forget to

complete the start command by -d (see also StorMan manual).

Storage Systems Administration Dynamic LUN Masking

176 Administration and Operation

4. If you want to use the StorManGUI, you have to adapt the correct host name in /opt/SMAW/SMAWstor/StorMan/StorManGui/stormanstart.jnlp. This

could be necessary, if the DNS name for the SMI-S provider, which has been

generated in this file, is not known in the DNS server, because FlexFrame is working

in a private, local net.

8.3.1.2.3 Preparation Works for Dynamic LUN Masking

First of all you should remove all existing LUN mappings on your storage system,

because this could cause problems, in the case of EMC CLARiiONs and NetApp storage

systems and for storage groups in which several hosts share the same devices.

Then you have to store all storage relevant informations into LDAP by using

ff_san_ldap_conf.pl (see 8.5.1). For all types of volume managers you must add

VLUN entries to your SAN configuration files (see 8.5.2). This is very important for the next step: Calling the procedure ff_storman_conf.pl to execute the configuration of

storage for the desired SAN configuration by using StorMan.

When using ff_san_ldap_conf.pl, you must define the arrayLUNs as

Storman needs it. For example, on a CLARiiON, you must complete them by

leading zeros to a 4 or 5 byte integer (depending on the SMI-S Provider version).

It is very important, that ff_storman_conf.pl is called after

ff_san_ldap_conf.pl, when adding VLUNs, and that it is called before

ff_san_ldap_conf.pl, when deleting VLUNs!

Additionally, on each NetApp storage system used for dynamic LUN masking, you need a

dummy LUN of minimal size which is statically attached with host LUN Id 0 to all

Application Nodes that will use LUNs from this storage system.

If using StorMan for dynamic LUN masking, this attachement must be done using

StorMan commands. You should do this after calling ff_storman_conf.pl, when the

Application Nodes are already defined in StorMan. To be able to do the attach, this LUN

must also be defined in StorMan.

Example:

cn1:~ # storcfg pool -add -poolid ff4s_dummy_0101192460 -system

ONTAP:0101192460

cn1:~ # storcfg volume -add -storid dummy_lun0_0101192460 -poolid

ff4s_dummy_0101192460 -system ONTAP:0101192460 -deviceid

/vol/vol_dummy/dummy:C4lMZ4OEnKar

cn1:~ # storattach -storid dummy_lun0_0101192460 -hostname bx602-11 -

hostlun 0

cn1:~ # storattach -storid dummy_lun0_0101192460 -hostname bx602-12 -

hostlun 0

Dynamic LUN Masking Storage Systems Administration

Administration and Operation 177

8.3.1.2.4 Script ff_storman_conf.pl

ff_storman_conf.pl is used to configure StorMan for FlexFrame SAN configuration.

The necessary information is taken from the LDAP. ff_storman_conf.pl generates or

deletes hosts, pools and devices in StorMan. You have the choice to do this for a certain

pool or for all defined pools. For each specified pool you can specify an SID, thus

generating StorMan configuration for this SID only. If you don't specify an SID, all SIDs of

the specified pool are taken for StorMan configuration.

For defining hosts in StorMan, all affected nodes must have defined HBAs and

WWPNs in their node definition. To define HBAs, ff_an_adm.pl with option

add or hba-change can be used (see section 7.2 and 7.6) and WWPN's are

supplied during boot of the node.

Synopsis

ff_storman_conf.pl –-op {add|del}

–-pool {<pool name> [--sid <SID>]|@ALL}

Options

--op (add|del)

Adds or deletes a StorMan configuration.

--pool {<pool name>|@ALL}

Defines from which pools the SIDs for StorMan configuration should be taken: a

certain SID of a specified pool (a pool name in combination with a given SID), all

SIDs of a specified pool (a pool name only), or all SIDs of all pools (@ALL).

--sid <SID>

Defines the SID, for which you want to configure StorMan.

8.3.1.2.5 On the Control Nodes: Start of Communication Server

To receive the attach/detach orders from the Application Nodes and transmit them to

StorMan, the communication server has to be started. This is normally done automatically

when StorMan has been started by Linux-HA. But if this one has been started manually,

the communication must be started manually by calling

startproc /opt/FlexFrame/bin/ff_comm_server.pl

It can be stopped manually by

killproc /opt/FlexFrame/bin/ff_comm_server.pl

Storage Systems Administration Dynamic LUN Masking

178 Administration and Operation

8.3.1.2.6 On the Application Node: Attach or Detach the Luns

Attach the LUNs of an SID to the current host

If in a SAN configuration an SID is started on a host, all LUNs of this SID have to be visible to the host. You can achive this visibility by calling ff_lunmasking.pl with

opcode attach, but only if the necessary preparation in StorMan is done, i.e. the LUNs are

defined as volumes in StorMan (see section above).

If calling ff_lunmasking.pl either the list of the concerned LUNs is given by caller

(with option lun_file) or ff_lunmasking.pl is collecting them by LDAP. Afterwards

all concerned LUNs are attached to the host by StorMan.

Normally this script is automatically used on Application Nodes during the start of an SID.

Detach the LUNs of an SID from the current host

If in a SAN configuration an SID is stopped on a host, the LUNs should no longer be

visible in the host.

You can finish this visibility by calling ff_lunmasking.pl with opcode detach, but only

if the necessary preparation in StorMan is done, i.e. the LUNs are defined as volumes in

StorMan (see section above).

If calling ff_lunmasking.pl either the list of the concerned LUNs is given by caller

(with option lun_file) or ff_lunmasking.pl is collecting them by LDAP. Afterwards

all concerned LUNs are attached to the host by storman.

Normally this script is automatically used on Application Nodes during the stop of an SID.

Caution must be applied when using this script directly. The configured luns

must not be in use. Not taking care of this may cause data corruption and make

the data on luns unusable.

Synopsis

ff_lunmasking.pl --op {attach|detach} -–sid <sid>

[--lun_file <file name>]

SRDF Support in FlexFrame Storage Systems Administration

Administration and Operation 179

Options

--op attach

Attached the LUNs of the SID to the current host.

--op detach

Detached the LUNs of the SID from the current host.

--sid <sid>

SID, whose LUNs are attached or deatched.

--lun_file <file name>

Name of a file, containing the list of all LUNs which should be attached or detached.

In lun_file every LUN has to be written in a single line. If lun_file is not given

by the caller, the script finds the LUNs by searching in LDAP.

8.4 SRDF Support in FlexFrame

As a prerequisite for disaster recovery procedures for FlexFrame landscapes using EMC

Symmetrix storage systems, a basic SRDF support is included in FlexFrame .

Please note that the SRDF support in FlexFrame is only a building block for a disaster

recovery solution. It is in no case a replacement for a detailed customer specific planning

and operations guide that takes into account the customer specific configuration and

requirements.

Besides this, one should be aware that a disaster is unpredictable by nature; the IT

components of the site hit by the disaster can be partially still functional, or come back to

life after a short outage. Therefore, human intelligence will be needed in most cases to

identify a disaster and act appropriately.

The SRDF support in FlexFrame includes the support of SRDF protected NAS systems

(Celerra) as decribed in chapter 8.1.16 "Celerra SRDF-NAS High Availability" and the

SRDF protection for SAN based SAP database file systems, as decribed in the next

paragraphs.

Symmetrix is EMC's product line of high-end storage solutions targeted to meet the

requirements of customer's mission critical databases and applications. EMC Symmetrix

Remote Data Facility (SRDF) is a Symmetrix-based family of business continuance and

disaster restart solutions.

Storage Systems Administration SRDF Support in FlexFrame

180 Administration and Operation

SRDF is a configuration of Symmetrix units whose purpose is to maintain multiple real-

time copies of the same logical volume data in more than one location. The Symmetrix

units can be in the same room, in different buildings within the same campus, or up to

hundreds of kilometers apart. The local SRDF device, known as the source (R1) device,

is associated with a target LUN (called R2 device). Data of the R1 device are mirrored to

the R2 device.

For more information on Symmetrix storage systems and SRDF refer to the appropriate

EMC documentation.

8.4.1 Storage System Configuration

As with SAN storage systems in general, this activity is outside the scope of FlexFrame

and will usually be done by EMC specialists.

The FlexFrame SAN SRDF support is limited to configurations where each SAN based

SAP database (referred to as a SID in the next paragraphs) with SRDF protection has its

source LUNs (R1 devices) on exactly one Symmetrix system and its target LUNs (R2

devices) on another single Symmetrix system. Other SIDs may use the same or another

pair of Symmetrix systems, or the same pair with opposite roles.

Furthermore, a cross-connected configuration is assumed, where each Application Node

on which a SID's database instance may run is connected via cabling and zoning to the

Symmetrix system holding the SID's source LUNs and to the Symmetrix system holding

the SID's target LUNs.

If possible, dedicated ports on the Symmetrix systems should be assigned for FlexFrame

SAN usage. It is not possible to use the same ports that are used for the connection of

the Symmetrix based Celerra. It is very important to use the bit settings for the ports used

to connect FlexFrame SAN Application Nodes according to the EMC Support Matrix for

the exact Storage Array type, Storage Operating Environment (Microcode) version and

host Operating System. A brief description can also be found in the EMC Host

Connectivity Guide for Linux.

As a part of the storage system preparation it is also very important to create a sufficient

number of gatekeeper devices on the Symmetrix systems for usage by the Application

Nodes on which the SAN SRDF functionality will be used. A number of at least three

gatekeeper devices per Application Node is recommended.

8.4.2 Configuring Application Nodes for SAN SRDF Usage

The FlexFrame scripts supporting SAN SRDF functionality use the EMC Solutions

Enabler to issue commands on the Symmetrix storage systems. This software is not

included in the FlexFrame Application Node images. Therefore it must be installed using

the Maintenance Cycle for Linux Application Node Image as described in section 11.4

―Maintenance of Application Nodes - Software‖ on page 318.

SRDF Support in FlexFrame Storage Systems Administration

Administration and Operation 181

After the preparation of the maintenance image as described in the first part of section

11.4 the installation itself is done on the selected maintenance Application Node. Thereby

some FlexFrame specific considerations must be taken into account:

1. /var/emc must be selected as working root directory instead of the proposed

default of /usr/emc.

2. After installing the rpm, enter the license keys using the symlmf command for at

least the following SYMAPI features: BASE / Symmetrix, SRDF / Symmetrix,

DevMasking / Symmetrix. These licenses will be used by all Application Nodes

running with this image.

3. Enable the usage of GNS (group name services) by uncommenting the line

containing the text #SYMAPI_USE_GNS = ENABLE in the file

/var/emc/API/symapi/config/options. This service facilitates the usage of

the same device group definitions on several hosts accessing the same Symmetrix

storage systems, as the definitions are maintained on the storage systems and not

only in the hosts symapi database as it is without this service.

4. Adjust the settings for the PATH and MANPATH variable for the root user to include

the newly installed software (see example below).

You should also consider to use this maintenance cycle to implement driver specific

adjustments in the Application Node image needed in conjunction with the Symmetrix

storage system, that are not covered by the default settings of the driver (as an example,

append the line options lpfc lpfc_nodev_tmo=10 to the file

/etc/modprobe.conf.local if using Emulex HBAs). Otherwise, these adjustments

must be done in a separate maintenance cycle, which is time-consuming.

The following example assumes that the Solutions Enabler software version 6.5 has been

downloaded as a compressed tar archive and copied to a location that can be accessed

from the Application Node.

Example for installing EMC Solutions Enabler on an Application Node named an1 that

has been prepared for maintenance:

an1:/mnt/emc/solution_enabler/6.5.0.0/LINUX # cp se6500-Linux-x86_64-ni.tar.Z /tmp

an1:/mnt/emc/solution_enabler/6.5.0.0/LINUX # cd /tmp

an1:/tmp # tar -xzvf se6500-Linux-x86_64-ni.tar.Z

symcli-core-V6.5.0-0.x86_64.rpm

symcli-datacore-V6.5.0-0.x86_64.rpm

symcli-datastorbase-V6.5.0-0.x86_64.rpm

symcli-oracle-V6.5.0-0.x86_64.rpm

symcli-srmbase-V6.5.0-0.x86_64.rpm

symcli-star_perl-V6.5.0-0.x86_64.rpm

symcli-storbase-V6.5.0-0.x86_64.rpm

symcli-storfull-V6.5.0-0.x86_64.rpm

Storage Systems Administration SRDF Support in FlexFrame

182 Administration and Operation

symcli-symcli-V6.5.0-0.x86_64.rpm

symcli-symrecover-V6.5.0-0.x86_64.rpm

se6500_install.sh

an1:/tmp # ./se6500_install.sh -install

#----------------------------------------------------------------------------

# EMC Installation Manager

#----------------------------------------------------------------------------

Copyright 2007, EMC Corporation

All rights reserved.

The terms of your use of this software are governed by the

applicable contract.

Solutions Enabler Native Installer[RT] Kit Location : /tmp

Install root directory [/opt/emc] :

Working root directory [/usr/emc] : /var/emc

Entered "/var/emc". To confirm type "Y" : Y

Checking for OS version compatibility......

Checking for previous installation of Solutions Enabler......

[... snip ...]

Do not forget to run 'symcfg discover' after the installation

completes and whenever your configuration changes.

You may need to manually rediscover remotely connected

arrays. Please see the installation notes for further

information.

#-----------------------------------------------------------------------------

# The following HAS BEEN INSTALLED in /opt/emc via the rpm utility.

#-----------------------------------------------------------------------------

ITEM PRODUCT VERSION

01 EMC Solutions Enabler V6.5.0.0

RT KIT

#----------------------------------------------------------------------

-------

an1:/tmp # /usr/symcli/bin/symlmf

SRDF Support in FlexFrame Storage Systems Administration

Administration and Operation 183

E M C S O L U T I O N S E N A B L E R

SOLUTIONS ENABLER LICENSE MANAGEMENT FACILITY

Register License Key (y/[n]) ? y

Enter License Key :

After entering the license keys, you can check their presence:

an1:/tmp # cat /var/symapi/config/symapi_licenses.dat

License Key: XXXX.XXXX-XXXX-XXXX SYMAPI Feature: BASE / Symmetrix

License Key: XXXX.XXXX-XXXX-XXXX SYMAPI Feature: SRDF / Symmetrix

License Key: XXXX.XXXX-XXXX-XXXX SYMAPI Feature: ConfigChange / Symmetrix

License Key: XXXX.XXXX-XXXX-XXXX SYMAPI Feature: DevMasking / Symmetrix

To enable GNS, open the file /var/emc/API/symapi/config/options with an

editor, search for the line starting with #SYMAPI_USE_GNS, remove the starting # and

save the modified file. Verify the setting:

an1:/tmp # grep SYMAPI_USE_GNS /var/emc/API/symapi/config/options

# Parameter: SYMAPI_USE_GNS

SYMAPI_USE_GNS = ENABLE

To adjust the PATH and MANPATH setting for the root user, enter:

an1:/tmp # cat >>/root/.bashrc

export PATH=$PATH:/usr/symcli/bin

export MANPATH=$MANPATH:/usr/storapi/man:/usr/storapi/storman

Do not forget to complete the Maintenance Cycle by applying the step described

in section 11.4.6 Step #3: Reverting the Maintenance Image on page 330.

At the end of this procedure, all Application Nodes on which a database instance of a

SAN based SID with SRDF protection may run should use the newly created image that

contains the EMC Solutions Enabler.

Please note that this procedure must be reapplied after upgrading the relevant

Application Nodes to a new standard FlexFrame Application Node image.

Storage Systems Administration SRDF Support in FlexFrame

184 Administration and Operation

8.4.3 FlexFrame SAN Configuration for SRDF

When using SRDF protection for SAN based SIDs, some additional data must be

recorded in the FlexFrame LDAP database. This is done together with the other SAN configuration data using the script ff_san_ldap_conf.pl. In the configuration file

used with ff_san_ldap_conf.pl when adding SAN data for SIDs, the SID entry for

each SID that uses SRDF must contain the parameter sanRemoteMirrorType with

value SRDF, and further related parameters (primaryDevGroup,

secondaryDevGroup, autoFailoverAllowed). Furthermore, besides the

virtualLUN list(s), that must contain the names of all LUNs used by a SID in normal

operation mode (the R1 devices), the list of the LUNs used after a failover (the R2

devices) must be given with the parameter secondaryVirtualLUN. For details about

using the script ff_san_ldap_conf.pl refer to chapter 8.5.

Example for the SRDF relevant part of a SAN configuration file

# SAN data of SID D01

SID pool1 D01 DB0

{

sanRemoteMirrorType = SRDF

primaryDevGroup = ff_pool1_D01_R1

secondaryDevGroup = ff_pool1_D01_R2

autoFailoverAllowed = yes

VOLUMEGROUP D01data

{

... snip ...

virtualLUN = D01_data_R1

secondaryVirtualLUN = D01_data_R2

... snip ...

}

VOLUMEGROUP D01logs

{

... snip ...

virtualLUN = D01_logs_R1

secondaryVirtualLUN = D01_logs_R2

... snip ...

}

# SAN data of VLUNs

VLUN D01_data_R1 pool1

{

hostLUN = 1

SRDF Support in FlexFrame Storage Systems Administration

Administration and Operation 185

arrayLUN = 00CFB

arrayID = 000290102621

LUNGUID = 60060480000290102621533030434642

}

VLUN D01_data_R2 pool1

{

hostLUN = 1

arrayLUN = 0221B

arrayID = 000190100207

LUNGUID = 60060480000190100207533032323142

}

VLUN D01_logs_R1 pool1

{

hostLUN = 5

arrayLUN = 00D03

arrayID = 000290102621

LUNGUID = 60060480000290102621533030443033

}

VLUN D01_logs_R2 pool1

{

hostLUN = 5

arrayLUN = 02223

arrayID = 000190100207

LUNGUID = 60060480000190100207533032323233

}

# SAN data of storage systems

STORSYS 000190100207 pool1

{

storageArrayType = Symmetrix

wwpn = 50060482d52cbbd8

wwpn = 50060482d52cbbd7

}

STORSYS 000290102621 pool1

{

storageArrayType = Symmetrix

wwpn = 5006048452a75767

wwpn = 5006048452a75768

}

Storage Systems Administration SRDF Support in FlexFrame

186 Administration and Operation

With operation mode list-all of the script ff_san_srdf.pl, a list of all SAN based SAP

systems with SRDF protection as defined in the LDAP database can be displayed, whereas operation mode list-by-symm of the same script shows these SIDs grouped

by the Symmetrix system they are using according to LDAP.

Examples

cn1:~ # ff_san_srdf.pl --op list-all

SIDs with SRDF usage sorted by pool and SID name

Pool pool1

SID D01

Primary Device Group: ff_pool1_D01_R1

Secondary Device Group: ff_pool1_D01_R2

Automatic failover allowed: yes

Secondary LUNs are in use: no

SID D02

Primary Device Group: ff_pool1_D02_R1

Secondary Device Group: ff_pool1_D02_R2

Automatic failover allowed: yes

Secondary LUNs are in use: no

cn1:~ # ff_san_srdf.pl --op list-by-symm

SIDs with SRDF usage grouped by used Symmetrix system

Pool pool1

SIDs using Symmetrix with Id 000290102621: D01 D02

The Symmetrix Device Groups specified in LDAP must exist on each Application Node

where the respective SID's database instance is started. With the usage of GNS that

should be enabled as described in chapter 8.4.2, it is sufficient to create them on one

Application Node. Care must be taken to add the same devices to the device groups as

defined in LDAP. When used with operation mode check-conf on an Application Node

with SAN connection and operational EMC Solutions Enabler, the script ff_san_srdf.pl

checks if this has been done correctly. Besides this, an additional option of this operation

mode gives the possibility to generate a template file with Symmetrix commands that can

be used to create the needed device groups.

Examples

cn1:~ # ff_san_srdf.pl --op check-conf --sid D01 --pool pool1 --outfile /tmp/dg-d01

Template for device groups written to /tmp/dg-d01

For a complete check, call this function on an Application Node

with SAN connection and operational EMC Solution Enabler.

cn1:~ # cat /tmp/dg-d01

SRDF Support in FlexFrame Storage Systems Administration

Administration and Operation 187

### File created on Thu Sep 4 18:07:19 2008 by ff_san_srdf.pl

# Primary device group:

symdg create ff_pool1_D01_R1 -type RDF1

symld -g ff_pool1_D01_R1 add dev 0CFB -sid 000290102621

symld -g ff_pool1_D01_R1 add dev 0D03 -sid 000290102621

# Secondary device group:

symdg create ff_pool1_D01_R2 -type RDF2

symld -g ff_pool1_D01_R2 add dev 221B -sid 000190100207

symld -g ff_pool1_D01_R2 add dev 2223 -sid 000190100207

an1:~ # ff_san_srdf.pl --op check-conf --sid D01 --verbose

Checking LDAP settings

Check storage systems in LDAP ... ok

Check device groups in LDAP ... ok

Checking storage system view for LDAP objects

Check availability of storage systems ... ok

Check availability of device groups ... ok

Check devices in device groups ... ok

an1:~ #

Besides this, it is also possible to get more detailed information about the SRDF

configuration of a SID or its state with other operation modes of the script ff_san_srdf.pl.

For details refer to the script decription in chapter 8.5.3.4.

As the FlexFrame SAN SRDF function uses dynamic LUN masking, it is also

required to make all preparations for this functionality as described in chapter

8.3.1.

8.4.4 SAN SRDF Usage in FlexFrame

With an SRDF configuration it is possible to continue, or, to be more exact, to resume

operation if a storage system is no longer available, by switching to the storage system

that holds the devices with up-to-date copies of the formerly used ones. This storage

system action, also known as an SRDF failover, can be automatically triggered by the

FlexFrame software when certain conditions are met, and the administrator has

requested the usage of the automatism by setting the parameter autoFailoverAllowed to yes for the affected SID in the LDAP database.

Please note that an automatic SRDF failover for the SAN devices of a SID is

only invoked if the FlexFrame software can find out that the secondary devices

are up-to-date copies of the primary devices. Otherwise, no automatic failover

will be done, even if the parameter autoFailoverAllowed has been set to

yes.

Storage Systems Administration SRDF Support in FlexFrame

188 Administration and Operation

During start or restart of the database instance of a SAN based SID with SRDF

protection, the concerned FlexFrame scripts check whether the needed Symmetrix

storage system is reachable, and if it is not reachable, the state of the relevant Symmetrix

device groups is checked to determine if an SRDF failover to the secondary devices is

possible without data loss.

The reaction to the outcome of this check depends on the setting of the parameter

autoFailoverAllowed. If this parameter has the value yes, and the check has a

positive outcome, the SRDF failover is invoked and after its successful completion, the

start or restart continues with the secondary devices.

If a failover is needed but the LDAP setting does not allow it, this is signalled to the

FlexFrame Autonomous Agents by means of the MonitorAlert interface and a message is

produced that describes the situation. In this case, the administrator must decide whether

a failover should be done. The failover can be invoked manually with the operation mode failover of the script ff_san_srdf.pl. The successful completion of the failover is

detected by the waiting start or restart processing, which continues with the secondary

devices.

In a situation where a storage system outage has been detected by other means, or when such an outage is planned, the failover operation of the script ff_san_srdf.pl

can also be invoked. Before calling this operation manually for a SID, it must be ensured

that the database instance of this SID is not running on an Application Node.

After completion of the failover operation on the Symmetrix storage system, the script ff_san_srdf.pl also sets an indicator in the LDAP data of the affected SID which

shows that the secondary devices are in use for this SID, so that subsequent actions for

the same SID can determine the correct devices.

To facilitate the detection of a storage system failure that can be handled by an SRDF

failover, the ServicePingDb detector of the FlexFrame Autonomous Agents should be

activated in configurations with SAN SRDF protection. For details on how to activate this

function, refer to the documentation of the FA Agents. Thereby take care to set also the flag SAN_SRDF_CHECK_ON in the ServicePingDb script. If the ServicePing script or

another detection mechanism of the FA Agents detects a failure of the database instance,

a restart of the service or a reboot of the Node followed by a service start will be trigerred

as defined by the FlexFrame Autonmous Agent's parameters. During this restart or start

processing, a more detailed check of the needed Symmetrix storage system and the

requirement for a failover is done, that can lead to the invocation of an SRDF failover in

some cases as described above.

Please note that while in failover mode, an SRDF protection is no longer available for the

affected devices.

After repairing the failed storage system, it is possible to switch processing back to the

primary devices and to reestablish the SRDF protection by using the operation mode

FlexFrame SAN Configuration Storage Systems Administration

Administration and Operation 189

failback of the script ff_san_srdf.pl. Before invoking this operation, the database

instance of the affected SID must be stopped.

8.5 FlexFrame SAN Configuration

8.5.1 Script: ff_san_ldap_conf.pl

ff_san_ldap_conf.pl is used to administrate the FlexFrame SAN configuration of

Application Nodes.

The arguments given with the command line differ with the operation mode.

To add new information to the FlexFrame database it is necessary to define a lot of data.

These data are stored in a configuration file which is specified as parameter of the add

operation.

Synopsis

ff_san_ldap_conf.pl –op add [--conffile <filename>]

[--outfile <filename>] [--sid {<SID>|*}]

[--pool {<poolname>|*}]

[--vlun {<vlunname>|*}]

[--storsys {<storsysname>|*}]

[--node {<nodename>|*}] [--dryrun] [--verbose]

ff_san_ldap_conf.pl –op del [--outfile <filename>]

[--sid {<SID>|*}] [--pool {<poolname>|*}]

[--vlun {<vlunname>|*}]

[--storsys {<storsysname>|*}]

[--node {<nodename>|*}] [--dryrun]

ff_san_ldap_conf.pl –op list [--sid {<SID>|*}]

[--pool {<poolname>|*}]

[--vlun {<vlunname>|*}]

[--storsys {<storsysname>|*}]

[--node {<nodename>|*}]

ff_san_ldap_conf.pl –-help

Storage Systems Administration FlexFrame SAN Configuration

190 Administration and Operation

Options

--op add

Adds new SAN information to the FlexFrame database.

--op del

Deletes SAN information from the FlexFrame database.

--op list

Displays SAN information from the FlexFrame database.

--conffile <filename>

Name of the SAN configuration file. The default for conffile is

/tftpboot/config/san.cfg.

--outfile <filename>

Name of the work file for writing data to LDAP. The default for outfile is

/tmp/ff_san_ldapdata.

--sid <SID>|*

The data for the given SID or all sids (*) of the specified pool(s) are selected.

--pool <poolname>|*

The data for the given pool or all pools (*) are selected.

--vlun <vlunname>|*

The data for the given vlun or all vluns (*) of the specified pool(s) are selected.

--storsys <storsysname>|*

The data for the given storage system or all storage systems (*) of the specified

pool(s) are selected.

--node <nodename>|*

The data for the given node or all nodes (*) are selected.

--dryrun

Performs the actions of the script but don't write the data to LDAP. The result is

written to outfile.

--verbose

All LDAP messages are displayed.

FlexFrame SAN Configuration Storage Systems Administration

Administration and Operation 191

--help

Displays usage.

8.5.2 FlexFrame SAN Configuration File

# definition of a node

NODE <node_name>

{

availableMultipathSW = { DM-MPIO }

listOfHBA = <value> ...

} ...

# definition of an SID

SID <pool_name> <sid> <instance_name>

{

sanRemoteMirrorType = {SRDF|USER}

primaryDevGroup = <value>

secondaryDevGroup = <value>

autoFailoverAllowed = <yes|no>

VOLUMEGROUP <volume_group_name>

{

usedMultipathSW = { DM-MPIO }

volumeManagerType = { LVM|USER}

volumeManagerUserScript = <value>

volumeManagerOptions = <value>

usage = {SAPDATA|SAPLOG}

virtualLUN = <value>

secondaryVirtualLUN = <value>

MOUNTPATH <file_system_mountpoint_destination>

{

fileSystemType = { ext3|USER}

fileSystemUserScript = <value>

fileSystemMountOptions = <value>

fileSystemCheckOptions = <value>

fileSystemMountpointSource = <value>

} ...

} ...

} ...

# definition of a VLUN

VLUN <vlun_name> <pool_name>

{

Storage Systems Administration FlexFrame SAN Configuration

192 Administration and Operation

hostLUN = <value>

arrayLUN = <value>

arrayID = <value>

LUNGUID = <value>

} ...

# definition of a SAN storage system

STORSYS <storage_array_ID> <pool_name>

{

storageArrayType = {Symmetrix|CLARiiON|NetAppFiler}

wwpn = <value>

wwpn = <value>

...

} ...

The file /opt/FlexFrame/etc/ff_san_ldap_conf.template contains a template of the

configuration file.

The configuration file consists of several lines:

Comment lines are marked with a ―#‖ in the first column of the line.

Empty lines or lines consisting only of white spaces are handled like comment lines.

All other lines describe the entries of the configuration file.

The configuration file has four entries: NODE entry, SID entry, VLUN entry and

STORSYS entry.

Each entry consists of the header line, the opening curly bracket line, the data lines and

the closing curly bracket line.

NODE entry

The NODE entry describes the properties of an Application Node and is identified by its

host name (as defined in the Management Tool and returned by uname -n). It consists of

the following parameters:

availableMultipathSW

This parameter is required and defines the multipath software which is available on

the node. The possible value is:

Value Meaning Available on

DM-MPIO Native multipath software of LINUX LINUX

FlexFrame SAN Configuration Storage Systems Administration

Administration and Operation 193

listOfHBA

This parameter is required and assigned a logical name to each available HBA.

The values of the logical names are not restricted by FlexFrame.

SID entry

The SID entry describes the SAN properties of an SID's database instance and is

identified by the name of the FlexFrame pool, the SID and the database instance. It

consists of a part describing the SAN remote mirroring method used (currently only

SRDF) and one or more VOLUMEGROUP entries.

sanRemoteMirrorType

This parameter must be set with the value SRDF if using EMC Symmetrix storage

with SRDF protection for this SID. It is also possible to specify the value USER, if

another storage system based mirroring method is used.

If using Host Based Mirroring, no entry for sanRemoteMirrorType must be

specified.

The following three parameters are only needed if sanRemoteMirrorType is SRDF:

primaryDevGroup

The name of the Symmetrix Device Group holding the R1 devices.

secondaryDevGroup

The name of the Symmetrix Device Group holding the R2 devices.

autoFailoverAllowed

Specifies whether an automatic failover to the secondary Storage system should be

done if certain conditions are met. Possible values: yes, no. Default: no

VOLUMEGROUP entry

A VOLUMEGROUP entry consists of parameters describing the properties of the volume

group and one or more MOUNTPATH entries. Each entry is identified by the name of the

volume group. The syntax of the volume group name depends on the used volume

manager. If no volume manger is used then the value is not restricted by FlexFrame.

Storage Systems Administration FlexFrame SAN Configuration

194 Administration and Operation

usedMultipathSW

This parameter is required and defines the used multipath software.

The possible value is:

Value Meaning Available on

DM-MPIO Native multipath software of LINUX LINUX

volumeManagerType

This parameter is required and defines the used volume manager.

The possible values are:

Value Meaning Available on

LVM Native volume manager of LINUX LINUX

USER User defined volume manager LINUX

The special value USER means that a non-standard volume manager is used. The

functionality of this volume manager is not guaranteed.

volumeManagerUserScript

If the value of the parameter volumeManagerType is USER then this parameter

specifies the name of a user script for handling the volume manger's start and stop

actions. The file name must be full qualified.

The call by FlexFrame is performed with the following parameters:

● Action code volume_on for actions performed during start of a SID or

volume_off for actions performed during start of an SID

● SID of the SAP system

● Service type

● Service number

● DN of the LDAP entry which caused the user script to be called

FlexFrame SAN Configuration Storage Systems Administration

Administration and Operation 195

volumeManagerOptions

This parameter is optional and contains parameters for the used volume manager.

The possible values depend on the used volume manager. For details see the

description of the used volume manager.

Examples for values are listed in the following table:

Value Meaning

forceimport Force the import (this value is mandatory)

waitforce=5 Wait time is five seconds

The different values are separated by a semicolon.

Usage

This parameter specifies the usage of the volume group. The allowed values are

SAPDATA, SAPLOG, or both.

virtualLUN

This parameter specifies the name(s) of the virtual LUN(s) which are contained in

the volume group and is only required if dynamic LUN masking is used. Each name

specified here refers to an entry in the VLUN section.

secondaryVirtualLUN

This parameter specifies the name(s) of the virtual LUN(s) which are used instead of the ones specified with virtualLUN after a switch to the secondary secondary

system in conjunction with the remote mirroring method specified with

sanRemoteMirrorType. Each name specified here refers to an entry in the VLUN

section.

MOUNTPATH entry

A MOUNTPATH entry consists of parameters describing the properties of the mount path

and is identified by the absolute pathname of the destination of the mount path. The

syntax of this name must be conform with the syntax of the Application Node's operating

system. The pathname must start with /oracle/<SID> or /sapdb/<SID>. This

pathname will be prefixed with /var/FlexFrame/SAN.

Storage Systems Administration FlexFrame SAN Configuration

196 Administration and Operation

fileSystemType

This parameter is required and defines the used file system. The possible values

are:

Value Meaning Available on

ext3 Extended file system version 3 LINUX

USER User defined file system LINUX

The special value USER means that a non-standard file system is used. The

functionality of this file system is not guarantied.

fileSystemUserScript

If the value of the parameter fileSystemType is USER then this parameter specifies

the name of a user script for handling the file system start and stop actions. The file

name must be full qualified.

The call by FlexFrame is performed with the following parameters.

● mount / umount specify that a mount or an umount action is required

● SID of the SAP system

● Service type

● Service number

● Mount point (full qualified filename).

● Volume name

● Volume group name

fileSystemMountOptions

This parameter is optional and defines options used during the mount of the file

system. The possible values depend on the used file system. For details see the

description of the file system. Examples for values are listed in the following table:

Value Meaning

forceumount Force the umount

mountopts=largefiles=yes Options for the mount command

The different values are separated by a semicolon.

FlexFrame SAN Configuration Storage Systems Administration

Administration and Operation 197

fileSystemCheckOptions

This parameter is optional and defines options used during the check of the file

system. The possible values depend on the used file system. For details see the

description of the file system. The special value nofsck indicates that no check of

the file system is to be performed. Examples for values are listed in the following

table.

Value Meaning

nofsck Skip file system check

fsckopts=-ya Options for the fsck command

fileSystemMountpointSource

This parameter is required and specifies the source of the data which are used for

this mount path. The source is a relative file name. For example, if using a volume

manger this is the volume name. The complete file name is constructed by

FlexFrame depending on the used software stack.

VLUN entry

The VLUN entry consists of data describing the properties of the virtual LUN and is

identified by the name of the virtual LUN. This name is a string starting with an

alphabetical character, followed by up to 254 alphabetical characters, numbers, '-', '_' or

'.'. It should not start with the prefix SM_ or _SSYS_ (regardless of case). Virtual LUN

names must be unique, regardless of case.

The specification of VLUN entries is needed if dynamic LUN masking with StorMan is

used. For details on StorMan usage refer to the section ―Using StorMan to Reconfigure

SAN‖. A VLUN entry consists of the following parameters:

hostLUN

This parameter is required and specifies the ID of the storage LUN as it is visible by

the application node.

arrayLUN

This parameter is required and specifies the ID of the storage LUN as it is visible by

the SAN storage subsystem.

arrayed

This parameter is required and specifies the ID of the SAN storage subsystem. It is

the SymmID of a Symmetrix array, the system ID of a NetApp Filer or the serial

number of a CLARiiON/FibreCAT CX.

Storage Systems Administration FlexFrame SAN Configuration

198 Administration and Operation

LUNGUID

This parameter is required and specifies the global unique ID of the LUN.

STORSYS entry

The STORSYS entry consists of data describing the properties of a SAN storage

subsystem and is identified by the ID of the SAN storage subsystem.

The specification of a STORSYS entry is needed for each SAN storage subsystem which

is referenced by the parameter arrayID of a VLUN entry. Use the same name here as

with the arrayID parameter of the corresponding VLUN entries. A STORSYS entry

consists of the following parameters:

storageArrayType

This parameter is required and specifies the type of the storage system. Possible

values are Symmetrix, CLARiiON (to be used also for FibreCAT CX) and

NetAppFiler.

Wwpn

This parameter is required and specifies a WWPN (World Wide Port Name) of a

storage system port that is used for connection to the FlexFrame Application Nodes. A wwpn parameter line must be specified for each used WWPN.

8.5.3 SAN Support Scripts

8.5.3.1 Script: ff_san_mount.sh

Mounting of SAN attached file systems

Synopsis

ff_san_mount.sh {pre|post} <service> <SID>

{start|stop|prestart|post-stop}

Description

This script is used on Application Nodes during start or stop of a database service. During

installation of a database this script must be called if the data files are to be placed onto

SAN based file systems. It is assumed that the information for all mount points is

maintained in the configuration database using ff_san_ldap_conf.pl.

FlexFrame SAN Configuration Storage Systems Administration

Administration and Operation 199

Caution must be applied when using this script directly. The configured file

systems and logical volumes must not be used by other hosts. Not taking care of

this may cause data corruption and make the data on these file systems

unusable.

In particular you must be aware that if the forceimport option has been

specified in the configuration database for the affected volume group, the

volume group will be made accessible on this host by forcing import with the

mechanisms of the used volume manager software.

Once the database is properly installed the ff_service.sh script is calling

ff_san_mount.sh during the start or stop phase implicitly. The prestart and prestop

options are used with SAP ACC. However, their functionality is identical to start and stop

respectively.

To mount a specific set of file systems for the database SID use

ff_san_mount.sh pre sapdb SID start

To unmount the file systems (after all services have stopped) use

ff_san_mount.sh post sapdb SID stop

This script acts upon information stored in the LDAP database. Refer to ff_san_ldap_conf.pl to configure this information.

Debugging

/tmp/ff_san_mount.sh.DEBUGLOG

This file is located at the Application Node and will hold debugging information. In case of

problems provide this file.

8.5.3.2 Script: ff_san_info.sh

SAN information utility

Synopsis

ff_san_info.sh {-g lunid|-t lunspec|-w guid|-n serno|-e|-i}

Description

This utility can be used to get various SAN information and convert them into other

formats.

Storage Systems Administration FlexFrame SAN Configuration

200 Administration and Operation

Options

-g lunid

Shows GUID of LUN(s) with LUN-ID.

-t lunspec

Shows WWNN of target by lunspec (format <host_no>:0:<target_no>:<lun_no>).

-w guid

Shows WWNN of target by GUID.

-n serno

Shows GUID of a NetApp LUN based on its serial number

(see lun show -v <path> on filer).

-e Shows instructions to find GUID of EMC systems.

-i Shows information on FC hosts and targets which are seen from this node.

Usage example on a SLES 9 Application Node:

bx2:~ # ff_san_info.sh -i

HBA WWNN WWPN

1 20:00:00:c0:9f:c6:45:08 21:00:00:c0:9f:c6:45:08

2 20:00:00:c0:9f:c6:45:09 21:00:00:c0:9f:c6:45:09

Target WWNN WWPN

1:0:0 50:06:01:60:b0:60:1f:31 50:06:01:60:30:60:1f:31

1:0:1 50:06:01:60:b0:60:1f:31 50:06:01:68:30:60:1f:31

2:0:0 50:06:01:60:b0:60:1f:31 50:06:01:61:30:60:1f:31

2:0:1 50:06:01:60:b0:60:1f:31 50:06:01:69:30:60:1f:31

bx2:~ # ff_san_info.sh -g 1

LUN(s): 1:0:1:1 2:0:1:1 1:0:0:1 2:0:0:1 GUID: 3600601603e8518008570a1841d38db11

tom1bx2:~ # ff_san_info.sh -t 1:0:1:1

LUN 1:0:1:1: Target WWNN: 5006016830601f31

Corresponding line from 'lsscsi':

[1:0:1:1] disk DGC RAID 5 0219 /dev/sdy

tom1bx2:~ # ff_san_info.sh -w 3600601603e8518008570a1841d38db11

GUID 3600601603e8518008570a1841d38db11: LUN 1:0:1:1: Target WWNN: 5006016830601f31

Corresponding line from 'lsscsi':

[1:0:1:1] disk DGC RAID 5 0219 /dev/sdy

FlexFrame SAN Configuration Storage Systems Administration

Administration and Operation 201

8.5.3.3 Script: ff_qlascan.sh

Scanning for new LUNs on LINUX using a QLA 2xxx HBA

Synopsis

ff_qlascan.sh [-d] -c <a-b> -l <x-y>

Description

This script can be used to query for new LUNs on a Linux Application Node, if the IO

subsystem is connected using QLA host bus adapters (HBA).

Options

-d

Turn on debugging. Currently not supported - use bash -x ff_qlascan.sh ...

instead.

-c <a-b>

Scan controllers from a to b (e.g. "1-2").

-l <x-y>

Scan for LUN-IDs from x to y (e.g. "10-20").

8.5.3.4 Script: ff_san_srdf.pl

FlexFrame(TM) SAN SRDF Support

Synopsis

ff_san_srdf.pl --op list-all [--pool <pool name>]

ff_san_srdf.pl --op list-by-symm [--pool <pool name>]

ff_san_srdf.pl --op list --sid <sid> [--pool <pool name>]

ff_san_srdf.pl --op check-conf --sid <sid>

[--pool <pool name>] [--outfile <output file>]

ff_san_srdf.pl --op check --sid <sid>

ff_san_srdf.pl --op check-and-activate --sid <sid>

[--service <service spec>]

ff_san_srdf.pl --op check-failover-done --sid <sid>

Storage Systems Administration FlexFrame SAN Configuration

202 Administration and Operation

ff_san_srdf.pl --op failover --sid <sid>

ff_san_srdf.pl --op failback --sid <sid>

Description

ff_san_srdf.pl is used to support SRDF usage for SAN based SAP systems in

FlexFrame. It complements the functionality provided for SAP systems (SIDs) with

database on SAN storage for the case of SRDF protected Symmetrix systems and allows

to get an overview about SRDF protected SIDs or detailed informations about a single

SID, check a SID's SRDF-specific FlexFrame configuration, check its state or initiate a

failover or a failback.

As a prerequisite, some SRDF specific SAN configuration data must be recorded in the

LDAP database. This is done together with the other SAN configuration data using the

script ff_san_ldap_conf.pl.

To provide full functionality, the script ff_san_srdf.pl must be called on an

Application Node with SAN connection and operational EMC Solutions Enabler (SYMCLI) for most operation modes (exceptions are list-all and list-by-symm). The EMC

Solutions Enabler must be installed in the Application Node image using the maintenance

cycle as described in chapter "Configuring Application Nodes for SAN SRDF Usage".

To get an overview about all SAN based SAP systems with SRDF protection, use

operation mode list-all. If a pool name is given with operand --pool, the output is

restricted to this pool. If called on an Application Node, the pool of this Node is always

used.

With operation mode list-by-symm, the SAN based SAP systems with SRDF

protection are grouped by needed Symmetrix system. Only the actually needed system is

taken into account; this means the primary system in normal mode and the secondary system after a failover. If a pool name is given with operand --pool, the output is

restricted to this pool.If called on an Application Node, the pool of this Node is always

used.

With operation mode list, more detailed informations are listed for a single SID

specified with operand --sid. If called on Control Node, a pool must also be specified

with operand --pool, and the output will contain only data form the LDAP database. If

called on an Application Node with SAN connection and operational EMC Solutions

Enabler, the storage system view for the relevant objects is also given.

With operation mode check-conf the SRDF-specific FlexFrame configuration for a SID

given with operand --sid can be checked. For a complete check, it must be called on

an Application Node with SAN connection and operational EMC Solutions Enabler.

Otherwise, only LDAP settings are checked. As a part of the check on an Application

Node it is also verified that the device groups specified in LDAP are available and contain the devices according to the LDAP specification. If operand --outfile is given, a list of

FlexFrame SAN Configuration Storage Systems Administration

Administration and Operation 203

commands is written to this file that can be used as a template for creation of the needed

device groups.

With operation mode check, the SRDF-specific state for a SID given with operand --

sid can be checked. It must be called on an Application Node with SAN connection and

operational EMC Solutions Enabler.

Operation mode check-and-activate is implicitly called during start and restart of the

database instance of the SAP system with SID given with operand --sid by

ff_san_mount.sh to check if a failover is needed, initiate the failover if it is needed, the

device state allows a secure failover and the LDAP setting autoFailoverAllowed for

this SID is yes, and activate the correct set of LUNS (primary or secondary). If a failover

is needed but the LDAP setting does not allow it, this is signalled to the FlexFrame

Autonomous Agents by means of the monitor-alert interface, a message that describes

the situation is sent and a specific exit code is produced. Calling this operation directly is

not supported.

Operation mode check-failover-done is used by ff_san_mount.sh in conjunction

with operation mode check-and-activate if a failover is needed for a SID, but the

LDAP setting for this SID does not allow it. It is also intended for internal use only.

With operation mode failover, a failover for the Symmetrix devices of the SID given

with operand --sid is done. It must be called interactively on an Application Node with

SAN connection and operational EMC Solutions Enabler. It is the user's responsibility not

to call this operation for a SID while it is active on an Application Node. The user will

always be asked for confirmation before invoking RDF actions on the Symmetrix system.

With operation mode failback, a failback for the Symmetrix devices of the SID given

with operand --sid is done. It must be called interactively on an Application Node with

SAN connection and operational EMC Solutions Enabler. It is the user's responsibility not

to call this operation for a SID while it is active on an Application Node. The user will

always be asked for confirmation before invoking RDF actions on the Symmetrix system.

Debugging

/tmp/ff_san_srdf.pl.DEBUGLOG

This file is located at the Application Node or Control Node where the script has been

called and contains debugging information. In case of problems provide this file.

Storage Systems Administration FlexFrame SAN Configuration

204 Administration and Operation

8.5.3.5 Script: ff_san_luns.pl

FlexFrame(TM) SAN LUN helper functions

Synopsis

ff_san_luns.pl --op list-all [--pool <pool name>]

ff_san_luns.pl --op list --sid <sid> [--pool <pool name>]

ff_san_luns.pl --op list-att --sid <sid> [--pool <pool name>]

[--symcli]

ff_san_luns.pl --op list-att-node --node <node name>

ff_san_luns.pl --op check-conf [--pool <pool name>]

ff_san_luns.pl --op attach --sid <sid>

[--primary|--secondary]

ff_san_luns.pl --op detach --sid <sid>

[--primary|--secondary]

ff_san_luns.pl --op switch-att-to --sid <sid>

{--primary|--secondary}

ff_san_luns.pl --op detach-all --node <node name>

ff_san_luns.pl --op switch-to --sid <sid>

{--primary|--secondary}

Description

ff_san_luns.pl is used for lun-related actions for SAN based SAP systems in

FlexFrame. It is mainly used internally by the scripts ff_san_mount.sh and

ff_san_srdf.pl on Application Nodes during start, stop and restart of a database

service to attach or detach the correct set of LUNs for a SID (primary or secondary LUNs,

dependig on the currently used ones in conjunction with the usage of storage system

based remote mirroring).

Besides this, the operation modes list-all and list give the possibility to get an

overview about all SAN based SAP systems (SIDs) with LUN usage or to get detailed

informations about the LUNs defined for a specific SID, respectively.

FlexFrame SAN Configuration Storage Systems Administration

Administration and Operation 205

Example

cn1:~ # ff_san_luns.pl --op list-all

SIDs with SAN usage sorted by pool and SID name

Pool pool1

SID D01

LUNs: 2 on 1 storage system(s)

Secondary LUNs: 2 on 1 storage system(s)

SID D02

LUNs: 2 on 1 storage system(s)

Secondary LUNs: 2 on 1 storage system(s)

cn1:~ # ff_san_luns.pl --op list --sid D01 --pool pool1

SAN LUNs of SID D01 from pool pool1

LDAP view:

Volume Manager Type: LVM

Groups: D01data D01logs

Multipath Software Type: DM-MPIO

File System Type: ext3

Number of FS: 12

Dynamic LUN masking with StorMan: yes

LUNs: 2 on 1 storage system(s)

LUN details - grouped by Storage System

VLUN hostLUN arrayLUN LUNGUID

--- Storage System 000290102621 ---

D01_data_R1 1 00CFB 60060480000290102621533030434642

D01_logs_R1 5 00D03 60060480000290102621533030443033

Secondary LUNs: 2 on 1 storage system(s)

Secondary LUNs are not in use

LUN details - grouped by Storage System

VLUN hostLUN arrayLUN LUNGUID

--- Storage System 000190100207 ---

D01_data_R2 1 0221B 60060480000190100207533032323142

D01_logs_R2 5 02223 60060480000190100207533032323233

Furthermore it is possible to verify that the LUN assignments for SIDs in LDAP are

consistent or to switch the LUN usage setting in the LDAP database.

The arguments given with the command line differ with the operation mode. The

functionality also depends on the fact if the call is done on a FlexFrame Control Node or

an Application Node.

Storage Systems Administration FlexFrame SAN Configuration

206 Administration and Operation

To get an overview about all SAN based SAP systems and the number of LUNs assigned to each SID, use operation mode list-all. If a pool name is given with operand

--pool, the output is restricted to this pool. If called on an Application Node, the pool of

this node is always used.

With operation mode list, more detailed informations are listed for a single SID

specified with operand --sid. If called on Control Node, a pool must also be specified

with operand --pool, and the output will contain only data form the LDAP database. If

called on an Application Node with SAN connection, the view of this Application Node is

also given. With the additional option --skip-ldap, the output of the LDAP information

can be suppressed.

Operation mode list-att is usually to be called on Control Node and displays for each

LUN of the SID specified with operand --sid from pool given with operand --pool to

which node it is attached. On Application Node it can be called only with option

--symcli and it then expects an operational EMC Solutions Enabler on this Node. In

this case, only a summary of LUN connections for this Application Node and the SID given with operand --sid is shown.

Operation mode list-att-node must be called on Control Node and displays for an

Application Node specified with operand --node the SIDs with LUNs that are attached to

this Application Node.

With operation mode check-conf, the LUN configuration in the LDAP database can be

checked. If a pool name is given with operand --pool, the check is restricted to this

pool. If called on an Application Node, the pool of this Node is always used. If called on

Control Node, it is also checked that all LUNs of SIDs configured for dynamic LUN

masking with StorMan are also known by StorMan.

Operation modes attach and detach are usually implicitly called during start and stop

of a SAP database instance configured for SAN usage with dynamic LUN masking on the

Application Node where the database instance is started or stopped.

Caution must be applied when calling this operations directly. This is only

allowed during setup of the SAN configuration for a SID, and the caller is

responsible that the LUNs are not in use by other hosts when attaching, or that

the usage of the LUNs has been ended on the host where the detach is done.

Operation mode attach must be called on Application Node and attaches the LUNs of

the SID given with operand --sid to the Application Node on which it is called, and also

makes sure that the operating system's device structures are updated with the newly

attached LUNs. For a SID, for which a storage based remote mirroring method is

configured, it can be specified which set of LUNs (primary or secondary) must be

attached, by using one of the options --primary or --secondary. If not specified, the

correct set of LUNs is selected according to the actual LDAP setting.

FlexFrame SAN Configuration Storage Systems Administration

Administration and Operation 207

Operation mode detach must be called on Application Node and detaches the LUNs of

the SID given with operand --sid from the Application Node on which it is called, and

also makes sure that the operating system's device structures are updated accordingly.

For a SID, for which a storage based remote mirroring method is configured, it can be

specified which set of LUNs (primary or secondary) must be detached, by using one of the options --primary or --secondary. If not specified, the correct set of LUNs is

selected according to the actual LDAP setting.

Operation mode switch-att-to is a combination of the two operations described

above and is used internally for SIDs with SRDF usage. Direct usage is not supported.

Operation mode detach-all must be called on Control Node and allows to detach the

LUNs of all SIDs configured for dynamic LUN masking with StorMan from the Application

Node specified with operand --node. This can be done only when the concerned

Application Node is not operational.

Operation mode switch-to allows to set the information in LDAP which set of LUNs

(primary or secondary) is to be used for the SID given with operand --sid. If called on

Control Node, the pool of this SID must also be specified with operand --pool. To select

the LUN set, one of the options --primary or --secondary must be given.

This operation is only relevant for a SID for which a storage based remote

mirroring method is configured. It must be used only when an action on the

storage system has been done directly that changed the LUN set to be used (as

an example, if an SRDF failback operation has been done without using the

ff_san_srdf.pl script, the LUN set setting must be switched to primary).

Debugging

/tmp/ff_san_luns.pl.DEBUGLOG

This file is located at the Application Node or Control Node where the script has been

called and contains debugging information. In case of problems provide this file.

Switch Administration Adding a Switch to a Switch Group

208 Administration and Operation

9 Switch Administration

A switch group consists of at least two switches of the Cisco Catalyst 3750 switch family

building a switch stack or of exactly two Nexus Switches building a vPC domain. The

switches building a switch stack are interconnected on the back with Cisco StackWise

technology and work like one switch.

For a description how to interconnect switches of the Cisco Catalyst 3750 switch

family, please refer to Cisco Catalyst 3750 Installation Manual or ―Cisco

StackWise Technology‖ White Paper.

Adding or removing a switch of the Cisco Catalyst 3750 switch family to or from

a switch group means to add or remove a switch to/from a switch stack.

To ensure safe operation we recommend doing this at a downtime to minimize

influence on running systems. This requires shutting down all systems con-

nected to this switch group. In case a NAS system is connected, all systems that

have mounted file systems from the NAS system have to be shut down as well.

Ensure all switches in the stack have a proper IOS Version. If necessary perform

an IOS upgrade resp. downgrade.

In a ―mixed‖ switch group that is with members of different types (3750g, 3750e,

3750x) an IOS upgrade can only be done step by step one switch after the other,

as different types may need different IOS versions. Please refer to original Cisco

documents for Catalyst 3750 for that task.

9.1 Adding a Switch to a Switch Group

A switch may be added to a switch group building a switch stack. See notes above for

recommendations.

The following processing has to be performed.

1. Mount the new switch next to the existing switches

2. Run ff_save_switch_config.pl to save configurations

3. Write down switch ids of each stack member of the switch stack as they may change

inserting the new switch to the stack.

4. Check the IOS versions of the switches in the existing switch stack and the new

switch. Switches of the same model (G or E model) must have the same version. If

the versions are different, upgrade resp. downgrade the IOS of the new switch.

Adding a Switch to a Switch Group Switch Administration

Administration and Operation 209

5. Power off all switches of the switch stack, connect the new switch to the stack using

the provided stacking cable and stacking ports at rear side (see Cisco installation

manual for details), power on all switches of the stack except the new one

6. Compare the actual stack member ids with your noticed ids In case of differences

use the following IOS command for renumbering: switch <Switch Number> renumber <New switch number>

7. Power on the new switch, once again compare the stack member ids, set all

interfaces of the new switch to shutdown

8. Use ff_swgroup_adm.pl --op add-sw to add the switch to the FlexFrame

configuration as described below.

The Switch group is now ready for use for further configuration.

Synopsis

ff_swgroup_adm.pl --op add-sw --group <switch group id>

--type <switch type> [--dryrun]

Options

--op add-sw

Adds a new member to the switch group and displays some information about

processing steps.

--group <switch group id>

Defines the switch group to be used.

--type <switch type>

Defines the type of the new switch to be added to the switch group. Call ff_swgroup_adm.pl without any parameter to get a list of supported switch types

The maximum number of switches per switch group is 9. For more than 4 switches

with 10GbE ports the StackWise cabling may be a bottleneck.

--dryrun

For test purpose you can perform the function without changing the LDAP Database.

Switch Administration Adding a Switch to a Switch Group

210 Administration and Operation

Example

cn1:/opt/FlexFrame/bin # ff_swgroup_adm.pl --op add-sw --group 1

--type cat3750g-24ts

If program is aborted by Ctrl-C or a failure remove left overs

by calling:

ff_swgroup_adm.pl --op rem-sw --group 1 --switch 3

Switch was added to LDAP data.

Keep in mind: INSERTING SWITCH TO STACK NEEDS A DOWN TIME !

To add the switch to switch stack (switch group 1) write down the current switch

ids as they may change inserting the new switch to stack.

To connect the switch to the stack use the provided stacking cable and stacking

ports at rear side. See Cisco installation manual for details.

If switch ids get scrambled use the IOS command

"switch <current_no> renumber <new_no>"

to put them in same order as before.

In short the to do list:

-> write down switch ids of each switch of group

-> power down entire switch group

-> insert switch into stack

-> power on entire switch group

-> look at switch ids and compare with your noticed

-> in case of differences use IOS command to renumber switches

Switch group is ready for use

See file /tmp/swgrp-add-1-3/next_steps for same instructions as above.

Removing a Switch from a Switch Group Switch Administration

Administration and Operation 211

9.2 Removing a Switch from a Switch Group

A switch may be removed from a switch group building a switch stack. Remove it first from the LDAP database with the command ff_swgroup_adm.pl --op rem-sw as

described below and then from the stack. See notes above for recommendations.

All ports of the switch must be unused.

Only the switch with the highest member ID within the stack can be removed.

Synopsis

ff_swgroup_adm.pl --op rem-sw --group <switch group id>

--switch <switch id> [--dryrun]

Options

--op rem-sw

Removes the last member from a switch group and displays some information about

processing steps.

--group <switch group id>

Defines the switch group to be used.

--switch <switch id>

Defines the stack ID of the switch to be removed from a switch group.

--dryrun

For test purpose you can perform the function without changing the LDAP Database.

Example

cn1:/opt/FlexFrame/bin # ff_swgroup_adm.pl --op rem-sw --group 1 --switch 3

Switch was successfully removed from LDAP data.

Keep in mind: REMOVING SWITCH FROM STACK NEEDS A DOWN TIME !

In short the to do list:

-> power down entire switch group

-> remove switch from stack

-> power on entire switch group

Switch group is ready for use

Switch Administration Listing a Switch Group Configuration

212 Administration and Operation

See file /tmp/swgrp-rem-1-3/next_steps for same instructions as above.

9.3 Listing a Switch Group Configuration

Invoking the command ff_swgroup_adm.pl with the list operation mode displays

the configuration of a switch group like used switch types, port channels, port usage

statistics and used switch ports.

Synopsis

ff_swgroup_adm.pl --op list --group <switch group id>

Options

--op list

Displays switch group configuration.

--group <switch group id>

Defines the switch group to be used.

Example

cn1:/opt/FlexFrame/bin # ff_swgroup_adm.pl --op list --group 1

Switch Group 1

Name/IP: switch-i / 192.168.13.15

Login: root

Password: passwort

SNMP Community: public;ro

Switch Types: (switch id, switch type)

1 cat3750g-24t

2 cat3750g-24t

3 cat3750g-24t

Port Channels: (channel id, switch ports, connected device)

2 1/5,2/5 swb-1-1/11,swb-1-1/12

3 1/6,3/1 swb-1-2/11,swb-1-2/12

4 1/15,3/7 intfiler

5 1/19,2/11 swb-2-1/39,swb-2-1/40

6 2/22,3/22 swb-2-2/39,swb-2-2/40

Switch port usage: (switch id, used, free tx, free fx, free 10GB ports)

1 11 used 13 free tx 0 free fx 0 free 10Gb

Listing a Switch Group Configuration Switch Administration

Administration and Operation 213

2 10 used 14 free tx 0 free fx 0 free 10Gb

3 8 used 16 free tx 0 free fx 0 free 10Gb

Switch port list: (switch id, port id, connected device, vlans)

1 1 BX cabinet 2 u13

1 2 --- unused ---

1 3 rx300-14 t10,t12,u11

1 4 rx300-15 t10,t12,u11

1 5 swb-1-1/11 t13,t10,t12,t11,t1

1 6 swb-1-2/12 t13,t10,t12,t11,t1

1 7 --- unused ---

1 8 --- unused ---

1 9 --- unused ---

1 10 --- unused ---

1 11 cn1 u13,t11,t12,t10

1 12 cn2 u13,t11,t12,t10

1 13 --- unused ---

1 14 --- unused ---

1 15 intfiler t11,t13

1 16 extern. Connect t13,t10,t11,t12

1 17 --- unused ---

1 18 --- unused ---

1 19 swb-2-1/39 t13,t10,t12,t11,t1

1 20 --- unused ---

1 21 --- unused ---

1 22 --- unused ---

1 23 Corporate LAN u10

2 1 BX cabinet 2 u13

2 2 --- unused ---

2 3 rx300-14 t10,t12,u11

2 4 rx300-15 t10,t12,u11

2 5 swb-1-1/12 t13,t10,t12,t11,t1

2 6 --- unused ---

2 7 --- unused ---

2 8 --- unused ---

2 9 --- unused ---

2 10 --- unused ---

2 11 swb-2-1/40 t13,t10,t12,t11,t1

2 12 cn2 mgmt u13

2 13 --- unused ---

2 14 --- unused ---

2 15 cn1 mgmt u13

2 16 --- unused ---

2 17 --- unused ---

Switch Administration Generating a Switch Group Configuration

214 Administration and Operation

2 18 --- unused ---

2 19 --- unused ---

2 20 --- unused ---

2 21 BX cabinet 1 u13

2 22 swb-2-2/39 t13,t10,t12,t11,t1

2 23 Corporate LAN u10

3 1 swb-1-2/11 t13,t10,t12,t11,t1

3 2 --- unused ---

3 3 --- unused ---

3 4 --- unused ---

3 5 cn1 u13,t11,t12,t10

3 6 cn2 u13,t11,t12,t10

3 7 intfiler t11,t13

3 8 --- unused ---

3 9 --- unused ---

3 10 --- unused ---

3 11 --- unused ---

3 12 --- unused ---

3 13 --- unused ---

3 14 --- unused ---

3 15 --- unused ---

3 16 --- unused ---

3 17 --- unused ---

3 18 --- unused ---

3 19 BX cabinet 1 u13

3 20 rx300-15 mgmt u13

3 21 rx300-14 mgmt u13

3 22 swb-2-2/40 t13,t10,t12,t11,t1

9.4 Generating a Switch Group Configuration

Invoking the command ff_swgroup_adm.pl with the gen-config operation mode

generates the configuration of switch group switches to be uploaded manually according

to the corresponding HW-Characteristics Quickguide.

If the connection to the NAS system may be lost because of the manually upload

change /tftpboot to be a local directory before generating the configuration. If the

generated configuration should be accessed by use of the tftp protocol also

restart tftpd.

mv /tftpboot /tftpboot.sav

mkdir /tftpboot

ff_ha_cmd.sh restart tftpd

After the upload go back using

Changing the Password of a Switch Group Switch Administration

Administration and Operation 215

mv /tftpboot/* /tftpboot.sav

rmdir /tftpboot

mv /tftpboot.sav /tftpboot

ff_ha_cmd.sh restart tftpd

Synopsis

ff_swgroup_adm.pl --op gen-config --group <switch group id>

Options

--op gen-config

Generates configuration of switch group switches.

--group <switch group id>

Defines the switch group to be used.

Example

Cn1:~ # ff_swgroup_adm.pl --op gen-config --group 1

Switch configuration generated. Upload has to be done manually.

Follow the instructions of the applicable Quick Start Hardware

FlexFrame for SAP document in the doc/hwinfo section of the Service CD

and use the following data:

Switch Group swg1

IP and Mask: 172.20.13.6 255.255.255.0

Config: scp://[email protected]//tftpboot/swg1.config

You can find a copy of the message above in

/var/log/FlexFrame/tmp/swgroup_adm-gen-config-1/msg_of_swgroup1_swg1.txt

9.5 Changing the Password of a Switch Group

To change the access password of a switch group it has to be changed at switch group

as well as LDAP database. The command ff_swgroup_adm.pl changes both.

Synopsis

ff_swgroup_adm.pl --op pass --group <switch group id>

--passwd <password> [--dryrun]

Switch Administration Changing the Host Name of a Switch Group

216 Administration and Operation

Options

--op pass

Changes the switch group access.

--group <switch group id>

Defines the switch group to be used.

--passwd <password>

Defines the new password as clear text.

--dryrun

For test purpose you can perform the function without changing the LDAP Database.

Example

cn1:/opt/FlexFrame/bin # ff_swgroup_adm.pl --op pass --group 1 --passwd berta

update switch 1/1 configuration

Notice: Update will take about 1 minute.

............+

Password changed from "anton" to "berta".

See file /tmp/swgrp-pass-1/info for same information as above.

9.6 Changing the Host Name of a Switch Group

To change the host name of a switch group it has to be changed at switch group as well

as in the LDAP database and host files on both control nodes. The command ff_swgroup_adm.pl changes all of them.

Synopsis

ff_swgroup_adm.pl --op name --group <switch group id>

--name <name> [--dryrun]

Options

--op name

Changing the Host Name of a Switch Group Switch Administration

Administration and Operation 217

Changes the switch group host name. The new name may consist of 2 up to 13

characters (letters, digits and dashes). It must start with a letter and must not end

with a dash.

--group <switch group id>

Defines the switch group to be used --name <name>

Defines the new host name to be used.

Switch Administration Displaying/Changing Common Network Configuration Parameters

218 Administration and Operation

--dryrun

For test purpose you can perform the function without changing the LDAP Database.

Example

cn1:/opt/FlexFrame/bin # ff_swgroup_adm.pl --op name --group 1 --name swg1

update switch 1/1 configuration

Notice: Update will take about 1 minute.

...+

Switch name changed from "swg-1" to "swg1".

See file /tmp/swgrp-name-1/info for same information as above.

9.7 Displaying/Changing Common Network Configuration Parameters

Some parameters of network configuration influence the switch group configuration and

were defined with the Management Tool at initial installation.

To display or change these parameters use the ff_swgroup_adm.pl command.

Synopsis

ff_swgroup_adm.pl --op parameter [–-parameter <parameter name>

--value <parameter value>] [--dryrun]

Options

--op parameter

Displays (without the following options) or changes (in conjunction with the following

options) network configuration parameters.

--parameter <parameter name>

Defines the name of the parameter to be changed. Known parameters are:

usePortsToClan

The parameter specifies if dedicated switch ports for Client LAN connection should

be used:

● no

No dedicated switch ports for Client LAN connection are used. Uplinks are

connected to customer corporate LAN and the Client LANs are part of the

uplink.

Displaying/Changing Common Network Configuration Parameters Switch Administration

Administration and Operation 219

● yes

Dedicated switch ports for Client LAN connection should be used.

clanPortPerVlan

The parameter depends the way the Client LANs are connected to the corporate

LAN (with the sapgui users). There are two modes:

● no

Use two dedicated switch ports (for redundancy) and use them for all pool

Client LANs (as tagged VLANs onto these two ports).

● yes

Use two dedicated switch ports (for redundancy) for each pool Client LAN

(as untagged VLAN on both ports).

useTxToClan

If the switch group has fiber optic ports (SFP ports) these may be preferred as ports

to the corporate LAN transferring the pool Client-LAN data. To be able to customize

port type the parameter has two values:

● no

The port type will be fiber optic. Be sure to have matching modules for the

SFP ports modules for the SFP ports.

● yes

The port type will be Cat5 twisted pair, commonly abbreviated as TX

useTxUplink

If the switch group has fiber optic ports (SFP ports) these may be preferred as ports

to connect another switch group directly or to use them as uplink ports to the

FlexFrame integrated LAN switch. The parameter has two values:

● no

The port type for uplink will be fiber optic. Be sure to have matching

modules for the SFP ports.

● yes

The port type for uplink will be Cat5 twisted pair, commonly abbreviated as TX

uplinkPortCnt

Defines the count of ports aggregated to the uplink. The uplink channel consists of at

least two ports and may have a maximum of eight ports. The parameter value is the

count of ports.

Switch Administration Adding a Switch Group

220 Administration and Operation

--value <parameter value>

Defines the value of the parameter to be changed. For uplinkPortCnt use a

range between 2 and 8. For all other parameters use yes or 1 and no or 0 as value.

--dryrun

For test purpose you can perform the function without changing the LDAP Database.

Examples

cn1:~ # ff_swgroup_adm.pl –op parameter

Common Network Configuration Parameters

Parameter Name Parameter Value

use ports to Client LAN no

Client LAN port per VLAN yes

use TX ports to Client LAN yes

use TX ports as uplink no

uplink port count 2

timezone Europe/Berlin

POSIX timezone CET

cn1:~ # ff_swgroup_adm.pl --op parameter --parameter useTxToClan --value

yes

Parameter successfully changed at LDAP.

9.8 Adding a Switch Group

To add an entire new switch group, use the ff_swgroup_adm.pl command. The

command adds data to the LDAP database and creates initial configuration which has to

be uploaded manually according to the corresponding HW-Characteristics Quickguide.

Values appropriate to the current configuration needed for the manual upload are

displayed.

One uplink channel is configured which should be connected to an uplink

channel of another switch group or to switches outside the FlexFrame

environment. If more than one uplink channel is needed you can use

ff_swgroup_adm.pl --op add-uplink to add more channels. Afterwards

use ff_swgroup_adm.pl --op gen-config to create a new initial

configuration for upload purposes.

For details about uplink channels and supported topologies see the FlexFrame™

for SAP® – Network Design and Configuration Guide.

Adding a Switch Group Switch Administration

Administration and Operation 221

Synopsis

ff_swgroup_adm.pl --op add --group <switch group id>

--type <list of switch types>

--name <switch group name>

--passwd <password> [--login <login name>]

[--host <ip host part>[,ip host part]]

[--snmp <community name>]

[--uplinkportcnt <port count per channel>]

[--uplinkportmedia {tx|fx|10gb}]

[--mgmtswgroup <switch group id>]

[--vpcdomain <vpc domain id>]

[--dryrun]

Options

--op add

Adds a switch group and displays some information about processing steps.

--group <switch group id>

Defines switch group to operate. Within this operation mode used to define the id of

the new switch group.

--type <list of switch types>

Defines the switch types of the switches the switch group will consist of. The types

are comma separated without any white spaces. The types must belong to the same

switch family as described above. Call ff_swgroup_adm.pl --help to get a list

of supported switch types.

In case of cat3750 types the first switch is intended to be the first member of a 3750

switch stack and so on. The maximum number of switches per switch stack is 9. For

more than 4 switches with 10GbE ports the StackWise cabling may be a bottleneck.

In case of Nexus types exactly two types have to be specified. It is recommended to

use switches with the same number of ports because ports are needed always in

same quantity from both switches.

For switch types providing expansion module slots, slots can immediately be

occupied by appending modules to the switch type separated with colon.

Switch Administration Adding a Switch Group

222 Administration and Operation

--name <switch group name>

Name string to be set as switch group node name.

--passwd <password>

Clear text password to be set at switch group.

--login <login name>

Login name to be set at switch group. Defaults to flexframe.

--host <ip host part>[,ip host part]

Host part to be used to build IP addresses for the control lan network. If this option is

omitted the script uses a free host number to calculate the IP address. If this option

is used the second definition is necessary if switches of type NEXUS are involved.

--snmp <community name>

Defines the community name. Defaults to public.

--uplinkportcnt <port count for uplink channel>

Count of ports to be used for an uplink channel (default: 2).

--uplinkportmedia {tx|fx|10gb}

Media type of ports to be used for the uplink channel (default: 10GB for NEXUS else

depends on FlexFrame global network settings).

--mgmtswgroup <switch group id>

Defines the switch group the switch management interface should be connected to.

The information is necessary if switches of type NEXUS are involved. Call

ff_swgroup_adm.pl --help to get a list of currently configured switch group IDs.

--vpcdomain <vpc domain id>

Defines a unique vPC domain within the network. The information is necessary if

switches of type NEXUS are involved.

Adding an Expansion Module Switch Administration

Administration and Operation 223

--dryrun

For test purpose you can perform the function without changing the LDAP Database

and updating the switchports.

cn1:~ # ff_swgroup_adm.pl --op add \

--type cat3750x-24t:2x10GbE,cat3750x-24t:2x10GbE \

--name swg6 --passwd passwort

If program is aborted by a failure remove leftovers by calling:

ff_swgroup_adm.pl --op rem --group 6

NOTICE: Ctrl-C will be ignored.

update LDAP

..................................................

........

New SwitchGroup 6 was added to LDAP.

Interconnect the Switch Group as noted below:

SwitchID Port

1 TE2 Uplink port of EtherChannel 1

2 TE2 Uplink port of EtherChannel 1

Upload of initial switch configuration has to be done manually.

Follow the instructions of the applicable Quick Start Hardware

FlexFrame for SAP document in the doc/hwinfo section of the Service CD

and use the following data:

Switch Group swg6

IP and Mask: 172.20.13.31 255.255.255.0

Config: scp://[email protected]//tftpboot/swg6.config

You can find a copy of the message above in

/var/log/FlexFrame/tmp/swgroup_adm-add-6/wiring_of_swgroup6_swg6.txt

9.9 Adding an Expansion Module

A switch may have slots where you can insert expansion modules to get more available

Ports. Use ff_swgroup_adm.pl --op add-module to add the expansion modules

ethernet ports to the FlexFrame configuration as described below.

Switch Administration Removing an Expansion Module

224 Administration and Operation

Synopsis

ff_swgroup_adm.pl --op add-module --group <switch group id>

--switch <switch id>

--slot <slot id>

--module <module type>

Options

--op add-module

Adds an expansion module to a switch.

--group <switch group id>

Defines the switch group to operate.

--switch <switch id>

Defines the switch to operate.

--slot <slot id>

Defines the slot to be used.

--module <module type>

Defines the type of the expansion module to be added.

9.10 Removing an Expansion Module

Before you remove an expansion module physically from a switch use

ff_swgroup_adm.pl --op rem-module to remove the expansion modules ethernet

ports from the FlexFrame configuration as described below.

All ports of the expansion module must be unused.

Synopsis

ff_swgroup_adm.pl --op rem-module --group <switch group id>

--switch <switch id>

--slot <slot id>

Options

--op rem-module

Removes an expansion module from a switch.

Removing a Switch Group Switch Administration

Administration and Operation 225

--group <switch group id>

Defines the switch group to operate.

--switch <switch id>

Defines the switch to operate.

--slot <slot id>

Defines the slot to operate.

9.11 Removing a Switch Group

To remove an entire switch group use the ff_swgroup_adm.pl command. The

command removes data from LDAP database. As a precaution the switch group may no

longer be in use by any FlexFrame component.

All ports of the switchgroup except the ports of one uplink must be unused.

Synopsis

ff_swgroup_adm.pl --op rem --group <switch group id> [--dryrun]

Options

--op rem

Removes named switch group.

--group <switch group id>

Defines switch group to operate. Within this operation mode used to define the id of

the switch group to be removed.

--dryrun

For test purpose you can perform without changing the LDAP Database.

Example

cn1:~ # ff_swgroup_adm.pl --op rem --group 2

update LDAP.

........

Unless any errors are reported disconnect uplink ports

1/49, 2/49

and switch group is removed from FlexFrame environment.

Switch Administration Adding an Uplink to Switch Group

226 Administration and Operation

9.12 Adding an Uplink to Switch Group

Use ff_swgroup_adm.pl --op add-uplink to create an additional uplink at a

switch group. Use the command line arguments to specify options like port count and port

media.

Synopsis

ff_swgroup_adm.pl --op add-uplink --group <switch group id>

[--uplinkportcnt <port count per channel>]

[--uplinkportmedia {tx|fx|10gb}]

[--dryrun]

Options

--op add-uplink

Creates a new uplink link aggregate.

--group <switch group id>

Defines the switch group to change.

--uplinkportcnt <port count per channel>

Count of ports to be used for an uplink channel (default: 2). The maximum number of

ports per channel is 8.

--uplinkportmedia <tx|fx|10gb>

Media type of ports to be used for the uplink channel (default: 10GB for NEXUS else

depends on FlexFrame global network settings).

--dryrun

For test purpose you can perform the function without changing the LDAP Database.

Example

cn1:~ # ff_swgroup_adm.pl --op add-uplink --group 2 --uplinkportcnt 4

--uplinkportmedia fx

If program is aborted by Ctrl-C or a failure remove left overs

by calling:

ff_swgroup_adm.pl --op rem-uplink --group 2 --channel 2

update LDAP

Extend an Uplink of Switch Group Switch Administration

Administration and Operation 227

....

update switch 2/1 configuration

Notice: Update will take about 1 minute.

..........

New uplink with channel id 2 was created for switch group 2.

It was added to switch group and LDAP.

See below the configured uplink ports and connect them to the peer

switch.

SwitchID / Port

1 / 51 Uplink port of EtherChannel 2 to peer switch

1 / 52 Uplink port of EtherChannel 2 to peer switch

2 / 51 Uplink port of EtherChannel 2 to peer switch

2 / 52 Uplink port of EtherChannel 2 to peer switch

Unless any errors are reported cable switch ports to use the uplink

channel.

Look at "/opt/FlexFrame/network/wiring_of_swgroup2__swg2.txt"

to get a copy of this message.

9.13 Extend an Uplink of Switch Group

Use ff_swgroup_adm.pl --op ext-uplink to extend an existing uplink at a switch

group. New ports are added until given port count is reached.

Synopsis

ff_swgroup_adm.pl --op ext-uplink --group <switch group id>

--channel <uplink channel id>

[--uplinkportcnt <port count per channel>]

[--dryrun]

Options

--op ext-uplink

Expands an existing uplink link aggregate.

--group <switch group id>

Defines the switch group to change.

--channel <uplink channel id>

Switch Administration Delete an Uplink of Switch Group

228 Administration and Operation

Defines the uplink channel of switch group to changed.

--uplinkportcnt <port count per channel>

Count of ports to be used for an uplink channel (default: 2). The maximum number of

ports per channel is 8.

--dryrun

For test purpose you can perform the function without changing the LDAP Database

and updating the switchports.

Example

cn1:~ # ff_swgroup_adm.pl --op ext-uplink --group 2 --channel 2 --

uplinkportcnt 8

update LDAP

.......

update switch 2/1 configuration

Notice: Update will take about 1 minute.

...................

Uplink with channel id 2 of switch group 2 was extended

up to 8 ports. Switch group configuration and LDAP are updated.

See below the new configured uplink ports and connect them to the peer

switch.

SwitchID / Port

2 / 51 Uplink port of EtherChannel 2 to peer switch

3 / 50 Uplink port of EtherChannel 2 to peer switch

3 / 51 Uplink port of EtherChannel 2 to peer switch

4 / 51 Uplink port of EtherChannel 2 to peer switch

Unless any errors are reported cable switch ports to use the new uplink

channel ports. Look at

"/opt/FlexFrame/network/wiring_of_swgroup2__swg2.txt"

to get a copy of this message.

9.14 Delete an Uplink of Switch Group

To delete an existing uplink of a switch group use operation mode rem-uplink of

ff_swgroup_adm.pl. It will remove the channel with its link aggregate and all

associated ports at switch group.

Delete an Uplink of Switch Group Switch Administration

Administration and Operation 229

Synopsis

ff_swgroup_adm.pl --op rem-uplink --group <switch group id>

--channel <uplink channel id>

[--dryrun]

Options

--op ext-uplink

Expands an existing uplink link aggregate.

--group <switch group id>

Defines the switch group to change.

--channel <uplink channel id>

Defines the uplink channel of switch group to removed.

--dryrun

For test purpose you can perform the function without changing the LDAP Database

and updating the switchports.

Example

cn1:~ # ff_swgroup_adm.pl --op rem-uplink --group 2 --channel 2

update LDAP

........

update switch 2/1 configuration

Notice: Update will take about 1 minute.

........................

Uplink with channel id 2 removed at switch group 2.

It was removed from switch group and LDAP.

See below the freed uplink ports.

SwitchID / Port

1 / 51

2 / 51

3 / 49

3 / 50

3 / 51

4 / 49

4 / 50

4 / 51

Switch Administration Migrating Switches of a Switch Group

230 Administration and Operation

Unless any errors are reported switch ports of uplink channel are now

unused.

Look at "/opt/FlexFrame/network/wiring_of_swgroup2__swg2.txt"

to get a copy of this message.

9.15 Migrating Switches of a Switch Group

Because of the various supported switch types within a switch group situations may occur

where a switch of a specific type should be replaced with a switch of another type. Such

situations are assisted by switch migration.

Migrating switches of a switch group of type CAT3750-STACK

The following figure illustrates the supported migration scenarios:

from to

cat3750g-24t cat3750g-24ts, cat3750e-24td, cat3750x-24t

cat3750g-24ts cat3750g-48ts, cat3750e-24td1)

, cat3750x-24t:4x1GbE

cat3750e-24td cat3750e-48td, cat3750x-24t:2x10GbE

cat3750g-48ts cat3750e-48td1)

, cat3750x-48t:4x1GbE

cat3750e-48td cat3750x-48t:2x10GbE

cat3750x-24t cat3750x-48t 1)

only possible if the SFP ports are not used

If migration from a to b to c is supported, the direct migration from a to c is also

supported.

This action requires a downtime of the affected switch group.

The following processing has to be performed.

1. Run ff_save_switch_config.pl to save the configuration for fall back.

2. Run ff_swgroup_adm.pl –-op migrate-sw with --dryrun option to check

the migration. If the migration fails on SFP ports used for uplinks and you want to

have other uplink ports after migration, remove the concerned uplink using

ff_swgroup_adm.pl –-op rem-uplink. If the uplink can not be removed

because it is the last uplink, add a temporary uplink with TX ports. If there are no free

TX ports available add additional temporary switches to the switch group.

3. Check the IOS version of the new switches. Switches of the same model (cat3750g,

cat3750e or cat3750x model) must have the same IOS version within the stack. If the

versions are different, upgrade resp. downgrade the IOS.

Migrating Switches of a Switch Group Switch Administration

Administration and Operation 231

4. If all checks are successful write down switch ids of each stack member of the switch

stack as they may change inserting new switches to the stack.

5. Power off all switches of the switch stack

6. Remove all network and backplane cables from the switches you want to replace.

7. Replace the switches.

8. Plug in all network cables and backplane cables into the new switches.

9. Power on all switches of the switch stack.

10. Compare the actual switch ids with your noticed ids. In case of differences use the

following IOS command for renumbering: switch <Switch Number> renumber <New switch number>

11. Use ff_swgroup_adm.pl –-op migrate-sw without --dryrun option to

complete the migration.

12. If necessary add a new uplink using ff_swgroup_adm.pl –-op add-uplink,

remove the temporary added uplink and the temporary added switches.

The switch group is now ready for use or further configuration.

Migrating switches of a switch group of type NEXUS5x00-VPC

The following figure illustrates the supported migration scenarios:

from to

nexus5010 nexus5020, nexus55481)

nexus5020 nexus55961)

nexus5548 nexus5596 1)

only possible if all switches of the switch group are migrated

If migration from a to b to c is supported, the direct migration from a to c is also

supported. If module ports are used the same ports must be available in the new

environment.

This action requires a downtime of the affected switch group.

The following processing has to be performed.

1. Run ff_save_switch_config.pl to save the configuration for fall back.

2. Check the NX-OS version of the new switches. Switches in the same vPC domain

should have the same NX-OS version. If the versions are different, upgrade resp.

downgrade the NX-OS.

Switch Administration Migrating Switches of a Switch Group

232 Administration and Operation

3. Run ff_swgroup_adm.pl –-op migrate-sw with --dryrun option to check

the migration. Omit --switch if all switches of the switch group should be

migrated to the same switch type.

4. Power off all switches of the switch group.

5. Remove all network cables from the switches you want to replace.

6. Replace the switches.

7. Plug in all network cables into the new switches.

8. Power on all switches of the switch group.

9. Use ff_swgroup_adm.pl –-op migrate-sw without --dryrun option to

accomplish the migration in LDAP.

10. Run ff_swgroup_adm.pl –-op gen-config to get the configuration of the

switch group switches and upload the configuration of the new switches according to

the corresponding HW-Characteristics Quickguide.

The switch group is now ready for use or further configuration.

Synopsis

ff_swgroup_adm.pl --op migrate-sw --group <switch group id>

[--switch <switch id>] --type <switch type>

[--dryrun]

Options

--op migrate-sw

Migrates the type of switches of a switch group.

--group <switch group id>

Defines the switch group to be used.

--switch <switch id>

Defines the switch within the switch group to be migrated. If omitted all switches of

the switch group are migrated to the given switch type.

--type <switch type>

Defines the type the switch should migrate to. Call ff_swgroup_adm.pl without

any parameter to get a list of supported switch types. Migration is allowed according

to the migration table above.

For switch types providing expansion module slots, slots can immediately be

occupied by appending modules to the switch type separated with colon.

Adding a Switch Port Configuration Switch Administration

Administration and Operation 233

--dryrun

For test purpose you can perform the function without changing the LDAP Database.

Example

cn1:/opt/FlexFrame/bin # ff_swgroup_adm.pl --op migrate-sw --group 1 –

switch 2 --type cat3750e-24td

9.16 Adding a Switch Port Configuration

Switch ports are typically directly configured by maintenance tools. But some issues like

configuring ports for gateways, backup or migration systems need a way to do this on a

per port basis. For this type of configuration a special peer type is used. The program ff_swport_adm.pl is used to configure or remove this type of port configuration.

Synopsis

ff_swport_adm.pl --op add --port <swgroup:switch:port>

[ --10gbit]

--lan <pool:lan[:lan][,pool:lan[:lan]]>

[--native <pool:lan>]

[--desc <description>]

[--dryrun]

Options

--op add

Adds a switch port configuration and displays some information about processing

steps.

--port <swgroup:switch:port>

Defines the switch group, switch and port ID of the port to be used.

--10gbit

Defines the port number used in --port to be the number of a TenGigabitEthernet

port.

--lan <pool:lan[:lan][,pool:lan[:lan]]>

Defines the accessible VLANs. For better readability, a VLAN is specified with its

pool and LAN name. Use only client, server or storage as LAN names. For more

than one LAN per pool the LAN names may be added to the same pool statement.

The VLANs are not restricted to belong to the same pool. To directly add VLAN IDs

not used within any pool, use '#' as pool name and the VLAN ID(s) as LAN(s).

Switch Administration Adding a Switch Port Configuration

234 Administration and Operation

The accessible VLANs. For better readability, a VLAN is specified with its pool and

LAN name. Use only client, server or storage as LAN names. For more than one

LAN per pool the LAN names may be added to the same pool statement. The

VLANs are not restricted to belong to the same pool. To directly add VLAN IDs not used within any pool, use '#' as pool name and the VLAN ID(s) as LAN(s).

If only a single VLAN is configured this is accessible as native VLAN. This means

the data packet contains no VLAN tag. This is the behavior used by a standard

server network interface. For more than one LAN they are configured as tagged. To

define which of them should be used as native VLAN use the option --native.

Examples:

--lan poolA:client:server,poolB:client:server

--lan poolA:client,poolA:server,poolB:client:server

--lan poolA:storage,poolB:storage

--lan poolA:server

--lan '#:417:891'

--lan poolA:server,'#:417:891'

--lan 'poolA:server,#:417:891'

--native <pool:lan>

Use this option to define the native VLAN of the accessible VLANs defined with

option --lan. To directly add VLAN ID not used within any pool, use '#' as pool

name and the VLAN ID as LAN.

Examples:

--native poolA:server

--native '#:417'

--desc <description>

The description is added to configuration of switch port and the LDAP data of the

switch port configuration.

--dryrun

For test purpose you can perform the function without changing the LDAP Database

and updating the switchports.

Example

cn1:/opt/FlexFrame/bin # ff_swport_adm.pl --op add --port 1:1:15

--lan ip1:storage:server,'#:4000' --native ip1:storage

Execution may take some minutes. If program is aborted

by Ctrl-C or a failure remove left overs by calling:

Removing a Switch Port Configuration Switch Administration

Administration and Operation 235

ff_swport_adm.pl --op rem --port 1:1:15

update switch 1/1 configuration

Notice: Update will take about 1 minute.

...........+

If not reported any error the port is configured and LDAP is updated

successfully.

9.17 Removing a Switch Port Configuration

To remove a switch port configuration that was previously configured by

ff_swport_adm.pl or as external connectivity with the Management Tool, use the

ff_swport_adm.pl command. Other ports are configured with maintenance tools like

ff_an_adm.pl, ff_pool_adm.pl or ff_bx_cabinet_adm.pl.

The switch port configuration will be removed from the switch and the LDAP database.

Synopsis

ff_swport_adm.pl --op rem --port <swgroup:switch:port> [ -10gbit]

Options

--op rem

Removes the configuration of a switch port and displays some information about

processing steps.

--port <swgroup:switch:port>

Defines the switch group, switch and port ID of the port to be used.

Switch Administration Displaying a Switch Port Configuration

236 Administration and Operation

--10gbit

Defines the port number used in --port to be the number of a TenGigabitEthernet

port.

--dryrun

For test purpose you can perform the function without changing the LDAP Database

and updating the switchports.

Example

cn1:/opt/FlexFrame/bin # ff_swport_adm.pl --op rem --port 1:1:15

Execution may take some minutes. If program is aborted

by Ctrl-C or a failure remove left overs by calling:

ff_swport_adm.pl --op rem --port 1:1:15

update switch 1/1 configuration

Notice: Update will take about 1 minute.

.............+

If not reported any error the port is unconfigured and LDAP is updated

successfully.

9.18 Displaying a Switch Port Configuration

To display the configuration of a switch port as known by LDAP database in detail, use the command ff_swport_adm.pl with operation mode list.

Synopsis

ff_swport_adm.pl --op list --port <swgroup:switch:port> [ -10gbit]

Options

--op list

Displays configuration of the switch port.

--port <swgroup:switch:port>

Defines the switch group, switch and port ID of the port to be used.

--10gbit

Defines the port number used in --port to be the number of a TenGigabitEthernet

port.

Displaying the Complete Switch Port Configuration Switch Administration

Administration and Operation 237

Examples

cn1:/opt/FlexFrame/bin # ff_swport_adm.pl --op list --port 1:1:4

Switch Port Configuration of

1:1:4 (Switch Group : Switch : Port)

assigned VLAN IDs: 24,25,26

assigned VLAN Names: pool1:client,pool1:server,pool1:storage

native VLAN: 26

Port Peer Type: AN

Peer Node: rx300-1

Display of an unconfigured port:

ERROR: wrong switch port "1:1:8".

Port configuration unknown.

9.19 Displaying the Complete Switch Port Configuration

To display the complete configuration of all switch ports as known by LDAP database in

detail, use the command ff_swport_adm.pl with operation mode list-all.

Synopsis

ff_swport_adm.pl --op list-all

Options

--op list-all

Displays configuration of all switch groups.

Switch Administration Displaying the Complete Switch Port Configuration

238 Administration and Operation

Examples

cn1:/opt/FlexFrame/bin # ff_swport_adm.pl --op list-all

Switch Port Configuration of

1:1:1 (Switch Group : Switch : Port)

assigned VLAN IDs: 1010,1011,1012

assigned VLAN Names: pool1:client,pool1:storage,pool1:server

native VLAN: 1011

Port Peer Type: GW

Peer Node: sno1apl5p1

Switch Port Configuration of

1:1:2 (Switch Group : Switch : Port)

assigned VLAN IDs: 1010,1011,1012

assigned VLAN Names: pool1:client,pool1:storage,pool1:server

native VLAN: 1011

Port Peer Type: AN

Peer Node: sno1apl2

Switch Port Configuration of

1:1:3 (Switch Group : Switch : Port)

assigned VLAN IDs:

1,90,91,92,93,94,95,1010,1011,1012,1013,1020,1021,1022,1030,1031,1032

assigned VLAN Names: -:default-

vlan,toni:storage,toni:client,toni:server,toni1:storage,toni1:client,ton

i1:server,

pool1:client,pool1:storage,pool1:server,-

:control,pool2:client,pool2:storage,pool2:server,pool3:client,pool3:stor

age,pool3:

server

Port Channel ID: 3

Port Peer Type: SWB

Switch Port Configuration of

1:1:4 (Switch Group : Switch : Port)

assigned VLAN IDs:

1,90,91,92,93,94,95,1010,1011,1012,1013,1020,1021,1022,1030,1031,1032

assigned VLAN Names: -:default-

vlan,toni:storage,toni:client,toni:server,toni1:storage,toni1:client,ton

Moving Device Connection to Core Switch Switch Administration

Administration and Operation 239

i1:server,

pool1:client,pool1:storage,pool1:server,-

:control,pool2:client,pool2:storage,pool2:server,pool3:client,pool3:stor

age,pool3:

server

Port Channel ID: 4

Port Peer Type: SWB

...

9.20 Moving Device Connection to Core Switch

Devices (e.g. application nodes or NAS storage etc) in FlexFrame are usually connected

to the FlexFrame internal (LAN) switch groups. When the configuration grows e.g. if a lot

of servers connected to several switch groups occupy the same NAS system it would be

more convenient if devices e.g. like this NAS system are directly connected to core

switches. Core switches are customer switches the switch groups have uplinks to. Core

switches are not under control of FlexFrame and not represented in LDAP. Devices di-

rectly connected to core switches are still represented in LDAP with a connection to an

imaginary switch group zero. For more detailed information to the core network concept

see also: FlexFrame for SAP Network Design and Configuration Guide.

ff_move_to_core.pl supports the move of device connections to core switches. The

switch group ports occupied from the device to be moved are displayed and if the --doit

option is used the configuration of the affected ports is changed and an update of LDAP

is performed.

Synopsis

ff_move_to_core.pl --help

ff_move_to_core.pl --version

Options

--help

Display usage.

--version

Display script version.

9.20.1 Move Control Center to Core Switch

Synopsis

ff_move_to_core.pl --op cn [--doit]

Switch Administration Moving Device Connection to Core Switch

240 Administration and Operation

Options

--op cn

The switch group ports of both control nodes should be released. The affected ports

are displayed.

--doit

The configuration of the affected ports is changed and an update of LDAP is per-

formed.

Example

cn1:/opt/FlexFrame/bin # ff_move_to_core.pl --op cn

9.20.2 Move Client LAN to Core Switch

Synopsis

ff_move_to_core.pl --op clan [--doit]

Options

--op clan

The switch group ports of the client LAN should be released. The affected ports are

displayed

--doit

The configuration of the affected ports is changed and an update of LDAP is per-

formed.

Example

cn1:/opt/FlexFrame/bin # ff_move_to_core.pl --op clan

9.20.3 Move NAS System to Core Switch

Synopsis

ff_move_to_core.pl --op nas --name <node name> [--doit]

Moving Device Connection to Core Switch Switch Administration

Administration and Operation 241

Options

--op nas

The switch group ports of a NAS system should be released. The affected ports are

displayed.

--name <node name>

The node name of the NAS system which should be moved.

--doit

The configuration of the affected ports is changed and an update of LDAP is per-

formed.

Example

cn1:/opt/FlexFrame/bin # ff_move_to_core.pl --op nas --name filer1

9.20.4 Move Application Node to Core Switch

Synopsis

ff_move_to_core.pl --op an --name <node name> [--doit]

Options

--op an

The switch group ports of an application node should be released. The affected

ports are displayed.

--name <node name>

The node name of the application node which should be moved.

--doit

The configuration of the affected ports is changed and an update of LDAP is per-

formed.

Example

cn1:/opt/FlexFrame/bin # ff_move_to_core.pl --op an --name rx300

Switch Administration Moving Device Connection to Core Switch

242 Administration and Operation

9.20.5 Move ESX Server to Core Switch

Synopsis

ff_move_to_core.pl --op esx --name <node name> [--doit]

Options

--op esx

The switch group ports of an application node should be released. The affected

ports are displayed.

--name <node name>

The node name of the application node which should be moved.

--doit

The configuration of the affected ports is changed and an update of LDAP is per-

formed.

Example

cn1:/opt/FlexFrame/bin # ff_move_to_core.pl --op esx --name rx300

9.20.6 Move BX Cabinet to Core Switch

Synopsis

ff_move_to_core.pl --op bx --name <cabinet name> [--doit]

Options

--op bx

The switch group ports of a BX cabinet should be released. The affected ports are

displayed.

--name <cabinet name>

The cabinet name of a BX chassis which should be moved.

--doit

The configuration of the affected ports is changed and an update of LDAP is per-

formed.

Moving Device Connection to Core Switch Switch Administration

Administration and Operation 243

Example

cn1:/opt/FlexFrame/bin # ff_move_to_core.pl --op bx --name bx61

SAP System Handling Moving Device Connection to Core Switch

244 Administration and Operation

10 SAP System Handling

This chapter describes the management of SAP System IDs (so-called SIDs) and their

respective instances within the FlexFrame environment. It further describes how to clone

SIDs as well as their instances for a different pool than the one they were installed to.

The tools described below only maintain the LDAP database entries, rather than adding

or removing the data and binaries of the respective SAP systems. These steps need to

be performed manually.

Listing, adding, removing and cloning the above entities in the LDAP server is supported

by two tools, ff_sid_adm.pl and ff_clone_sid.pl. Both scripts will take care of

keeping the data accessed by the operating system's naming service mechanism in sync

with the FlexFrame internal configuration data, both of which reside in the LDAP server.

This data should not be manipulated manually.

Preventing port conflicts

Please take into account that depending on the instance number ports are

reserved for the SAP application. That could mean that you conflict with other

non-SAP applications on your nodes (e.g. SAP's ICM uses ports 80nn which

maybe conflicts with an application which uses port 8081). Please refer to

SDN document ―TCP/IP Ports Used by SAP Applications‖ to prevent conflicts.

If a new SAP system is added by ff_sid_adm.pl and the modification of the LDAP

database fail, ff_sid_adm.pl always tries to remove all changes in LDAP which are

made until the failure This rollback may cause error messages because the rollback

operation does not exactly know where the error occurred (it cannot recognize the failed

command). Those messages can be ignored.

The script ff_sid_adm.pl tries to generate files to support diagnostics. Those files are:

/var/log/FlexFrame/tmp/ff_sid_adm.pl/ldap.ldifs.<pool>.<SID>

/ var/log/FlexFrame/tmp/ff_sid_adm.pl/ldap.error.<pool>.<SID>

/ var/log/FlexFrame/tmp/ff_sid_adm.pl//ldap.log.pool>.<SID>

/ var/log/FlexFrame/tmp/ff_sid_adm.pl//rollback.<pool>.<SID>

In addition to the decribed SAP System handling in this chapter here there is another

possibility to handle SAP systems using ACC (Adaptive Computing Controller (ACC) -

SAP system for monitoring and controlling SAP environments). For more information see

manual "Installation ACC" and additional documentation from SAP (ACCImplementa-

tion.pdf, ACCSecurity.pdf, ACCCustomizing.pdf).

Listing SAP SIDs and Instances SAP System Handling

Administration and Operation 245

10.1 Listing SAP SIDs and Instances

Synopsis

ff_sid_adm.pl –-op list-sids [ –-pool <pool name> ]

ff_sid_adm.pl --op list-ids [ --pool <pool name> --sid <SID> ]

ff_sid_adm.pl –-op list-all [ –-pool <pool name> --sid <SID>]

Options

--op list-sids

The SIDs of the specified pool/all pools are displayed with information on SAP

version, database type and database version

--op list-ids

The instance numbers of the specified SID are displayed related to the given

pool/SID or all pools/all SIDs.

--op list-all

Shows extended information about a specific SID (instance types, instances

numbers, used ip@, used hostnames) in a specific pool.

--pool <pool name>

Specifies the FlexFrame pool to which the operation should be applied.

--sid <SAP system id>

The instance numbers of the specified SID are displayed related to the given pool.

Examples

%> ff_sid_adm.pl –-op list-ids –-pool Pan –-sid sht

Configuration Data of SID – instance number and types: SHT (Pool: Pan)

05 scs

DB0 db

07 ci

SAP System Handling Listing SAP SIDs and Instances

246 Administration and Operation

%> ff_sid_adm.pl –-op list-sids –-pool Pan

List of SIDs in pool: Pan

O01 SAP-7.1 Oracle-10

S02 SAP 7.0 MaxDB-76

%> ff_sid_adm.pl –-op list-all –-pool Pan –-sid SHT

Configuration data of SID: SHT (Pool: Pan)

SID global Data:

SAP version specified: SAP-7.0

Database specified: Oracle-10

Instance Data:

Instance type: ci

Instance number: 00

Client-Lan Host-Ip: nnn.nn.nn.nnn

Client-Lan Hostname: cisht

Server-Lan Host-IP: nnn.nn.mm.nnn

Serber-Lan Hostname: cisht-se

Updating System Configuration Files SAP System Handling

Administration and Operation 247

10.2 Updating System Configuration Files

Synopsis

ff_sid_adm.pl --op db2adm --pool <pool name> --sid <SAP system id>

Options

--op db2adm

Updates the system configuration files with DB2 specific data (host name, service

names) for a specific SID.

--pool <pool name>

Specifies the FlexFrame pool to which the operation should be applied.

--sid <SAP system id>

Specifies the SID being used.

10.3 Adding/Removing/Modifying SAP SIDs and Instances (Classic SAP Services)

Synopsis

ff_sid_adm.pl –-op add –-pool <pool name> --sid <SAP system id>

--sapversion {4.6|6.20|6.40|7.0| 7.1|7.2|7.3}

--db {ORACLE{9|10|11}|MAXDB{75|76|77|78}|DB2V{91|95|97}}

:{<db_loghost_ip>|*}

--lc {MAXDB75|MAXDB76|MAXDB77|MAXDB78} {<lc_loghost_ip>|*}

--group <groupname1>:<gidnumber1>,<groupname2>:<gidnumber2>,...

--user <username1>:<uidnumber1>,<username2>:<uidnumber2>,...

:{<db_loghost_ip>|*}

--sap {ci|app|jc|j|scs|ascs|ers}:<SYSNR>:

:{<loghost_client_ip>|*}

:{<loghost_server_ip>|*}

[--os <instance_type>:<os>,<instance_type>:<os>,...]

[--ips <old_ip>:<new_ip>,<old_ip>:<new_ip>,...]

[--db2srv sapdb2<SID>:<port>,DB2_db2<sid>:<port>,

DB2_db2<sid>_1:<port>, DB2_db2<sid>_2:<port>,

DB2_db2<sid>_END:<port>]

SAP System HandlingAdding/Removing/Modifying SAP SIDs and Instances (Classic SAP Services)

248 Administration and Operation

[--sapdata <nas system>:/vol/<volume path>]

[--saplog <nas system>:/vol/<volume path>]

ff_sid_adm.pl –-op mod –-pool <poo _name> --sid <SAP system id>

[--os <instance_type>:<os>,<instance_type>:<os>,...]

ff_sid_adm.pl –-op mod –-pool <pool name> --sid <SAP system id>

[--ips <old_ip>:<new_ip>,<old_ip>:<new_ip>,...]

ff_sid_adm.pl –-op del –-pool <pool name> --sid <SAP system id>

[--sysnr <SYSNR>]

Options

--op add

Determines the add operation.

--op mod

Determines the mod operation. This option is only used to modify OS specifications

and exchanges IP adresses of specific SID instances.

--op { del | rem }

Determines the delete operation.

--pool <poo _name>

Specifies the FlexFrame pool to which the operation should be applied.

--sid <SAP system id>

Specifies the SID being used.

--sapversion {4.6|6.20|6.40|7.0|7.1}

Specifies the SAP basis version being used.

--db {ORACLE{9|10}|MAXDB{75|76|77|78}|DB2V{91|95|97}}

Specifies the database type as well as the respective version being used. The

specification of the database type is not case-sensitive. Please take care of the

restrictions given by SAP (e.g. SAP 7.3 only allows Oracle11, MaxDB78 or

DB2V97).

Adding/Removing/Modifying SAP SIDs and Instances (Classic SAP Services)SAP System Handling

Administration and Operation 249

--lc {MAXDB75|MAXDB76|MAXDB77}

Specifies the database type as well as the respective version being used. The

specification of the database type is not case-sensitive.

--group <groupname1>:<gidnumber1>,<groupname2>:<gidnumber2>,...

--user <username1>:<uidnumber1>,<username2>:<uidnumber2>,...

User and group enable specially selected user numbers and group numbers to be

assigned to SAP users and SAP groups respectively. In this case a check is made to

see whether the user or group has already been defined for the DB system involved.

A user or group is created only if they do not already exist. For example, a group

dba which already exists cannot be assigned a group number which deviates from

the default value.

{<db_loghost_ip>|*}}

The logical host name is used for the database (generated automatically: db<sid>-

se) as well as the IP address for that host name. Use an asterisk if you want it to be

chosen automatically. All the entries need to be specified in a colon separated for-

mat.

You can omit the network part of the IP, e.g. 10.10.12, and specify only the last

tupel of the IP, e.g. 123.

{<lc_loghost_ip>|*}}

The logical host name is used for the database (generated automatically: lc<sid>-

se) as well as the IP address for that host name. Use an asterisk if you want it to be

chosen automatically. All the entries need to be specified in a colon separated for-

mat.

You can omit the network part of the IP, e.g. 10.10.12, and specify only the last

tupel of the IP, e.g. 123.

--sap {ci|app|jc|j|scs|ascs|ers}:<SYSNR>:

{<loghost_client_ip>|*}

{<loghost_server_ip>|*}

Specifies an SAP instance (optionally multiple of those) through its type (ci, app,

jc, j, scs, ascs, ers), its SAP system number, the logical host name

(generated automatically: <type><sid>) in the client network, the respective IP

address, the logical host name (generated automatically: <type><sid>-se) in the

server network and the respective IP address. Again, the IP addresses can be

replaced with asterisks in order to have them chosen automatically.

ERS is only supported with --sapversion 7.0 or higher. For Enqueue Repli-

cated Servers (ERS) the asterisk should be used because we do not need an IP

address from the client and server LAN (given IP addresses will be ignored).

SAP System HandlingAdding/Removing/Modifying SAP SIDs and Instances (Classic SAP Services)

250 Administration and Operation

All the entries need to be specified in a colon separated format.

With SAP 7.1 up SAP systems do not know instance type jc for JAVA systems.

Therefore the hostname used changed from jc<sid> to j<sysnr><sid> and from

jc<sid>-se to j<sysnr><sid>-se.

ff_sid_adm.pl still requires the type jc used in option —-sap to make a distinc-

tion between central instance and application instance. Please take into account the

different hostname specifications depending on SAP version used.

--sysnr <SYSNR>

Removes a specific SAP instance instead of the entire system (SID).

--os <instance_type>:<os>,<instance_type>:<os>,...]

Specifies the OS type for the given SID and/or their instances.

instance_type::= {default|ci|app|jc|j|scs|ascs|ers}

os::= {SLES-10.x86_64|SLES-11.x86_64}

In combination with the –-add option, default:<os> sets the operating system of

the SID itself and all their instances which no own operating system is assigned to.

In combination with the –-mod option, default:<os> sets the operating system of

the SID only. The specifications of their instances are not changed.

Examples:

" default:SLES-10.x86_64 - all instances are set to SLES-10.x86_64

" default:SLES-10.x86_64,ci:SLES-11_x86_64 - all instances are set to SLES-

10.x86_64 except instance CI is set to SLES-11

" DB: SLES-11.x86_64,SCS: SLES-10.x86_64, … - for each instance an OS is set

explicitely

The consistence of the given instance type/OS combinations is not checked

by FlexFrame. For a list of allowed combinations see SAP note 1067221.

--ips <old_ip>:<new_ip>,<old_ip>:<new_ip>,...]

Allows to exchange the IP addresses of specific SID instances. This option is only

used with —-op mod. You have to specify the full IP you want to exchange.

ff_sid_adm.pl searches for the specific instances within a SID and exchanges all

corresponding entries in LDAP concerned by that request.

Please pay attention that you make critical changes within your configuration.

We strongly recommend to save a backup of the LDAP database before changing

any IP addresses.

Adding/Removing/Modifying SAP SIDs and Instances (Classic SAP Services)SAP System Handling

Administration and Operation 251

--db2srv sapdb2<SID>:<port>,DB2_db2<sid>:<port>,

DB2_db2<sid>_1:<port>,DB2_db2<sid>_2:<port>,

DB2_db2<sid>_END:<port>

This option is only used for DB2, otherwise it is ignored. You can specify a list of

services you need. The services are written to LDAP and the pool specific

/etc/services.

--sapdata <nas system>:/vol/<volume path>

Allows to request an implicit call of ff_sid_mnt_adm.pl to relocate location to store

sapdata. It is expected that the requested volume already exists. The option is

valid only with services using a database.

--saplog <nas system>:/vol/<volume path>

Allows to request an implicit call of ff_sid_mnt_adm.pl to relocate location to store

saplog. It is expected that the requested volume already exists. The option is

valid only with services using a database.

Examples

Adding an SID with one Central Instance:

control1:~ # ff_sid_adm.pl –-op add –-sid SHT –-pool Otto --sapversion

6.40

–-db ORACLE9:192.168.1.1 --sap ci:00:\*:\*

Adding an instance to an existing SAP System:

control1:~ # ff_sid_adm.pl –-op add –-sid SHT –-pool Otto --sapversion

6.40

--sap app:01::\*:\*

Adding a SID for LiveCache:

control1:~ # ff_sid_adm.pl --op add --pool test --sid LCA --sapversion

7.0

--lc MAXDB76:\*

Adding a SID with ERS support:

control1:~ # ff_sid_adm.pl --op add --pool pool1 –-sid S04 --sapversion

7.0

--db ORACLE10:dbs04-se:10.10.12.100 --sap ers:12:\*:\*

SAP System HandlingAdding/Removing/Modifying SAP SIDs and Instances (Classic SAP Services)

252 Administration and Operation

--sap ci:13:10.10.11.102:10.10.12.102

Adding a SID with operation system:

control1:~ # ff_sid_adm.pl --op add --pool pool1 –-sid OS1 --sapversion

7.0

--db ORACLE10:10.10.12.159

--sap ci:57:10.10.10.157:10.10.12.157

--sap ascs:55::10.10.10.155:10.10.12.155

--os default:SLES-10.x86_64,ascs:SLES-11.x86_64

Adding a SID with DB2 services:

control1:~ # ff_sid_adm.pl --op add --pool pool1 –-sid LB5 --sapversion

7.0

--db DB2V91:10.10.12.159

--sap ci:57:10.10.10.157:10.10.12.157

--sap ascs:55:10.10.10.155:10.10.12.155

--db2srv

sapdb2LB5:60000,DB2_db2lb5:60001,DB2_db2lb5_1:60002,DB2_db2lb5_2:60003,

DB2_db2lb5_END:60004

Adding a SID

control1:~ # ff_sid_adm.pl --op add --pool pool1 –-sid JA0 --sapversion

7.0

--db MAXDB77:d10.10.12.158

--sap jc:57:10.10.10.157:10.10.12.157

--sap j:58:10.10.10.156:10.10.12.156

--sap ascs:55:10.10.10.155::10.10.12.155

Adding a SID with old syntax (SAP 7.1, JAVA)

control1:~ # ff_sid_adm.pl --op add --pool pool1 –-sid JA1 --sapversion

7.1

--db ORACLE10:10.10.12.158

--sap jc:57:10.10.10.157:10.10.12.157

--sap j:58:10.10.10.156:10.10.12.156

--sap ascs:55:10.10.10.155:10.10.12.155

Removing an entire SID (including its instances):

%> ff_sid_adm.pl –-op del –-sid SHT –-pool Otto

Removing an Application Server:

%> ff_sid_adm.pl –-op del –-sid SHT –-pool Otto –-sysnr 01

Removing SAP SIDs and Instances SAP System Handling

Administration and Operation 253

Modifying IP adresses:

%> ff_sid_adm.pl –-op mod –-sid SHT –-pool Otto

–ips 10.10.10.123:10.10.10.132,10.10.12.123:10.10.12.132

Modifying OS version of the SID entry:

%> ff_sid_adm.pl –-op mod –-sid SHT –-pool Otto

-–os "default:SLES-10.x86_64,ci:SLES-9.x86_64"

10.4 Removing SAP SIDs and Instances

Synopsis

ff_sid_adm.pl –-op del –-pool <pool name> --sid <SAP system id>

ff_sid_adm.pl –-op [ del | rem ] –-pool <pool name> \

--sid <SAP system id> [--sysnr <SYSNR>]

Options

--op { del | rem }

Determines the delete operation.

--pool <pool name>

Specifies the FlexFrame pool to which the operation should be applied.

--sid <SAP system id>

Specifies the SID being used.

--sysnr <SYSNR>

Removes a specific SAP instance instead of the entire system (SID).

Examples

Removing an entire SID (including all instances):

%> ff_sid_adm.pl –-op del –-sid SHT –-pool Otto

Removing an specifc instance:

%> ff_sid_adm.pl –-op del –-sid SHT –-pool Otto –-sysnr 01

SAP System Handling Adding/Removing SAP SIDs (addon services)

254 Administration and Operation

10.5 Adding/Removing SAP SIDs (addon services)

10.5.1 BOBJ – Business Objects

Synopsis

ff_sid_adm.pl –-op add –-pool <pool name> --sid <SAP system id> \

--sapversion {3.1|3.2|4.0} \

--bobj <client lan hostip> --os <spec>

ff_sid_adm.pl –-op del –-pool <pool name> --sid <SAP system id>

Options

--op add | del

Determines the operation, add or delete.

--pool <pool name>

Specifies the FlexFrame pool to which the operation should be applied.

--sid <SAP system id>

Specifies the SID being used.

--sapversion {3.1|3.2|4.0}

Specifies the SAP basis version being used.

--bobj <client lan hostip>

Specifies the hostip in the client lan for BOBJ service.

--group <groupname1>:<gidnumber1>,<groupname2>:<gidnumber2>,...

--user <sid>adm:<uidnumber>

user and group enable specially selected user numbers and group numbers to be

assigned to SAP users and SAP groups respectively. In this case a check is made to

see whether the user or group has already been defined for the SAP system

involved. A user or group is created only if they do not already exist. For example, a

group sapsys which already exists cannot be assigned a group number which

deviates from the default value.

--bobj <client lan hostip>

Adding/Removing SAP SIDs (addon services) SAP System Handling

Administration and Operation 255

The IP address for the clien lan host name (generated internally: bobj<sid>). Use

an asterisk if you want it to be chosen automatically.

You can omit the network part of the IP, eg. 10.10.12, and specify only the last

tupel of the IP, e.g. 123.

--os <instance_type>:<os>,<instance_type>:<os>,...]

Specifies the OS type for the given SID and/or their instances.

instance_type::= {default|bobj}

os::= {SLES-10.x86_64|SLES-11.x86_64}

In combination with the –-add option, default:<os> sets the operating system of the

SID itself and all their instances which no own operating system is assigned to.

In combination with the –-mod option, default:<os> sets the operating system of the

SID only. The specifications of their instances are not changed.

The consistence of the given instance type/OS combinations is not checked

by FlexFrame. For a list of allowed combinations see SAP note 1067221.

Examples

Adding a SID with BOBJ service:

control1:~ # ff_sid_adm.pl –-op add –-sid bob –-pool Otto --sapversion

3.2 \

––bobj 160

Removing a BOBJ service:

control1:~ # ff_sid_adm.pl –-op del –-sid bob –-pool Otto

10.5.2 Content Server (CMS)

Adding a CMS service at first time means that you need to specify all the instances

required by CMS (database and cms client spec).

CMS service consists of

a database instance

a client instance

SAP System Handling Adding/Removing SAP SIDs (addon services)

256 Administration and Operation

Synopsis

ff_sid_adm.pl –-op add –-pool <pool name> --sid <SAP system id> \

--db MaxDB76:{<db_loghost_ip>|\*} \

--cms <client lan_ip>|\*} \

--sapversion 6.40 [--os <spec> ] \

[ --sapdata <nas_system>:/>volume path> ]

[ --saplog <nas_system>:/>volume path> ]

ff_sid_adm.pl –-op del –-pool <pool name> --sid <SAP system id>

Options

--op add | del

Determines the operation, add or delete.

--pool <pool name>

Specifies the FlexFrame pool to which the operation should be applied.

--sid <SAP system id>

Specifies the SID being used.

--sapversion {6.40}

Specifies the SAP basis version being used.

--db MaxDb76:{<db loghost ip> | \*}

Specifies the database version being used and the server lan host IP

--cms <client lan hostip>

Specifies the hostip in the client lan for BOBJ service.

{<db_loghost_ip>|*}}

The logical host name (generated internally: db<sid>-se) is used for the database as

well as the IP address for that host name. Use an asterisk if you want it to be chosen

automatically. All the entries need to be specified in a colon separated format.

You can omit the network part of the IP, e.g. 10.10.12, and specify only the last

tupel of the IP, e.g. 123.

--cms <client lan hostip>

Adding/Removing SAP SIDs (addon services) SAP System Handling

Administration and Operation 257

The IP address for that host name (generated internally: cms<sid>). Use an asterisk

if you want it to be chosen automatically.

You can omit the network part of the IP, e.g. 10.10.12, and specify only the last

tupel of the IP, e.g. 123.

--group <groupname1>:<gidnumber1>,<groupname2>:<gidnumber2>,...

--user <username1>:<uidnumber1>,<username2:uidnumber2>,…

user and group enable specially selected user numbers and group numbers to be

assigned to SAP users and SAP groups respectively. In this case a check is made to

see whether the user or group has already been defined for the SAP system

involved. A user or group is created only if they do not already exist. For example, a

group sapsys which already exists cannot be assigned a group number which

deviates from the default value.

--os <instance_type>:<os>,<instance_type>:<os>,...]

Specifies the OS type for the given SID and/or their instances.

instance_type::= {default|bobj}

os::= {SLES-10.x86_64|SLES-11.x86_64}

In combination with the –-add option, default:<os> sets the operating system of the

SID itself and all their instances which no own operating system is assigned to.

--sapdata <nas system>:/vol/<volume path>

Allows to request an implicit call of ff_sid_mnt_adm.pl to relocate location to store

sapdata. It is expected that the requested volume already exists. The option is valid

only with services using a database.

--saplog <nas system>:/vol/<volume path>

Allows to request an implicit call of ff_sid_mnt_adm.pl to relocate location to store

saplog. It is expected that the requested volume already exists. The option is valid

only with services using a database.

In combination with the –-mod option, default:<os> sets the operating system of the

SID only. The specifications of their instances are not changed.

The consistence of the given instance type/OS combinations is not checked

by FlexFrame. For a list of allowed combinations see SAP note 1067221.

Examples

Adding a SID with CMS service:

SAP System Handling Adding/Removing SAP SIDs (addon services)

258 Administration and Operation

control1:~ # ff_sid_adm.pl –-op add –-sid cmx –-pool Otto --sapversion

6.40 \

--db MaxDB76:170 \

––cms 170 ]

Adding/Removing SAP SIDs (addon services) SAP System Handling

Administration and Operation 259

Adding a CMS client instance:

control1:~ # ff_sid_adm.pl –-op add –-sid cmx –-pool Otto --sapversion

6.40 \

––cms 170 ]

Removing a CMS Service:

control1:~ # ff_sid_adm.pl –-op rem –-sid cmx –-pool Otto

Removing the CMS client instance:

control1:~ # ff_sid_adm.pl –-op rem –-sid cmx –-pool Otto –sysnr CMS

10.5.3 MDM – Master Data Management

Adding a MDM service for the first time means that you need at least to specify the

database instance.

MDM service contains:

a database instance

a number of services of type 'mds'

a number of services of type 'mdss'

a number of services of type 'mdis'

SAP System Handling Adding/Removing SAP SIDs (addon services)

260 Administration and Operation

Synopsis

ff_sid_adm.pl –-op add –-pool <pool name> --sid <SAP system id> \

--db {ORACLE{9|10|11}|

MAXDB{75|76|77|78}|

DB2V{91|95|97}}:{<db_loghost_ip>|*}

--mdm mds:<nr>:{<client lan_ip>|\*}>:{<server

lan_ip>|\*} \

--mdm mdss:<nr>:{<client lan_ip>|\*}>:{<server

lan_ip>|\*} \

--mdm mdis:<nr>:{<client lan_ip>|\*}>:{<server

lan_ip>|\*} \

--sapversion 7.1 [--os <spec> ] \

[ --sapdata <nas_system>:/<volume path> ] \

[ --saplog <nas_system>:/<volume path> ] \

ff_sid_adm.pl –-op del –-pool <pool name> \

--sid <SAP system id>

--op add

Determines the add operation.

--op mod

Determines the mod operation. This option is only used to modify OS specifications

and exchanges IP adresses of specific SID instances.

--op del

Determines the delete operation.

--pool <pool name>

Specifies the FlexFrame pool to which the operation should be applied.

--sid <SAP system id>

Specifies the SID being used.

--sapversion {7.1}

Specifies the SAP basis version being used.

Adding/Removing SAP SIDs (addon services) SAP System Handling

Administration and Operation 261

--db {ORACLE{9|10|11}|MAXDB{75|76|77|78}|DB2V{91|95|97}}

Specifies the database type as well as the respective version being used. The

specification of the database type is not case-sensitive.

--group <groupname1>:<gidnumber1>,<groupname2>:<gidnumber2>,...

--user <username1>:<uidnumber1>,<username2>:<uidnumber2>,...

user and group enable specially selected user numbers and group numbers to be

assigned to SAP users and SAP groups respectively. In this case a check is made to

see whether the user or group has already been defined for the DB system involved.

A user or group is created only if they do not already exist. For example, a group

dba which already exists cannot be assigned a group number which deviates from

the default value.

{<db_loghost_ip>|*}}

The logical host name is used for the database (generated automatically: db<sid_-

se) as well as the IP address for that host name. Use an asterisk if you want it to be

chosen automatically. All the entries need to be specified in a colon separated for-

mat.

You can omit the network part of the IP, e.g. 10.10.12., and specify only the last

tupel of the IP, e.g. 123.

--mdm

{mds|mdss|mdis}:<SYSNR>:{<loghost_client_ip>|*}:{<loghost_server_i

p>|*}

Specifies an SAP instance (optionally multiple of those) through its type, its SAP

system number, the client network IP address, the the server network IP address.

Again, the IP addresses can be replaced with asterisks in order to have them cho-

sen automatically.

All the entries need to be specified in a colon separated format.

FlexFrame expects that the syntax of loghost-client follows the rule <type>sid and

loghost-server follows <type>sid-se.

--sysnr <SYSNR>

Removes a specific SAP instance instead of the entire system (SID).

SAP System Handling Adding/Removing SAP SIDs (addon services)

262 Administration and Operation

--os <instance_type>:<os>,<instance_type>:<os>,...]

Specifies the OS type for the given SID and/or their instances.

instance_type::= {default|mds|mdss|mdis}

os::= {SLES-10.x86_64|SLES-11.x86_64}

In combination with the –-add option, default: <os> sets the operating system of the

SID itself and all their instances which no own operating system is assigned to.

In combination with the –-mod option, default:<os> sets the operating system of the

SID only. The specifications of their instances are not changed.

--sapdata <nas system>:/vol/<volume path>

Allows to request an implicit call of ff_sid_mnt_adm.pl to relocate location to store

sapdata. It is expected that the requested volume already exists. The option is valid

only with services using a database.

--saplog <nas system>:/vol/<volume path>

Allows to request an implicit call of ff_sid_mnt_adm.pl to relocate location to store

saplog. It is expected that the requested volume already exists. The option is valid

only with services using a database.

The consistence of the given instance type/OS combinations is not checked

by FlexFrame. For a list of allowed combinations see SAP note 1067221.

--ips <old_ip>:<new_ip>,<old_ip>:<new_ip>,...]

Allows to exchange the IP addresses of specific SID instances. This option is only

used with —-op mod. You have to specify the full IP you want to exchange.

ff_sid_adm.pl searches for the specific instances within a SID and exchanges all

corresponding entries in LDAP concerned by that request.

Please pay attention that you make critical changes within your configuration. So we

strongly recommend you to take a backup of LDAP database before changing IP

addresses.

Adding/Removing SAP SIDs (addon services) SAP System Handling

Administration and Operation 263

Examples

Adding a SID with MDM services:

control1:~ # ff_sid_adm.pl –-op add –-sid mdm –-pool Otto --sapversion

7.1 \

--db MaxDB76:170 \

––mdm mds:01:171:171

––mdm mdss:02:172:172

––mdm mdis:03:173:173

Adding an instance of type mdss to an existing MDM SID:

control1:~ # ff_sid_adm.pl –-op add –-sid mdm –-pool Otto --sapversion

7.1 \

––mdm mdss:04:174:174

Removing an specific instance of a MDM SID:

control1:~ # ff_sid_adm.pl –-op rem –-sid mdm –-pool Otto –sysnr 02

Removing a SID with MDM services:

control1:~ # ff_sid_adm.pl –-op rem –-sid mdm –-pool Otto

10.5.4 SMD – Solution Manager Diagnostics

Before SAP 7.1 – EHP 1 a Solution Manager Diagnostics Agent specification was

optional. From SAP 7.1 – EHP 1 up services like 'ci', 'app' etc. request a SMD service SID

to continue installation process.

You do not need to create a SMD SID for each SAP system SID you want to install. You

can specify a general SMD SID ans specify SMDs for different SAP services, even from

different SAP service SIDs. You have to take into account that each SMD instance needs

a unique instance number. This may lead to a flood of instance numbers within your

specific pool.

Synopsis

ff_sid_adm.pl –-op add –-pool <pool name> --sid <SAP system id> \

--smd <nr>:<client lan_hostname –

monitored host> \

--sapversion 7.1 [--os <spec> ]

ff_sid_adm.pl –-op del –-pool <pool name> --sid <SAP system id>

SAP System Handling Adding/Removing SAP SIDs (addon services)

264 Administration and Operation

--op add

Determines the add operation.

--op del

Determines the delete operation.

--pool <pool name>

Specifies the FlexFrame pool to which the operation should be applied.

--sid <SAP system id>

Specifies the SID being used.

--sapversion {7.1}

Specifies the SAP basis version being used.

--group <groupname1>:<gidnumber1>,<groupname2>:<gidnumber2>,...

--user <username1>:<uidnumber1>,<username2>:<uidnumber2>,...

user and group enable specially selected user numbers and group numbers to be

assigned to SAP users and SAP groups respectively. In this case a check is made to

see whether the user or group has already been defined for the DB system involved.

A user or group is created only if they do not already exist. For example, a group

dba which already exists cannot be assigned a group number which deviates from

the default value.

--smd <SYSNR>:<client lan name – monitored host>

Specifies the name of the host which should be monitored. The name of the host

depends on the SAP service type

● app – app<nr><sid>

● j – j<nr><sid>

● ci – ci<sid>

● jc – jc<sid> (less than SAP 7.1), j<nr><sid> (SAP > 7.1)

--sysnr <SYSNR>

Removes a specific SAP instance instead of the entire system (SID).

--os <instance_type>:<os>,<instance_type>:<os>,...]

Specifies the OS type for the given SID and/or their instances.

instance_type::= {default|smd}

Adding/Removing SAP SIDs (addon services) SAP System Handling

Administration and Operation 265

os::= {SLES-10.x86_64|SLES-11.x86_64}

In combination with the –-add option, default:<os> sets the operating system of the

SID itself and all their instances which no own operating system is assigned to.

In combination with the –-mod option, default:<os> sets the operating system of the

SID only. The specifications of their instances are not changed.

The consistence of the given instance type/OS combinations is not checked

by FlexFrame. For a list of allowed combinations see SAP note 1067221.

Examples

Adding a SID with SMD services:

control1:~ # ff_sid_adm.pl –-op add –-sid SMD –-pool Otto --sapversion

7.1 \

––smd 01:j12abc

––smd 02:cixyz

Adding an instance of type mdss to an existing SMD SID:

control1:~ # ff_sid_adm.pl –-op add –-sid SMD –-pool Otto --sapversion

7.1 \

––smd 04:app12xyz

Removing an specific instance of a SMD SID:

control1:~ # ff_sid_adm.pl –-op rem –-sid mdm –-pool Otto –sysnr 02

Removing a SID with SMD services:

control1:~ # ff_sid_adm.pl –-op rem –-sid mdm –-pool Otto

10.5.5 TREX (Search and Classification Service)

Synopsis

ff_sid_adm.pl –-op add –-pool <pool name> --sid <SAP system id> \

--trx <nr>:<client lan_ip>>:<server

lan_ip> \

--sapversion 7.1 [--os <spec> ]

ff_sid_adm.pl –-op del –-pool <pool_name> --sid <SAP_system_id>

SAP System Handling Adding/Removing SAP SIDs (addon services)

266 Administration and Operation

--op add

Determines the add operation.

--op del

Determines the delete operation.

--op mod

Determines the mod operation. This option is only used to modify OS specifications

and exchanges IP adresses of specific SID instances.

--pool <pool name>

Specifies the FlexFrame pool to which the operation should be applied.

--sid <SAP system id>

Specifies the SID being used.

--sapversion {7.1}

Specifies the SAP basis version being used.

--group <groupname1>:<gidnumber1>,<groupname2>:<gidnumber2>,...

--user <username1>:<uidnumber1>,<username2>:<uidnumber2>,...

user and group enable specially selected user numbers and group numbers to be

assigned to SAP users and SAP groups respectively. In this case a check is made to

see whether the user or group has already been defined for the DB system involved.

A user or group is created only if they do not already exist. For example, a group

dba which already exists cannot be assigned a group number which deviates from

the default value.

--smd <SYSNR>:<client lan hostname – monitored host>>

The logical host name is used for the database (generated automatically:

trx<nr><sid>, trx<nr><sid-se>) as well as the IP address for that host

name. Use an asterisk if you want it to be chosen automatically. All the entries need

to be specified in a colon separated format.

You can omit the network part of the IP, e.g. 10.10.12. and specify only the last

tupel of the IP, e.g. 123.

--sysnr <SYSNR>

Removes a specific SAP instance instead of the entire system (SID).

Adding/Removing SAP SIDs (addon services) SAP System Handling

Administration and Operation 267

--os <instance_type>:<os>,<instance_type>:<os>,...]

Specifies the OS type for the given SID and/or their instances.

instance_type::= {default|trx}

os::= {SLES-10.x86_64|SLES-11.x86_64}

In combination with the –-add option, default:<os> sets the operating system of the

SID itself and all their instances which no own operating system is assigned to.

In combination with the –-mod option, default:<os> sets the operating system of the

SID only. The specifications of their instances are not changed.

The consistence of the given instance type/OS combinations is not checked

by FlexFrame. For a list of allowed combinations see SAP note 1067221.

--ips <old_ip>:<new_ip>,<old_ip>:<new_ip>,...]

Allows to exchange the IP addresses of specific SID instances. This option is only

used with —-op mod. You have to specify the full IP you want to exchange.

ff_sid_adm.pl searches for the specific instances within a SID and exchanges all

corresponding entries in LDAP concerned by that request.

Please pay attention that you make critical changes within your configuration. So we

strongly recommend you to take a backup of LDAP database before changing IP

addresses.

Examples

Adding a SID with TRX services:

control1:~ # ff_sid_adm.pl –-op add –-sid TR1 –-pool Otto --sapversion

7.1 \

––trx 01:201:201

––trx 02:200:200

Adding an instance of type mdss to an existing TRX SID:

control1:~ # ff_sid_adm.pl –-op add –-sid TR1 –-pool Otto --sapversion

7.1 \

––smd 04:202:202

Removing an specific instance of a TRX SID:

control1:~ # ff_sid_adm.pl –-op rem –-sid tr1 –-pool Otto –sysnr 02

Removing a SID with TRX services:

control1:~ # ff_sid_adm.pl –-op rem –-sid tr1 –-pool Otto

SAP System Handling Cloning a SAP SID into a Different Pool

268 Administration and Operation

10.6 Cloning a SAP SID into a Different Pool

10.6.1 Script: ff_clone_sid.pl

The script ff_clone_sid.pl allows users to clone (basically copy) an entire SID from

one pool to another. It needs to be clearly understood that only the FlexFrame-specific

administrational data in the LDAP server as well as the required information for the

operating system's naming services are copied and/or added to the LDAP database. Any

additional work (like copying SAP/database binaries and database content) will not be

performed by this tool and needs to be performed, as well.

To get a real clone the script has to expect that there are no conflicts in userids, groupids,

services and other properties between the origin and the target pool. You can check it by

using option ‗—dryrun‘.

If there are conflicts and you are sure it does not matter you can clone the SID and the

script will select other values than defined in origin. Otherwise the cloning could mean

inconsistencies in your LDAP database. Be careful !

Synopsis

ff_clone_sid.pl –-sid <sid>

--srcpool <pool name>

--trgtpool <pool name>

[--dryrun ]

[--renew ]

Options

--sid <sid>

Specifies the SID which should be cloned.

-srcpool <pool name>

Specifies the pool with the origin SID specification.

--trgtpool <pool name>

Specifies the pool wherein the clone should be created.

[ --dryrun ]

Shows you what happens if you request a clone. You can use the option to check if there

are inconsistencies between the source and the target pool.

[ --help ]

Shows the usage.

Cloning a SAP system into a different Pool SAP System Handling

Administration and Operation 269

[ --renew ]

You can enforce cloning if you have only conflicts with IP addresses between source and

target. If you are sure that it does not matter you can use this option and the script selects

new IP addresses independent from the source.

10.6.2 Changing User and Group IDs after Cloning

The ff_change_id.pl script is not further available.

To change the UID of an OS user you can use ff_user_adm.pl.

To change the GID of an OS group you can use ff_group_adm.pl.

10.7 Cloning a SAP system into a different Pool

10.7.1 Script: ff_clone_sap.sh

The script ff_clone_sap.sh allows users to clone (basically copy) an entire SAP

installation from one pool to another. It means also the binaries are copied to the target

pool and so you get a complete independent 'copy' of your source system. The command

tries to do all configuration tasks needed to run the SAP system in the target pool.

You can clone a SAP system also if the system is active and the database is online. But

we recommend to clone in offline state if possible.

At the current release the command only supports SAP services with

configured databases, Services like BOBJ, SMD or TRX are not

supported.

The script cannot check if your clone request overwrites newer versions of

files already copied earlier with older file versions. You are responsible for

that issue. Use cloning with care !

With ff_clone_sap.sh you copying a complete SAP system. You are

responsible for license issues using SAP software.

SAP System Handling Cloning a SAP system into a different Pool

270 Administration and Operation

The script copies the SAP system files (which each clone request)

located in /usr/sap/<SID>

located in /sapmnt/<SID>

home directories of SAP specific users like <sid>adm

/usr/sap/sapservices

Database files (software, configuaration) depending on database type

Some files are only copied if not available in the target configuration or on force request

/usr/sap/hostctrl

/usr/sap/tmp

/oracle/client

/sapdb/data

/sapdb/programs

/sapdb/sql

There are some preconditions you should take into account if you want to clone SAP

systems

the environment used (users, groups, etc.) has to be equal in source and target

pool. Otherwise you may get problems accessing files and directories. It is an

essential precondition.

do not clone SIDs from different source pools to one target pool. We

recommend to create always an own target pool to each pool from which you

want to clone SIDs.

only SAP systems residing on NetApp NAS storage are supported.

Be careful when cloning with different SAP versions. The script cannot prevent

overwriting files with older revisions.

It is necessary to store the files from sapdata/saplog in an own volume to

optimize disk storage.

Cloning a SAP system into a different Pool SAP System Handling

Administration and Operation 271

Newer SAP release require an SMD agent per default. Do not forget to clone

also those installation (if SMD is supported in later). There is no automatic clone

of depending installations.

If you want to unload volFF we recommend to relocate SID specific directories in

the target pool using relocation feature.

At start of the clone process it is not known how much storage is necessary for

the running clone. You are responsible to supply enough storage. In the worst

case the clone needs storage resources as much as the origin.

Synopsis

ff_clone_sap.sh [ --op | -o ] { create | destroy | crclone |

rmclone | initpool | check }

--srcpool <pool name>

--trgtpool <pool name>

-- sid | -s <sid>

--flexdata <sapdata clone volume >

--flexlog <saplog clone volume>

[ --reloc <filer:volume> ]

[ --dbhost <virtual DB hostname> ]

[ --config <user defined config file> ]

[ --prefix <prefix used for snapshot name> ]

[ -- fcopy ]

[ --help | - h ]

[ --version | -v ]

[ --debug | -d ]

[ --dryrun ]

Options

--op initpool

After introducing the target pool by ff_pool_adm.pl use this opcode to create global users

and groups to get a consistent environment between source and target. This step has to

be done before creating first clone.

--op create

SAP System Handling Cloning a SAP system into a different Pool

272 Administration and Operation

Creates a complete clone of your SAP system (cloning of database files, binaries and

global files if necessary)

--op destroy

Deletes the clone completely. Only global files are not deleted.

--op check

Runs a check to examine your environment. It is useful to recognize violations of most of

the preconditions

--op crclone

Creates only a clone of your database

--op rmclone

Removes the database clone only

--srcpool <pool name>

Name of the pool which contains the SID to be cloned..

--trgtpool <pool name>

Name of the pool the SID should be cloned into.

--sid <sid>

Specifies the SID to be cloned.

--flexdata <filer name>:<volume name>

Name of the clone volume for sapdata. This volume is created through FlexClone feature,

not by the caller of the script.

The name of the filer is the filer‘s name as used in first pool. It does not contain the name

of the target pool.

--flexlog <filer name>:<volume name>

Name of the clone volume for saplog. This volume is created through FlexClone feature,

not by the caller of the script.

The name of the filer is the filer‘s name as used in first pool. It does not contain the name

of the target pool.

--reloc <filer name>:<volume name>

Name of specific volume in target pool to store SID specific data (unload volFF). You are

responsible for the creation and availability of the volume. You can use the command

ff_relocate_sid_data.pl to modify filer exports and create a script to create the necessary

directories on your volume.

--dbhost <dbhost name>

Cloning a SAP system into a different Pool SAP System Handling

Administration and Operation 273

Contains the virtual name of the database host (db<sid>-se). It is used only if the clone

should be taken from an inactive system (database offline).

--fcopy

Enforce copy of global files. If the directory already exists in target configuration it is

overwritten (use it with care !).

--prefix <name>

Name of the prefix used for naming snapshot. Default is FF_DATA resp. FF_LOG

--help

Shows the usage

--debug

Switch on debug mode

--dryrun

Starts a cloning without changing anything. You will see what will happen if you are

requesting a real clone.

--version

Shows the script version.

Return Codes

0 means all o.k.

9 at least one warning, but the clone request succeeded

1 abortion of the request because of errors

10.7.1.1 Cloning SAP systems with MaxDB

Some files of MaxDB database configuration are stored in /sapdb/data/config or

/sapdb/FLM/data/config. To ensure consistency you have to move those files to

/sapdata/sapdata<n>/config and create a link at origin site before you start the cloning

operation

<SID>.<nn>

<SID>.cfg

<SID>.pah

<SID>.upc>

SAP System Handling Cloning a SAP system into a different Pool

274 Administration and Operation

<SID>

.M*

10.7.1.2 Post cloning steps – DB2

After cloning a SAP system based on DB2 database you have to execute

exchange hostname in /home_sap/db2<sid>/sqllib/db2nodes.cfg with the name

of the application node you want to start the cloned system first

/home_sap/db2<sid>/db2_software/instance/db2iupdt db2<sid> under control

of user ‗root‘

start the database using ff_service.sh

login as db2<sid> and execute ‗db2inidb <SID> as snapshot‘ if there are

connection problems (error message SQL20153 by executing ‗db2 connect‘)

For further informations please refer to ‗DB2: Cloning a Database Using NetApp

FlexClone Technology‘ by NetApp

10.7.1.3 Post cloning steps – Oracle

If your clone is based on an online system you have to execute a ‗recover database‘

using ‗sqlplus‘ under control of user ‗ora<sid>‘.

10.7.1.4 Post cloning steps – MaxDB

If you are cloning a system with MaxDB less than 7.8 the configuration files from

/sapdb/data/config (7.8: /sapdb/<SID>/data/config) are copied at the time the first clone

operations took place. All further cloning operations for MaxDB do not change

/sapdb/data/config and the configuration files are possibly not up to date. This may cause

problems while starting database.

After cloning the SAP system ensure the consistency of the configuration files by

executing

CN1: cd /FlexFrame/volFF/pool-<target>/sapdb/<platform>/data/config

CN1: rm <SID>.???

CN1: cd /FlexFrame/volFF/pool-<source>/sapdb/<platform>/data/config

CN1: find . | cpio –dump \

CN1: /FlexFrame/volFF/pool-<target>/sapdb/<platform>/data/config

Cloning a SAP system into a different Pool SAP System Handling

Administration and Operation 275

10.7.1.5 Cloning a SAP system – example session

We guess you have a pool ‗source‘ with installed SAP services using SID ‗ORA‘. You

want to clone that installation to a pool ‗target‘. We further guess that your sapdata/saplog

are stored on a SID specific volume (command ff_sid_mnt_adm.pl used before executing

ff_setup_sid_folders.sh). Otherwise your clone will include other databases which

belongs to SIDs you do not want to clone. This is just a matter of needed disk space.

At first you have to create your target pool by executing

Control1: ff_pool_adm.pl –op add –name target –storage 16,172.20.16.0,255.255.255.0

–server 17,172.20.17.0,255.255.255.0 –client 15,172.20.15.0,255.255.255.0 –dns

ff4s.mch.fts.net –defrouter 172.20.18.251

The used IPs are just examples, you insert your own configuration values. If you want to

specify a pool specific volFF volume you have to extent the command by ‗—volFF

<filer>,/vol/<volume name>‘. Before using the volume within this command you have to

create it manually by NetApp filer commands or ask your storage administrator.

In the next step you initialize your target pool to make sure that UIDs and GIDs are

in consistent state with the source pool. Please execute

Control1: ff_clone_sap.sh –op initpool –srcpool source –target target

Now you can clone your SAP system. Please take into account that SAP systems with

MaxDB needs to move configuration files from /sapdb/data/config or

/sapdb/FLM/data/config to a sapdata directory to ensure consistencies (for requested files

see above).

We guess our filer is named ‗jer1na2‘ and the volume names for sapdata/saplog are

‗/vol/sid_ora_data‘/‘/vol/sid_ora_log‘.

You don not need to create the volumes for the clone, you just give it names. Those

volumes are created by the FlexClone feature of NetApp. You are only responsible to

supply enough disk space for clone operation and the later operation of the clone in

target pool.

To clone your system (SAP services are down; database offline) you execute

Control1: ff_clone_sap.sh –op create –srcpool source –trgtpool target \

--sid ORA \

--flexdata jer1na2-st:/vol/sid_ora_data \

--flexlog jer1na2-st:/vol/sid_ora_log \

--dbhost dbora-se

Some directories/files are copied one time (e.g. if you have MaxDB 7.7 database

software and it is valid for all cloned systems, /sapdb/programs needs to be copied only

with the first clone operation).

If you want to force the copy of those files you additionally use the option ‗—fcopy‘. With

SAP System Handling Relocating sapdata/saplog by ff_sid_mnt_adm.pl

276 Administration and Operation

older versions of MaxDB software you should use ‗—fcopy‘ for each clone because of

global directory ‗/sapdb/data‘

If you need additional diagnostic support you can use the option ‗—debug‘.

Please check the filer configuration by executing

Control1: ff_nas_adm.pl –op list –nas jer1na2

If you cannot find ‗jer1na2-target-st‘ in output of the command you need to execute

Control1: ff_nas_adm.pl –op list –nas jer1na2 --pool target

Now you should be ready to operate your clone in target pool. Because of changes within

the automount configuration you should execute

applnode: /etc/init.d/autofs restart

Please check also for necessary manual operations depending on database type

described above.

10.8 Relocating sapdata/saplog by ff_sid_mnt_adm.pl

To unload volumes containing database files you can use ff_sid_mnt_adm.pl. See also

descriptions made in chapter ‗Multiple NAS Systems and Multiple Volumes‘.

Synopsis

ff_sid_mnt_adm.pl --op { add | del | rem | list }

--sid <sid>

--pool <pool name>

[ --sapdata

<filer>:<volume>/<pool name>/<SID> ]

[ --saplog

<filer>:<volume>/<pool name>/<SID> ]

[ --help ]

Options

--op add

Adds new automount entries into LDAP to relocate sapdata and/or saplog.

--op { del | rem }

Relocating sapdata/saplog by ff_sid_mnt_adm.pl SAP System Handling

Administration and Operation 277

Deletes the relocation entries for sapdata and/or saplog

--op list

Displays information about relocation of sapdata and/or saplog for specific SID.

--sid <sid>

Specifies the SID for which the relocation entries should be handled. <SID> has to be

always written in uppercase characters.

--pool <pool namel>

Specifies the corresponding pool which contains the SID specification.

--sapdata <filer>:<volume path>/<pool>/<SID>

Specifies the complete path on the volume to relocate sapdata. <SID> has to be always

written in uppercase characters. For the path specification ‗>:<volume path>:/<pool>‘

there is an implicit modification of filer‘s export file in case of opcode ‗add‘. If deleting the

entry filer‘s export is not modified implicitely.

--saplog <filer>:<volume>/<pool>/<SID>

Specifies the complete path on the volume to relocate saplog. <SID> has to be always

written in uppercase characters. For the path specification ‗>:<volume path>:/<pool>‘

there is an implicit modification of filer‘s export file in case of opcode ‗add‘. If deleting the

entry filer‘s export is not modified implicitely.

--help

Shows the usage.

Hints:

With FlexFrame 5.1 the filer exports are modified internally by ff_sid_mnt_adm.pl with

opcode ‗add‘. This feature is only available for NetApp.

If you are removing relocated data the exports have to be modified manually by

ff_exports.pl. If the volume should be deleted no action is necessary.

Administration and Operation 279

10.9 Multiple NAS Systems and Multiple Volumes

During the installation process, FlexFrame assumes that there is one NAS system with

sapdata and saplog volumes. Larger installations may require more than one NAS

system or multiple volumes on the same NAS system.

It is possible to distribute SAP databases across multiple NAS systems and multiple

volumes under the following conditions:

1. All NAS systems were entered in the FlexFrame Management Tool, prior to

installation of the FlexFrame landscape.

2. The software for SAP and the database are located in the centralized volume volFF

of the first NAS system or alternatively (if wanted) in a pool-specific volFF volume in

an arbitrary NAS system (defined within the FlexFrame environment).

3. For each SID you can assign a sapdata volume for the database's data files. This

sapdata volume can be shared with other SIDs or solely for this SID. But if you

want to share multiple SIDs on a volume we recommend you to use pool-specific

volumes for sapdata. It will minimize your efforts for administration tasks.

4. For each SID you can assign a saplog volume for the database's online redolog

files. This saplog volume can be shared with other SIDs or solely for this SID. But if

you want to share multiple SIDs on a volume we recommend you to use pool-specific volumes for saplog. It will minimize your efforts for administration tasks.

SAP System Handling Multiple NAS Systems and Multiple Volumes

280 Administration and Operation

10.9.1 NetApp Filer

The volumes must be created manually on the Filer, e.g. like:

filer2> vol create dataC11 10

filer2> vol create logC11 4

Here, dataC11 and logC11 are the new names of the volumes and 4 and 10 are the

numbers of disks to be used.

We recommend using the volume names sapdata and saplog (if on a different

Filer) or data<SID> and log<SID> for SID specific volumes on the same Filer.

You may use FlexVols (ONTAP 7G) or regular volumes.

The following options must be set for the volume:

filer2> vol options dataC11 nosnap on

filer2> vol options dataC11 nosnapdir on

filer2> vol options dataC11 minra off

filer2> vol options dataC11 no_atime_update on

filer2> vol options logC11 nosnap on

filer2> vol options logC11 nosnapdir on

filer2> vol options logC11 minra off

filer2> vol options logC11 no_atime_update on

Next, create qtrees for each FlexFrame pool which will store data and logs in those

volumes:

filer2> qtree create /vol/dataC11/pool1

filer2> qtree create /vol/logC11/pool1

If you use more than the first Filer, make sure it is reachable using, e.g.:

control1:~ # ping filer2-st

PING filer2-st (192.168.10.203) from 192.168.10.201 : 56(84) bytes

ofdata.

64 bytes from filer2-st (192.168.10.203): icmp_seq=1 ttl=255 time=0.117

ms

64 bytes from filer2-st (192.168.10.203): icmp_seq=2 ttl=255 time=0.107

ms

64 bytes from filer2-st (192.168.10.203): icmp_seq=3 ttl=255 time=0.103

ms

Multiple NAS Systems and Multiple Volumes SAP System Handling

Administration and Operation 281

Now the LDAP database has to be told that this SID is not using the default (first) Filer

and sapdata/saplog volumes:

control1:~ # ff_sid_mnt_adm.pl –-op add –-pool pool1 –-sid C11

--sapdata filer2-st:/vol/dataC11/pool1/C11 –-saplog filer2-

st:/vol/logC11/pool1/C11

The step to modify filer‘s export file is not necessary with FlexFrame 5.1 up because this

step in integrated in command ff_sid_mnt_adm.pl. This feature is only available with

NetApp.

The network 192.168.10.0/24 must match the Storage LAN segment of pool pool1.

Before you can install SAP and the database on those volumes, some folders for the SID

in question have to be created in advance. To do so, run the following command for each

SID (replace "pool1" with your pool name and "C11" with your SID:

control1:~ # ff_setup_sid_folder.sh –p pool1 –s C11

Now you can continue with the SAP installation.

10.9.2 EMC Celerra

Regard a Celerra with at least two active data movers: What is to be done, if volFF is on

one data mover, and the SAP volumes should be on the other ?

Define the volumes as described in the Installation Guide, but you have to export the

corresponding file systems on the data mover 2 (=server_3), e.g.:

server_export server_3.../dataC11

server_export server_3.../logC11

For the data mover, which is to access the SAP volumes, you will have to define the

storage VLANs of all FlexFrame pools, that should have access to these volumes. In the example it is pool1 and VLAN-ID 11.

Now the LDAP database has to be told that this SID is not using the default NAS and

sapdata/saplog volumes:

control1:~ # ff_sid_mnt_adm.pl –-op add –-pool pool1 –-sid C11

--sapdata filer2:/vol/dataC11/pool1/C11 –-saplog

filer2:/vol/logC11/pool1/C11

Now the volumes on the Control Nodes need to be mounted. To do so you should add the following lines to each Control Node's /etc/fstab:

SAP System Handling Upgrading a SAP System

282 Administration and Operation

<Datamover2-st>:/vol/dataC11/pool1 /FlexFrame/<Datamover2-

st>/pool1/dataC11

nfs nfsvers=3,rw,bg,udp,soft,nolock,wsize=32768,rsize=32768

<Datamover2-st>:/vol/logC11/pool1 /FlexFrame/<Datamover2-

st>/pool1/logC11

nfs nfsvers=3,rw,bg,udp,soft,nolock,wsize=32768,rsize=32768

Repeat the sapdata and saplog lines for each pool, if there's more than one

pool for those volumes.

Use the volume name for the last directory in the mount point.

Now we need the mount points:

control1:~ # mkdir -p /FlexFrame/<Datamover2-st>/pool1/dataC11

control1:~ # mkdir -p /FlexFrame/<Datamover2-st>/pool1/logC11

(sapdata and saplog for each pool again)

Now we can mount the volumes:

control1:~ # mount -a

Before you can install SAP and the database on those volumes, some folders for the SID

in question have to be created in advance. To do so, run the following command for each

SID (replace pool1 with your pool name and C11 with your SID:

control1:~ # ff_setup_sid_folder.sh –p pool1 –s C11

If you have several Celerras, the FlexFrame administrator will have to configure the time

service and the SSH keys for each of them (see Installation Guide).

Now you can continue with the SAP installation.

10.10 Upgrading a SAP System

10.10.1 Service Port

If you plan to upgrade your SAP release, you have to add a special service port (called

shadow instance) to LDAP.

SAP shadow service ports are required during an SAP release upgrade. To list, add or

remove the service ports in the LDAP database for service entries, you can use this

tool.

Upgrading a SAP System SAP System Handling

Administration and Operation 283

Synopsis

ff_sap_shadowport.sh [-d] -l -p <pool name> [-s <sid>]

[-o <portnr>]

ff_sap_shadowport.sh [-d] {-a|-r} -p <pool name> -s <sid>

[-o <portnr>]

Options

-d

Writes debugging information.

-l

Lists all SAP shadow service ports of the pool provided with the -p option.

If the option -s is used, only the port of the specified SID is displayed.

-a

Adds an entry.

-r

Removes an entry.

-p <pool name>

Specifies the name of the pool (e.g. pool1).

-s <sid>

Specifies the SAP System ID (SID) by a 3 character string (e.g. C11).

-o <portnr>

Specifies the service port number. The default is 3694 (optional).

10.10.2 FA Agents

Please make sure that the FA Application Agents are stopped on the hosts while you are

installing, updating or removing any SAP or database software:

Stop the FA Agent and check the status:

control1:~ # /etc/init.d/myAMC.FA_AppAgent stop

control1:~ # /etc/init.d/myAMC.FA_AppAgent status

There should be no running processes listed.

SAP System Handling Upgrading a SAP System

284 Administration and Operation

10.10.3 Support SAP Upgrade

This command is used to support the migration from a previous SAP base version to a

higher one. It could be also useful if you are migrating from an older release of FF to a

newer one.

Upgrading a SAP System SAP System Handling

Administration and Operation 285

You are able to change the SID specific settings in LDAP database like version of SAP or

database installation. It also tries to introduce the LDAP data which are requested with

newer SAP versions. The script tries to cover all requirements concerning version of SAP

or database service.

Sometimes users want to migrate from one database type to an other. The script cannot

execute all actions requested by such a migration. Manual actions could be required

Synopsis

ff_sap_upgrade.pl --pool <pool name> --sid {<SID>|\*}

--sapvo <old version> --sapvn <new version>

--dbvo <old version> --dbvn <new version>

--type <service type>

[--version] [--debug] [--dryrun]

ff_sap_upgrade.pl [--help]

Options

--pool <pool name>

Name of the pool the SID should belong to.

--sid <SID>

SAP system identifier. Either you specify a specific SID or you use the asterisk.

Asterisk means all SIDs of the specific pool which match to sapvo and dbvo are

modified.

--sapvo <old version>

Current value of the SAP version stored in LDAP (without leading SAP-).

<old version> can be one of 4.6, 6.20, 6.40, 7.0, 7.1 or 7.2. The type

‗BOBJ‘ allows ‗3.1‘ or ‗3.2‘

--sapvn <new version>

SAP version the SID is migrated to. <new_version> is 7.0, 7.1, 7.2 or 7.3.

Type ‗BOBJ‘ allows ‗3.2‘ or ‗4.0‘

--dbvo <old version>

Version of previous used database version. <old version> can be one of

ORACLE9, ORACLE10, SAPDB73, SAPDB73, SAPDB74, MAXDB75, MAXDB76,

MAXDB77, DB2V91 or DB2V95.

SAP System Handling SAP Kernel Updates and Patches

286 Administration and Operation

--dbvn <new version>

Version of database the SID is migrated to. <new version> can be one of ORACLE9, ORACLE10, ORACLE11,MAXDB76,

MAXDB77, MAXDB78, DB2V95, or DB2V97.

--type <service type>

Specifies the SAP service. Default value is ‗SAP‘, other values are BOBJ, CMS,

MDM, SMD and TREX. Except ‗SAP‘, all services are new. This is the reason why

option ‗type‘ has been introduced.

--version

Show version of command.

--debug

Sets the debug option to enlarge logging.

--dryrun

Just to see which changes would be made in LDAP.

--help

Display usage.

Tips and Tricks

It is not allowed to specify 7.1 with option sapvo. But in some installations it could be

necessary to make system changes (e.g. automount entries). So you can do that with a

little workaround:

ff_sid_adm.pl –op add –pool pool1 –sid ttt –sapversion 4.6

--db ORACLE9:dbttt-se:\* --sap ci:90:\*:\*

ff_sap_upgrade.pl –pool pool1 –sid ttt –sapvo 4.6 –sapvn 7.1

--dbvo ORACLE9 –dbvn ORACLE10

ff_sid_adm.pl –op del –pool pool1 –sid ttt

10.11 SAP Kernel Updates and Patches

For an SAP kernel update (binary patches), logon to the Application Node with the CI of

the SAP system that is to be updated. Please make sure that the FA Application Agents

are stopped on the host while you are updating the SAP kernel (see also section ―FA

Agents‖ on page 283).

SAP's OSS note 19466 describes where to find kernel patches and how to handle the

installation.

Unloading volFF SAP System Handling

Administration and Operation 287

For the installation of SAP ABAP patches or similar, please refer to the SAP

documentation. There are no FlexFrame specific changes.

10.12 Unloading volFF

10.12.1 Status Quo/Solution

All SID specific data are located on the same volume. In case of request of diagnostics it

could happen that the volume space is not sufficient or it is not easy to find the specific

data you need because of all SIDs writes into the same location.

With FlexFrame for SAP, Release 5.0A, it is possible to separate some directories with

SID specific data to a pool-specific volFF volume.

The SID specific data are stored in

oracle

home_sap/sidadm,home_sap/sqdsid or /home_sap/db2sid

sapmnt

usr_sap

sapdb

db2

To weaken that situation we offer

a way to move large SID specific data to its own storage location

a monitor function which logs the usage of volFF

10.12.2 ff_relocate_sid_data.pl

Sometimes you are requested to move SID specific data from volFF to a separated volume. ff_relocate_sid_data.pl is used to change the automount information in

LDAP to use separated volume for data access.

If you take a snapshot of a configuration and you want to use it for an installation

on another system, the changes made in LDAP automount entries with

ff_relocate_sid_data.pl are lost. You must repeat the commands you

have executed in the origin installation to recreate the LDAP data for relocation.

The new command does not support the tasks to configure a volume and to make the corresponding changes in system files needed (e.g. exports file).

SAP System Handling Unloading volFF

288 Administration and Operation

Synopsis

ff_relocate_sid_data.pl –-op add --pool <pool name> [ --sid <SID> ]

--name <type>:<nas system>:<volume_path>/<pool name>/<SID>

[--force] [--dryrun]

ff_relocate_sid_data.pl –-op { del rem } --pool <pool name> --sid

<SID> --name <type> ... [--dryrun]

ff_relocate_sid_data.pl –-op list --pool <pool name> --sid <SID>

ff_relocate_sid_data.pl –-help

Options

--op add

Determines the add operation.

--op { del | rem }

Determines the delete operation.

--pool <pool name>

Name of the pool the SID(s) belongs to.

--sid <SID>

SAP system identifier. If you omit this option informations about all SIDs within the

specified pool are displayed.

--name <type>:<nas system>:<volume path>/<pool name>/<SID>

Type, name of the NAS system and path to the new storage. This option could be

repeated several times within one call. Each type can only be used one time. Moving

data of SIDs from different pools need an own command for each SID and pool

which data are moved. The subdirectory on the new volume has to be created

manually if not already available.

For <type> you can ―usr_sap‖ or ―sapmnt‖ or ―oracle‖ or ―sapdb‖ or ―db2‖ or

―<sid>adm ― or ―sqd<sid>‖ or ―db2<sid>‖ or sap<sid> or sap<sid>db or ―oraarch‖

―oraarch‖ is a special case because it is part of /oracle (link

to/saplog/oraarch/<SID>).

You have to take this into account if you are copying the data from the origin storage

to the new location. The data on the origin location have to be deleted manually. A workaround is introduced by using /FlexFrame/scripts/sapdb. To mount the

relocated directory in a shell, use sapdb <SID> mount, and to umount the relocated

data, use sapdb <SID> umount. That workaround is not further needed and not

available since FlexFrame release 5.0.

Unloading volFF SAP System Handling

Administration and Operation 289

--force

The process continues even if errors occur at a specific name option. If force is not

set, the command rollback the previous changes made in LDAP until the error

occurs.

--dryrun

Displays which changes would be made in LDAP.

--help

Displays usage.

Hints

Moving sapmnt is a special case. If new SID instances are introduced by

ff_sid_adm.pl the mount points SID and SID_exe are already created.

ff_relocate_sid_data.pl renames these two nodes and generates new entries. If

you are delete sapmnt, the previous entries are restored.

The installation of a Celerra must be done by the EMC customer support. Please refer to

the Fujitsu Solution Facts document "Guidelines for the Installation of an EMC-Celerra

within FlexFrame", which can be downloaded from the public "FlexFrame for SAP

Softbooks" site http://manuals.ts.fujitsu.com/index.php?id=9215-9217-9263 (select

"FlexFrame Further Public Documents).

To generate a new volume (on NetApp) use

rsh <filer_name-st> vol create <volume_name> aggr0 5g

e.g.

rsh jer1na-st vol create z02_vol aggr0 5g

The next step is the insertion into the corresponding export file executing ff_exports.pl

ff_exports.pl –op add –nas <filer> --path <volume_name> -option ‗-

sec=sys,rw=<cn1 ip@>:<cn2_ip@>:<server_lan>/24,anon=0‘

e.g.

ff_exports.pl –op add –nas jer1na1-st –path /vol/z02_vol –option ‗-

sec=sys,rw=10.10.11.104:10.10.11.103:10.10.11.0/24,anon=0‘

If there is already an exports entry available for this volume read the current options using

‗ff_exports.pl –op list …‘ and extend the options with ‗:<server_lan>/24‘ if necessary.

SAP System Handling Unloading volFF

290 Administration and Operation

With FlexFrame 5.1 the filer exports are modified internally by ff_relocate_sid_data.pl with

opcode ‗add‘. This feature is only available for NetApp.

If you are removing relocated data the exports have to be modified manually by

ff_exports.pl. If the volume should be deleted no action is necessary.

Unloading volFF SAP System Handling

Administration and Operation 291

10.12.3 LDAP

A new entry in LDAP looks like this:

Administration and Operation 293

10.12.4 Move Data from volFF (how to)

10.12.4.1 Specify Volume Before Installing SAP

The following steps are necessary:

1. Define a new SID with

ff_sid_adm.pl

2. If you want to switch SAPDATA/SAPLOG to a specific volume, use

ff_sid_mnt_adm.pl --op add --pool <pool name> --sid <SID>

--sapdata <nas system>:/vol/<data volume>/<pool

name>/<SID>

--saplog <nas system>:/vol/<log volume>/<pool

name>/<SID>

3. Create directories with

ff_setup_sid_folder.sh –p <pool name> -s <SID>

4. Prepare the moving of specific data with

ff_relocate_sid_data.pl --op add --pool <pool name> --sid <SID>

--name <type>:<nas_

ystem>:/vol/new_volume/<pool name>/<SID> ...

5. Create directories (e.g. usr_sap) on the new volume with

mkdir /tmp/mnt

mount <nas system>:/vol/new_volume /tmp/mnt

cd /tmp/mnt

mkdir –p /tmp/mnt/<SID>/usr_sap

6. Move the data to the new volume

cd /FlexFrame/volFF/pool-<pool name>/sap/usr_sap/<SID>

mv * /tmp/mnt/<SID>/usr_sap

The last two steps are integrated into a script generated with ff_relocate_sid_data.pl (ff_relocate_sid_data.pl.<pool>.<KHE>.sh.

SAP System Handling Unloading volFF

294 Administration and Operation

10.12.4.2 Moving an Existing /usr/sap/SID

The following steps are necessary (example uses usr_sap):

1. Use nowatch to remove corresponding SAP instances from the monitoring of FA

Agents.

2. Stop all corresponding SAP instances.

3. Take a backup of the origin directory.

4. Rename /usr/sap/<SID> with

mv /FlexFrame/volFF/pool-<pool name>/sap/usr_sap/<SID>

/FlexFrame/volFF/pool-<pool name>/sap/usr_sap/<SID>_bak

5. Prepare the moving of specific data with

ff_relocate_sid_data.pl --op add --pool <pool name> --sid <SID>

--name <type>:<nas system>:/vol/new_volume/<pool name><SID> ...

6. Move original directories from /usr/sap/<sid> to volume specific directories (on any

application node)

mkdir /tmp/mnt

mount <nas system>:/vol/new_volume /tmp/mnt

cd /tmp/mnt

mkdir –p <sid>/usr_sap

cd /FlexFrame/volFF/pool-<pool name>/sap/usr_sap/<SID>_bak

mv * /tmp/mnt/<SID>/usr_sap 7. After moving a directory we recommend to restart automounter on the concerned

Application Nodes with /etc/init.d/autofs restart.

Steps 5 and 6 are integrated into a script generated with ff_relocate_sid_data.pl

(ff_relocate_sid_data.pl.<pool>.<sid>.sh. You must change the script

because of rename of <sid> to <sid>_bak of the source directory.

10.12.5 Delete Entries from LDAP

If you delete a SID you must also delete the specific automount information of the SID:

ff_relocate_sid_data.pl --op del --pool <pool name> --sid <SID> --name

<type>

Maybe the information on movement of sapdata/saplog have to be deleted, too:

ff_sid_mnt_adm.pl --op del --pool <pool name> --sid <SID>

Administrating SAP Services SAP System Handling

Administration and Operation 295

10.13 Administrating SAP Services

In a FlexFrame environment, any type of SAP instance, i.e. database (DB), SAP Central

Services (SCS), Java central instance (JC), Central Instance (CI), application instance

(APP) or Java-only stack (J), is called an SAP service. The management can be done

either web based with the FA WebGUI or script based with the adapted start/stop scripts.

If the SAP ACC is running, SAP services can be managed by ACC.

10.13.1 Displaying Status of SAP Services

10.13.1.1 myAMC.FA WebGUI

The myAMC.FA WebGUI can be used to display the states of SAP services in a

FlexFrame environment. To see all active SAP services and nodes in the FA WebGUI,

the FA Application Agent should be started on all Application Nodes. The displayed active

SAP services can be shown system-related or node-related.

The following table shows the node and service states of FlexFrame Autonomy and the

corresponding colors of these states in the FA WebGUI:

SAP System Handling Administrating SAP Services

296 Administration and Operation

white

inactive or no

further

information

green

normal,

everything ok

yellow

warning

red

critical

black

critical

Node states

RUNNING

SWITCH_INT

SWITCH_EXT

PowerOff

Service states

SHUTDOWN

DOWN

WATCH

NOWATCH

UNKNOWN

NULL

RUNNING

STOPPING

STARTING

REBOOT

RESTART

RESTARTING

REBOOT

REBOOTING

RBGET

SWITCH

SWITCHOVER

SWGET

ERROR

If you don't want to configure the parameters in the plain xml files, you may use the FA

WebGUI to do this more conveniently. Details are provided in the ―FA Agents –

Installation and Administration― manual.

10.13.1.2 List SAP Services

The script ff_list_services.sh lists the status of all installed SAP services. If no

pools are specified, services of all pools will be shown.

Administrating SAP Services SAP System Handling

Administration and Operation 297

Synopsis

ff_list_services.sh [-ivcCH] [<pool> ...]

Options

-i

Shows inactive services

-v

Verbose mode

-c

Force use of colors for status information

-C

Suppress colors for status information

-H

Suppress headers

-h or -?

Shows usage

10.13.2 Starting and Stopping Application Services

Virtualization of the applications and services demands special measures for starting,

stopping, restarting them etc. These measures are cared for by an SAP service script for

each service type.

With FlexFrame 5.0 up there is an incompatibility concerning starting and

stopping SAP services. The standard sapxx scripts are deprecated and all

requests are now handled by ‗ff_service.sh‘ script.

We also now use Linux command ‗ip‘ to start and stop virtual interfaces.

Therefore you will not see the interfaces using ‗ifconfig‘ as expected. To show

the interfaces you should use now ‗ip -4 –o addr show‘ or ‗ff_service.sh mit

action code ‗icheck‘.

Only project specific solutions like sapxprint will be supported furthermore.

SAP System Handling Administrating SAP Services

298 Administration and Operation

The application or service must not be started directly, e.g. for an SAP instance as <sid>adm with startsap, since in this case the interfaces are

neither supplied with IP addresses, nor is the service control file maintained.

The started application will not work due to the lack of a network connection.

We strongly recommend you not to logon at application node with the client

LAN interface which you want to stop. In case of stopping a service you would

loose your connection and your terminal would hang up.

10.13.2.1 Starting and Stopping SAP Services Without Root Privileges

In order to enable user to start and stop SAP services without having root privileges a

sudo (see man sudo) configuration is shipped with the Application Nodes. This

configuration allows the SAP and the DB administration users to start their respective

services (including their virtual hostname's IP addresses). The user command line

requires a prefix of sudo in addition to the standard command line (see example below):

> sudo /FlexFrame/scripts/ff_service.sh –s <SID> -t db -a start

The sudo mechanisms will check against its configuration file if the user (or the

responsible group) is allowed to run the command and run it with root privileges.

The configuration file is stored on the Application Node and can be found in

Linux: /etc/sudoers

If FlexFrame itself needs extensions these are made in our standard images.

If there are any requirements to extent the configuration at customer site you can make

these extensions in the corresponding configuration files. We strongly recommend to

adjust the intended modifications with professional service support because in the worst

case your system will not work correctly furthermore.

To make these modifications persistent in Linux you can modify the root image directly

without a maintenance cycle. But with each update of FlexFrame you need to modify your

image again.

Administrating SAP Services SAP System Handling

Administration and Operation 299

Content of FlexFrame configuration file on Linux:

# sudoers for FlexFrame Application Nodes

Command definitions

Cmd_Alias SERVICE = /FlexFrame/scripts/ff_service.sh,

/FlexFrame/scripts/view_hosts

Cmd_Alias SAN = /FlexFrame/scripts/ff_san_*

This configuration means, all members of the operating system group sapsys are

allowed to execute the script named ff_service.sh and view_*. Members of oper,

sdba or dba just have the permission to execute the scripts ff_san_*

additionally. Entering a password is not needed.

10.13.2.2 SAP Service Scripts

All Services (except project specific ones) are started by:

ff_service.sh –s <SID> -t <type> [-i <id> ] –a <action> \

[ -d ] [ –u <user> -c <password> ]

The call parameters are:

-i <id>

Distinction of several similar instances of a service type of an SID; 2 digits,

numerical.

-s <SID>

System ID (SID),.

-a <action>

The action to be performed with the application or service. The actions are start,

stop, restart, status, cleanup, watch, nowatch, istart and istop.

-t <type>

Distinction of the different service types.

ascs (classic, message server)

app (classic, application server, ABAP)

bobj (BOBJ - Business Objects)

ci (classic, central instance, ABAP)

cms (CMS - Content Server)

SAP System Handling Administrating SAP Services

300 Administration and Operation

db (Database)

ers (classic, enqueue replicated server)

j (classic, application server, Java,; also used for jc if version ge

SAP 7.1)

jc (classic, central instance, Java)

lc (Live Cache)

mds (MDM server)

mdss (MDM syndication server)

mdis (MDM import server)

scs (classic, message server)

smd (SMD -Solution Manager Diagnostics)

trx (TREX – Ssearch and Classification)

-d

Switch on the debug mode.

If you need the Debuglog from requests done by FA Agents you create a file by

‗touch /tmp/switch_on_debug.<SID>‘ for the SID you want to debug. All requests for

‗SID‘ store a Debugfile in /FlexFrame/scripts/log. Please do not forget to delete the

file and the Debuglogs if you will not need Debuglogs furthermore.

-u <user>

To get the state of BOBJ 4.0 an authentification is needed.

<user> is the login user for CMC specified in installation

procedure. Default user is ‗Administrator‘.

-c <password>

Is the password of <user> used for authentification to access

BOBJ.

10.13.2.3 SAP Service Script Actions

In the following, the SAP service script actions are described:

Administrating SAP Services SAP System Handling

Administration and Operation 301

start

This checks whether the application or service in question is already running. If it is,

it is not restarted. It also checks whether required applications or services are

running.

The required virtual IP addresses are assigned to the relevant interfaces for the

Client LAN and Server LAN (ifconfig <ifc> <ip-adr> netmask <netmask>

up). The application is started and the service control file is written.

SAP System Handling Administrating SAP Services

302 Administration and Operation

stop

This checks whether the application or service in question is running. If it is running,

it is stopped.

The application is terminated, the service control file is deleted and the virtual IP

addresses are separated again from the interfaces (ifconfig <ifc> down).

status

This checks the logical status of the application or service. The functional availability

is not tested.

With BOBJ 4.0 (SBOP) the request to get the state by using

‗/usr/sap/<SID>/…/ccm.sh‘ requires an authentification

restart

This merges the actions stop, cleanup and start in one call; restart is

intended for restarting a malfunctioning application.

cleanup

This kills application processes that are still running and deletes occupied resources

such as shared memory, semaphores and the message queue.

Note that this action may only be performed after an attempt to stop the application

has failed.

nowatch

This removes the application from monitoring by the high-availability software FA

Agent without the application having to be restarted. The application itself retains its

current status. The default status after application startup is 'watch'.

watch

This includes the application again in monitoring by the high-availability software

(FA Agents) without having to restart the application. The application itself retains its

current status.

istart

Start the service interfaces. You do not need to make a distinction between server

and client lan. The script itself selects the necessary interfaces.

This action code cannot be used with type ‗ERS‘ because it does not have a own IP

in client or server lan during operation.

istop

Stop the service interfaces.

This action code cannot be used with type ‗ERS‘ because it does not have a own IP

in client or server lan during operation.

Administrating SAP Services SAP System Handling

Administration and Operation 303

10.13.2.4 SAP Service Script logging

For diagnostic issues SAP service script writes different levels of logging information

standard logging is always written to stdout

/FlexFrame/scripts/log/sapservice.log logs each call of ff_service.sh and its

exitcode. You should reorganize it periodically to get rid of old entries.

/FlexFrame/scripts/log/sap<type>_<SID>_<action>.log logs each call of

ff_service.sh and its exitcode. Additionally the messages written to stdout are

shown, too. A new request overwrites an already existing file for the same SID,

service and action.

/FlexFrame/scripts/log/sapservice.<pid>.debug.<SID>.<type>.<inr>.<action>

contains an extended debuglog if option ‗-d‘ is set by ff_service.sh.

10.13.2.5 SAP Service Script User Exits

The FlexFrame script for SAP Services (ff_service.sh) provide a user exit as a shell

script /FlexFrame/scripts/user_script. If this user_script exist, the

FlexFrame script will call it twice, 1st in the beginning, 2

nd at the end of the start-/stop-

scripts. This means two phases, a pre-phase and a post-phase. The pre-phase will be

run before any action, the post-phase will run after all action. Because of plausibility

errors while executing the FlexFrame start-/stop-script, it is possible that the post-phase

will be omitted.

A sample user_script is delivered as user_script.TEMPLATE.

With FlexFrame 5.0 new service types were introduced. Those new service

types have to be added to customer specific user script. The name is built by

the rule ‗sap<service>‘, e.g. ‗sapsmd‘, ‗saptrx‘. Please refer to template file

available in directory /FlexFrame/scripts.

To achieve a pool dependent function of user_script, put the script into the pool image

and create a symbolic link to it in FlexFrame/scripts. Seen from the Control Node:

Control1: # ln -s

/FlexFrame/pooldata/config/scripts_config/user_script/FlexFrame/scripts/

user_script

If no pool dependent function is needed, the script user_script can be located directly

in /FlexFrame/scripts.

SAP System Handling Administrating SAP Services

304 Administration and Operation

10.13.3 Return Code of the SAP Service Script

The SAP service scripts issue a return code (exit code). The meaning of this code can be looked up in the sapservice_functions file:

#================================================================

# common exit codes for service scripts

#================================================================

no_error=0 # no error

wrong_parameter_count=1 # Bit 0, wrong number of parameters

plausibility_error=2 # Bit 1, plausibility error

interface_server_lan_error=4 # Bit 2, error at server lan interface

up/down

interface_client_lan_error=8 # Bit 3, error at client lan interface

up/down

service_start_stop_error=16 # Bit 4, error at service

start/stop/status/...

any_error=32 # Bit 5, any other error

san_mount_error=64 # Bit 6, san mount error

user_script_error=128 # Bit 7, user_script error

For actions like start, stop, nowatch and watch, the exit code should be no_error

normally. For the action status, the exit code is dependent from various factors:

The installation check against the SAP service may have failed

(plausibility_error)

the SAP service is possibly not running or not running on this host (any_error)

the SAP service is running on this host (no_error).

In addition to these exit codes, the SAP service scripts print out messages as described

in the section ―Start/Stop Script Errors‖ on page 365.

10.13.4 Starting and Stopping Multiple SAP Services

The following scripts are provided for starting and stopping multiple applications and

services:

Script name Application Place of execution

start_all_sapservices Initial start of all configured

applications

Only on a Control

Node

stop_all_sapservices Stopping all running

applications

Only on a Control

Node

stop_all_sapservices_SID Stopping all running

applications of one SID

Only on a Control

Node

Administrating SAP Services SAP System Handling

Administration and Operation 305

Script name Application Place of execution

stop_all_sapservices_local Stopping all running

applications on the local

node

Only on an

Application Node

Call syntax for pool-dependent scripts:

stop_all_sapservices [<pool name>]

stop_all_sapservices_SID <sid> [<pool name>]

Functionality of the SAP service scripts for multiple services:

start_all_sapservices

Initial start script for customer defined sap service landscapes. This can be the

whole FlexFrame landscape, a pool, a SAP system landscape with a productive

system, a quality assurance system and a development system, or a single SAP

system with database, central instance and application instances. This script can be

duplicated to create various initial start scripts with different names, e.g.

start_all_sapservices_pool2. A sample start_all_sapservices script is

delivered as start_all_sapservices.TEMPLATE.

stop_all_sapservices

Shutdown script for all running sap services in a dedicated pool. The advantage of

this script is that sap services have not to be stopped one-by-one.

stop_all_sapservices_SID

Shutdown script for a singe SAP system (one SID) with all belonging sap services

like application instances, central instances und database instance in the right order.

stop_all_sapservices_local

Shutdown script for all running sap services on one Application Node. This script will

be integrated via run levels. The information base for these

stop_all_sapservices* scripts are the /FlexFrame/scripts/log/*_host

files.

10.13.5 Removing an Application from Monitoring by FA Agents

If the applications or services are started with the scripts for virtualization, they are

monitored by the FA Agents. If you do not want this, be it for tests, installation or

upgrades, you have to inform the high-availability software of this, using the additional parameter nowatch when the application is started.

SAP System Handling Administrating SAP Services

306 Administration and Operation

Example:

The central instance of BW1 is to be started without monitoring by the high-availability

software:

blade1 # ff_service.sh -s bw1 –t ci –a nowatch

blade1 # ff_service.sh –s bw1 –t ci –a start

or from a Control Node:

control1 # ssh blade1 ff_service.sh -s bw1 –t ci –a nowatch

control1 # ssh blade1 ff_service.sh –s bw1 –t ci –a start

If a running application is to be excluded from monitoring by the high-availability software

without being restarted, this is possible, using the nowatch option. The application then

retains its current status.

Example:

The central instance of BW1 is to be excluded from monitoring by the high-availability

software while running:

blade1 # ff_service.sh -s bw1 –t ci –a nowatch

Alternatively, call the following from a Control Node:

control1 # ssh blade1 ff_service.sh -s bw1 –t ci –a nowatch

If a running application is to be included (again) into monitoring by the high-availability software without being restarted, this is possible using the watch option. The application

then retains its current status.

Example:

The central instance of BW1 is to be included into monitoring by the high-availability

software while running:

blade1 # ff_service.sh -s bw1 –t ci –a watch

or from a Control Node:

control1 # ssh blade1 ff_service.sh -s bw1 –t ci –a watch

Administrating SAP Services SAP System Handling

Administration and Operation 307

10.13.5.1 Stopping and Starting an Application for Upgrades

Using r3up

The r3up upgrade control program starts and stops the central instance or application

server in various upgrade phases, using the conventional startsap and stopsap. The

start and stop scripts for visualization are therefore left out.

Suitable measures must be taken to ensure that the virtual host name for <sid>adm is

always available and that the interfaces are supplied with the correct IP addresses. In

addition, the application has to be removed from monitoring by the high-availability

software.

Note that for the following solution steps, the application must have been started

correctly beforehand with start and stop scripts for virtualization, so that the

matching virtual host name is set by default in the $HOME/hostname_default

script.

Remove the application from monitoring by the high-availability software and run it.

r3up can now shut down and start the application directly with stopsap and

startsap; the IP addresses remain assigned to the interfaces and the host name is

retained for <sid>adm without any changes on the correct virtual host name.

Include the application again into monitoring by the high-availability software after the

upgrade.

10.13.6 Service Switchover

There is no command line tool to enforce a direct switchover. To switch over an SAP

service like a Central Instance, Database Instance, Application Instance etc., it is required

to first stop this SAP service on the actual Application Node and to start this SAP service

on the destination Application Node thereafter.

Examples:

The central instance of OL4 from pool pool1 is running on klinge4 and should

switchover to klinge5.

First, check whether the Central Instance OL4 is running on klinge4:

For a first overview to see which SAP service should be running on which Application Node in the pool pool1, run

control1 # view_hosts pool1

app_44_ol4 should be running on klinge4

app_56_ml4 should be running on Baby_4

ci_ml4 should be running on RX300-01

SAP System Handling Administrating SAP Services

308 Administration and Operation

ci_ol4 should be running on klinge1

ci_osm should be running on Baby_3

db_osm should be running on Baby_3

db_ml4 should be running on RX300-01

db_ol4 should be running on klinge4

The output should be means that the script view_hosts does not really check the

state of the SAP services. It only checks the FlexFrame/script/log/*_host files

which are created immediately after a successfully start of the corresponding SAP

service.

For a detailed status information you can use:

control1 # ssh klinge4 ff_service.sh -s ol4 –t ci –a status

The exit code from our SAP service script will be ―0‖ in this case. For any other errors

while executing our SAP service script, it will return an exit code not equal ―0‖.

control1:/ # ssh klinge1 ff_service.sh -s ol4 –t ci –a status

INFO: Version of ff_service.sh: 1.20.2.21

INFO: …….

INFO: Service ci for SID CL4 should be running on this host

INFO: ….

INFO: server_lan interface cis02-se 172.20.20.129 is (already) up

INFO: Exit code: 0 (no error)

Stop the running Central Instance OL4 on klinge4:

control1 # ssh klinge4 ff_service.sh -s ol4 –t ci –a stop

And now start the stopped Central Instance OL4 on the new destination klinge5:

control1 # ssh klinge5 ff_service.sh -s ol4 –t ci –a start

The manual switch over should now be complete.

Administrating SAP Services SAP System Handling

Administration and Operation 309

10.13.7 Use ServicePings from FA Agents

With newer SAP releases it could be insufficient of FA Agents just check the process list

of a host to decide if o service is running or not.

Now the FA Agents offers the possibility to check it using SAP executables like

‘sapcontrol … GetProcessList‘. To activate that feature you

Switch to /opt/myAMC/scripts

Remove ‗exit 99‘ from ServicePings*.sh

Activate ‗ServicePing‘ for each pool you want to use it by the myAMC WebGUI

Restart ‗ff_manage‘ either bei ‗hb_gui‘ or by ‗ff_ha_cmd.sh‘

Administration and Operation 311

11 Software and Hardware Update and Maintenance

11.1 Upgrading the Entire FlexFrame Landscape

Updating the entire FlexFrame to the 5.1A release depends on the release you are

starting from:

Upgrading from FlexFrame 3.x or lower version

Since this release has reached "Extended Support End" in its lifecycle you need to

upgrade to FlexFrame 4.2B first. Please contact the Competence Center SAP

(CCSAP). There exists a process, which has to be adapted on project specific

needs.

Upgrading from FlexFrame 4.0A, 4.1A or 4.2A

Since these releases have reached "General Support End" in their lifecycles you

need to upgrade to FlexFrame 4.2B first.

Upgrading from FlexFrame 4.2B or 5.0A

See the document ―Upgrading FlexFrame 4.2B or 5.0A to 5.1A‖.

11.2 Software Upgrade on the Control Node

This section describes FlexFrame software installation and upgrades on the Control

Node. For upgrade of the Control Node see previous chapter. Control Nodes are not

designed to run any third party software (except saprouter).

Since FlexFrame 5.1 most of the installation of FlexFrame specific software (products) on

top of the underlying SLES takes place during first boot of the Control Node. You know

this from the ―Quick Installation‖ section in the ―Installation of a FlexFrame Environment‖

manual.

That way the installation of these products moved from Fujitsu Mastering Site to

customer‘s site. The advantage is that all those products are now available at your site

which might become important in case of needs to reinstall them.

In addition this means that these software products are no longer inside the pre-built

Control Node image.

To find out more details about the performed installation of these software products you

can inspect the well-known log file directory /var/log/FlexFrame. The very first

log_<control node name> file contains the complete installation logs.

To find out details about the software products before the actual installation you may

inspect the Service CD. Insert the Service CD and run the following commands:

Software and Hardware Update and Maintenance

312 Administration and Operation

control1: mount /media/dvd

control1: cd /media/dvd/CONOS

control1:/media/dvd/CONOS # ./installer.sh –l

This will list (-l) the products and product versions.

Do not run this installer in any other than the mentioned list (-l) mode unless

instructed to do so! Running in install (-i) or upgrade (-u) mode may damage your FlexFrame.

New versions of these mentioned software products will be deployed with new Service

CDs. These will contain instructions how to install new versions if needed (see also

chapter ―11.2.1 FlexFrame Control Node Software‖

You may want to install other software products (i.e. Third Party Products) beside the

mentioned from the Service CD. Remember (as stated in chapter 2.1):

The FlexFrame Control Nodes are seen as an appliance

Installing third party software on Control Nodes or Application Nodes may cause

functional restrictions and other malfunctions of the FlexFrame system software.

Fujitsu shall have no liability whatsoever whether of a direct, indirect or

consequential nature with regard to damages caused by the third party software

or its erroneous installation.

SLES patches may have effects on the functionality of FlexFrame Control Node

and Application Node images. Therefore security and recommended patches for

SLES will be evaluated, whether they are applicable within a FlexFrame Control

Node or Application Node image. The results and further remarks on installation

will be published within the scope of the solution contract.

If you install or upgrade software products on a project specific base you must

use the ff_projectinfo.sh command to document this change!

See chapter 11.9 Project specific changes for details on this command.

Nevertheless we will describe here the general approach to update/install software on the

Control Nodes, on a few examples. In your special case this approach can differ to the

ones described here.

Software and Hardware Update and Maintenance

Administration and Operation 313

11.2.1 FlexFrame Control Node Software

As stated in the previous chapter, new versions of the FlexFrame Control Node products

will be deployed with a new Service CD.

If needed or instructed by Fujitsu support an update of these components is possible by

launching the respective installer command: Basically you just have to change the

directory to the product location and start the installer.

Example: To (re-) install the ―Java Runtime Environment‖ product insert the Service CD

and run the following commands:

control1: mount /media/dvd

control1: cd /media/dvd/CONOS/jre

control1:/media/dvd/CONOS/jre # ./installer.sh –iv

Run these installers only if needed or instructed by Fujitsu support.

Normally there is a README or other installation instructions available which

guides through the process.

All of these product installer cover the Control Node only where the Service CD has

been mounted. You will have to run them on the other Control Node as well. The

mentioned README or installation instructions will point out to that fact.

11.2.2 Updating/Installing a New Linux Kernel

The following sections describe how to install a new Linux kernel for Control Nodes in

FlexFrame. This only has to be done if instructed to do so by Fujitsu support.

The process will be described on examples accomplished on a development system. The

version numbers here in this document are not generally binding; they must be replaced

with the correct information for your system.

This approach describes the kernel installation without the recommended Server Start

CD.

11.2.2.1 Software Stage

To install and update any of the software packages in FlexFrame, it is useful to mount

your NAS system or jukebox or any other software stage on a local mount point.

For example, your software stage is /FlexFrame/volFF/FlexFrame/stage/ and

the new kernel to install is 2.6.16.60-0.63.1.

Software and Hardware Update and Maintenance

314 Administration and Operation

Copy the delivered software packages to the software stage:

control1: # cd /FlexFrame/volFF/FlexFrame/stage

control1:/FlexFrame/volFF/FlexFrame/stage # mkdir -p CN/kernel

Kernel:

control1: # cp <somewhere>/kernel-smp-2.6.16.60-0.63.1.x86_64.rpm

/FlexFrame/volFF/FlexFrame/stage/CN/kernel

Kernel kdump:

control1: # cp <somewhere>/kernel-kdump-2.6.16.60-0.63.1.x86_64.rpm

/FlexFrame/volFF/FlexFrame/stage/CN/kernel

Kernel source:

control1: # cp <somewhere>/kernel-source-2.6.16.60-0.63.1.x86_64.rpm

/FlexFrame/volFF/FlexFrame/stage/CN/kernel

With SLES 10 (Service Pack 1 and later) Novell has introduced the Partner

Linux Driver Process (PLDP). There is no need to download drivers for the

update kernel. The existing drives are reused for the new kernel.

11.2.2.2 Install the New Kernel

To install the new kernel, execute the following command:

control1: # cd /FlexFrame/volFF/FlexFrame/stage/CN/kernel

control1:/FlexFrame/volFF/FlexFrame/stage/CN/kernel #

rpm -ivh kernel-smp-2.6.16.60-0.63.1.x86_64.rpm

control1:/FlexFrame/volFF/FlexFrame/stage/CN/kernel # rpm –ivh kernel-

kdump-2.6.16.60-0.63.1.x86_64.rpm

control1:/FlexFrame/volFF/FlexFrame/stage/CN/kernel # rpm –ivh kernel-

source-2.6.16.60-0.63.1.x86_64.rpm

11.2.2.3 Reboot the Control Node

Reboot the Control Node and test the functionality of the updated Control Node, Linux-

HA, network connectivity, and so on.

The uname -r command should display the new kernel version:

control1:~ # uname -r

2.6.16.60-0.63.1-smp

Software and Hardware Update and Maintenance

Administration and Operation 315

11.2.3 Backup/Restore of FlexFrame Control Nodes

To backup or restore Control Nodes use the ff_cn_backup.sh script.

Synopsis

ff_cn_backup.sh [-dr] [-f <file name>] [-l <list>]

Options

-d

Debug

-r

Restore

-f <file name>

Use file name for restore

-l <list>

Blank separated list of files (quoted)

11.2.3.1 Backup of a Control Node

Most of the files on a Control Node are shipped with the installation DVD. Necessary data

can be backed-up using ff_cn_backup.sh.

The default behavior (without any options) is to backup the configuration files in a zipped tar file at /FlexFrame/volFF/FlexFrame/backup/<name_of_cn><date>.tgz.

Additional files can be backed-up using the -l option.

The Control Nodes are not intended to have additional software installed on.

Therefore backup in the regular sense is not required. Installation from the DVD

is much faster than any restore.

11.2.3.2 Restore of a Control Node

Restore can be done after installation using the Installation DVD along with the

configuration media. Make sure to install the same patches and RPM packages of the

FlexFrame tools as before. You have to use the Service CD as well.

After the installation, ff_cn_backup.sh must be called using the option -r.

By default, the restore will pick the latest backup it can find. If this should not be the

appropriate backup file, use the option -f <file name>.

Software and Hardware Update and Maintenance

316 Administration and Operation

In detail, follow these steps to restore a Control Node:

1. Boot the Control Node Installation DVD.

2. Provide the configuration file ff_config.xml with the USB stick or floppy. It may be

obtained manually either from the /opt/FlexFrame/etc folder of the remaining

Control Node or out of the *.autosave.tgz backup files in the

/FlexFrame/volFF/FlexFrame/backup/ folder which is accessible from the

remaining Control Node..

You will get a warning message about wrong checksums if you reuse a

configuration file which has had been installed on a Control Node. This

warning message is only intended to give you a hint to double-check the

configuration file. It can be ignored safely.

3. Insert the Service DVD with the same patch level as installed before (same as the

remaining Control Node) and continue the installation as described in the ―Installation

of a FlexFrame Environment‖ manual.

4. If the passwords are requested, enter the current passwords.

5. Enter the correct Control Node (first or second) for this Control Node during

installation.

6. Once the system is booted do not follow the installation instructions from the

―Installation of a FlexFrame Environment‖ manual. Instead copy the file

FF_XMLHash.pm out of the *FF_XMLHash*autosave backup files in the

/FlexFrame/volFF/FlexFrame/backup/ folder which is accessible from the

remaining Control Node to /opt/FlexFrame/etc. Then execute the following

command:

control1:~# ff_cn_folders.pl -notrans

7. Check the access to /FlexFrame/volFF/FlexFrame/backup

8. Use ff_cn_backup.sh with the option -r and optionally -f <file name> to

restore the original contents of the Control Node's configuration files.

control1:~# ff_cn_backup.sh –r –f <file name>

9. Add additional security related changes to some config files using the command:

control1:~# ff_enh_sec.sh

10. Compare the two Control Nodes using the command:

control1:~# ff_cn_cmp.sh

If there are differences in files listed, copy them over from the other Control Node

using:

Software and Hardware Update and Maintenance

Administration and Operation 317

scp -p control2:/<path> /<path>

11. Reboot the Control Node using the init 6 command.

12. Start the Linux HA software stack: control1:~# rcheartbeat start

After the command returned, it will take a while until the complete Linux-HA software stack is up and running. This can be verified with the "crm_mon" command; both

Nodes should be shown as online and the ldap master or replica should be switched

over to the restored Control Node. A "ff_ha_cmd.sh cleanup" of some

resources might be necessary in some cases, especially if the old Control Node was

hard powered off.

13. If the Linux-HA cluster works as expected, activate it permanently: control1:~# insserv heartbeat

11.3 Maintenance of the Control Node - Hardware

11.3.1 Exchanging a Control Node

If a Control Node must be replaced for any reason, attention has to be paid to the fact

that the replacing hardware has to be the same as the replaced component. Otherwise,

problems with missing Linux drivers might occur. Problems due to hard disk failures

should usually not occur, since the Control Node's hard disks are mirrored through

hardware-based RAID-1.

11.3.1.1 Hardware Failed, Hard Disk and Installed OS are not Affected

In case that other hardware components than one of the HDs are broken (e.g. memory,

CPU, etc.) and the installed OS image is still operational, these hardware components

have to be replaced with equivalent ones. Even the entire hardware can be replaced -

excluding the HDs. The HDs can be removed from the existing hardware, plugged into

the new hardware and booted, after checking the RAID controller's BIOS for hardware

RAID settings being enabled. Even if one of the hard disks is broken, it can be replaced

with the same model and synchronized with the working hard disk through the RAID

controller's BIOS.

In this case the Control Node hardware must be replaced by the same model with the

same disk controller and the same network interface controllers. The approach is very

simple, plug in your hard disks in the new Control Node in the same order as the old one,

power on the Control Node, enter the RAID controller BIOS and check the parameters;

hardware raid should be enabled as before.

See also the manual ―Installation of a FlexFrame Environment―, chapter ―Control Node

Host RAID Configuration‖.

Software and Hardware Update and Maintenance

318 Administration and Operation

11.3.1.2 One Hard Disk is Defect, the Other One is Undamaged

The Control Nodes OS installation must be on a RAID-1 disk array.

Replace the defect hard disk with the same model and synchronize it with the

undamaged one via the RAID controller BIOS. This approach is dependent from the

RAID controller model and its BIOS.

11.3.1.3 The Control Nodes OS is Damaged

If, for any reason, the operating system of a Control Node should be damaged, it has to

be installed anew with the original Control Node installation DVD. The configuration has

to be recovered from a previous backup. See section ―Backup/Restore of FlexFrame

Control Nodes‖ on page 315.

11.3.2 Replacing a Network Card – Control Node

If the motherboard or the PCI network card in a Control Node is replaced, no further

actions are required since there is no reference to the replaced MAC addresses.

11.4 Maintenance of Application Nodes - Software

11.4.1 Introduction

Maintenance of Linux Application Node Images – often referred to as ―the Maintenance

Cycle‖ – allows modification of the Application Node‘s software which is normally hardly

possible owing to the shared image concept.

After a short schematic overview this section describes the following:

installing Application Node images from installation media

creating a Maintanance Base Image

assigning the Maintenance Image, booting and maintaining

reverting the Maintenance Image

migrating remaining Application Nodes

maintaining use cases

FlexFrame 5.0 introduced a new Application Node Image concept. It is still a shared

image concept as known from earlier FlexFrame releases but was revolutionized

particularly in order to improve the Maintenance of Application Node Images.

For details refer to the section "Setting up Application Nodes" of the "Installation of a

FlexFrame Environment" manual.

By help of the new concept the Maintenance of an Application Node Image is now as

easy as 1-2-3 as you will see in this section.

Software and Hardware Update and Maintenance

Administration and Operation 319

11.4.2 Schematic Overview

This section gives an idea how the Maintenance Cycle works. Basically this cycle

consists of just three steps:

1. Creating a Maintenance Base Image

(ff_install_an_linux_images.sh -m ...)

2. Assigning the Maintenance Image, creating boot image, booting the Application

Node and maintaining

(ff_an_adm.pl –op os ...; ff_new_an.sh –n ...)

3. Reverting the Maintenance Image (ff_install_an_linux_images.sh –r ...)

Then all of the Application Nodes can use the maintained image instead of the old one to

boot.

The following figure visualizes the entire process of maintaining of an Application Node

Image i.e. the ―Maintenance Cycle‖. You will recognize the figure since it has been

introduced in the ―Setting up Application Nodes‖ section of ―Installation of a FlexFrame

Environment‖ manual.

Software and Hardware Update and Maintenance

320 Administration and Operation

1

`

os/Linux/FSC_50A*/

productive mode

generation 1

/FlexFrame/volFF/

Base Im

ag

es

root_img and

var_img of Node #n

root_img and

var_img of Node #2root_img and

var_img of Node #1

base_img/

root_img/

os/Linux/CST_50A*/

root_img and

var_img of Node #n

root_img and

var_img of Node #2root_img and

var_img of Node #1

base_img/

root_img/

os/Linux/MNT_50A*/

root_img of

Maintenance-Node

base_img/

root_img/

maintenance mode

generation 1’

productive mode

generation 2

Ro

ot

Imag

es

1.

2.

3.

Software and Hardware Update and Maintenance

Administration and Operation 321

Please take notice of:

There is always a ―Base Image‖ needed for starting a Maintenance Cycle.

This ―Base Image‖ does not have to be a newly installed image from an ―Application

Node – Images Linux SLES1x‖ DVD - you can use an image from an earlier

Maintenance Cycle. This implies the option to ―maintain a maintained image‖.

There is no need to disable/enable FA Agents manually.

There is no need to change exports on NAS manually.

11.4.3 Installing Application Node Images from Installation Media

This section describes how to install an Application Node Image delivered by Fujitsu. You

have to follow the instructions in this section only if you did not install a FlexFrame 5.1A

Application Node Image yet.

Due to the changed Application Node Image concept it is not possible to use old

Application Node Images (i.e. images from FlexFrame 4.2B and earlier). You

have to install FlexFrame 5.1A Application Node Images in order to run a

Maintenance Cycle.

You need to install an Application Node image if you did not have installed it yet.

This Application Node image comes as a so called ―Base Image‖ and contains all needed

components in a single file delivered by the ―Application Node – Images Linux SLES1x‖

DVD(s). To install the Application Node image you just have to launch one command:

ff_install_an_linux_images.sh which was already introduced in the ―FlexFrame

for SAP – Installation of a FlexFrame Environment‖ guide‘s section ―Setting up

Application Nodes‖. The following sections describe the

ff_install_an_linux_images.sh command in more detail since it plays an

important role in the Maintenance of Linux Application Node Images.

11.4.3.1 Installing the Application Node Image

To install Application Node Images delivered by Fujitsu insert one of the ―Application

Node – Images Linux SLES1x‖ DVDs into the DVD device of one of the Control Nodes.

You can only install one ―Application Node – Images Linux SLES1x‖ DVD at the

same time onto your FlexFrame system.

Since the ff_install_an_linux_images.sh command locks system wide

you must install the needed DVDs one after another.

The following examples use the first Control Node and the ―Linux SLES10‖ DVD. At first

mount it:

Software and Hardware Update and Maintenance

322 Administration and Operation

cn1:/ # mount /media/dvd

After changing to the /media/dvd directory you will see the

ff_install_an_linux_images.sh command. This is actually only a wrapper to the

/opt/FlexFrame/bin/ff_install_an_linux_images.sh command in order to

increase the ease of use. It has been placed there since the FlexFrame 4.2B DVDs had

that command as well and you can use it in the same way:

cn1:/media/dvd # ./ff_install_an_linux_images.sh

You might consider using the –v option when running the command for the first

time.

This will install the contents of the DVD to the Control Node as defined by the DVD itself: the cfg/install.cfg configuration file contains the information. In the example you

will find after installation the following directory:

cn1:/FlexFrame/volFF/os/Linux # ls -l

total 20

drwxr-xr-x 5 root root 4096 May 10 11:43 ./

drwxr-xr-x 4 root root 4096 May 4 10:26 ../

drwxr-xr-x 4 root root 4096 May 10 12:14 FSC_5.1A00-001.SLES-10.x86_64/

If you want to install for any reason the contents of the DVD to a different location keep in mind that only images in /FlexFrame/volFF/os/Linux can be used by FlexFrame

administration commands.

To do so use the –t option of the ff_install_an_linux_images.sh command:

cn1:/media/dvd # ./ff_install_an_linux_images.sh \

-t /FlexFrame/volFF/os/Linux/FSC_5.1A00-001.SLES-10.x86_64-another-

example

It is worthwhile not changing the prefix of the directory name to keep track of the

images. All images delivered by Fujitsu are named

FSC_<FlexFrame_version>.<OS_version>.<OS_architecture> where

FSC_ stands for FlexFrame Software Container.

Consult the ff_image_types(8) manual page for a suggestive naming of the

arguments to the -s and -t options..

The ff_install_an_linux_images.sh(8) manual page contains further compre-

hensive information about ff_install_an_linux_images.sh.

Software and Hardware Update and Maintenance

Administration and Operation 323

11.4.3.2 Understanding the Application Node Image

Looking closer to the installation results of the example it will reveal the following

structure (trimmed):

cn1:/FlexFrame/volFF/os/Linux # ls -l FSC_5.1A00-001.SLES-10.x86_64/

drwxr-xr-x 32 root root 4096 May 10 12:14 base_img

-rw-r--r-- 1 root root 632 May 10 12:14 ff_pxe.conf

-rw-r--r-- 1 root root 170 May 10 12:14 modules.conf

drwxr-xr-x 32 root root 4096 May 10 12:14 root_img

The most important result of the installation process is the ―Base Image‖ base_img: only

a ―Base Image‖ is suitable as a source for a Maintenance Cycle. This ―Base Image‖ has

been created from the DVD‘s (virgin) ―Base Image‖ by copying and configuring during the

installation process. This configuration is a ―generically‖ configuration by help of the

contents of the fflanic RPM and does not involve any personalization of this ―Base

Image‖. This means that no Linux Application Node can use such a ―Base Image‖ to boot

from.

During the installation process you certainly have seen the following output:

cn1:/media/dvd # ./ff_install_an_linux_images.sh

Info: Extracted /media/dvd/images/base_img_FSC_5.1A00-001.SLES-

10.x86_64.cgz ... 100%

Info: Starting ff_prep_linux_an.sh

Info: ff_prep_linux_an.sh done

Info: Finally copying base image to root image .. 100%

Info: ff_install_an_linux_images.sh done

Notice the line ―Finally copying base image to root image‖: at this point the

entire ―Base Image‖ base_img has been copied to a ―Root Image‖ root_img. This

―Root Image‖ works similar to the ―Root Images‖ you might know from FlexFrame 4.2B:

the Application Node(s) can use it to boot from. The process of copying the ―Base Image‖

to the ―Root Image‖ is almost as time consuming as the installation of the ―Base Image‖

itself but needs to be done only once. Therefore it was done during the installation and it

will speed up the following processes.

A ―Root Image‖ will get personalized later on during the ff_new_an.sh and will provide

the real ―/‖ (root) image and the real ―/var‖ image of the Application Nodes.

11.4.4 Step #1: Creating a Maintenance Base Image

As stated in the ―Schematic overview‖ the first step in the Maintenance Cycle is to create

a so called ―Maintenance Base Image‖.

Software and Hardware Update and Maintenance

324 Administration and Operation

1

`

os/Linux/FSC_50A*/

productive mode

generation 1

/FlexFrame/volFF/

Base Im

ag

es

root_img and

var_img of Node #n

root_img and

var_img of Node #2root_img and

var_img of Node #1

base_img/

root_img/

os/Linux/CST_50A*/

root_img and

var_img of Node #n

root_img and

var_img of Node #2root_img and

var_img of Node #1

base_img/

root_img/

os/Linux/MNT_50A*/

root_img of

Maintenance-Node

base_img/

root_img/

maintenance mode

generation 1’

productive mode

generation 2

Ro

ot

Imag

es

1.

2.

3.

This ―Maintenance Base Image‖ is ultimately a ―Base Image‖ and therefore it can be

created by help of the already known ff_install_an_linux_images.sh command.

Since a ―Maintenance Base Image‖ has to be created the –m option has to be used.

Based on an existing Linux Application Node Base Image (specified by -s option) the

ff_install_an_linux_images.sh command will spawn another Linux Application

Node Base Image (specified by -t option) with special properties: The spawned image

can be used later on to assign one system to this image by using ff_an_adm.pl and

create one Node which allows software installation. This non-productive Node will be in

Software and Hardware Update and Maintenance

Administration and Operation 325

an exclusive state – the ―Maintenance Mode‖ - and not recognizable by FA Agents. This launch of ff_install_an_linux_images.sh command will copy the entire ―Base

Image‖ to a ―Root Image‖ as known since it is needed to boot a real Application Node

from. (In fact the ―Base Image‖ will be moved to the ―Root Image‖ since in this particular

case the ―Base Image‖ is no longer used. And it saves time and disk space as well.)

Synopsis

ff_install_an_linux_images.sh -m -s <srcdir> -t <trg> \

[-v] [-d] [-i]

Example

cn1:/ # cd /FlexFrame/volFF/os/Linux

cn1:/FlexFrame/volFF/os/Linux # ff_install_an_linux_images.sh –m \

-s FSC_5.1A00-001.SLES-10.x86_64/ -t MNT_5.1A00-001.SLES-10.x86_64/

Notice that the ff_install_an_linux_images.sh command accepts

relative and absolute path names as arguments to the –s and –t option. You

may save some typing when you first change to the

/FlexFrame/volFF/os/Linux/ directory as shown in the example above.

It is worthwhile using a meaningful prefix for the directory name to keep track of

the images. A good start for choosing the name is to replace the original prefix

(here: FSC_) by MNT_ indicating a ―Maintenance‖ image.

Consult the ff_image_types(8) manual page for a suggestive naming of the

arguments to the -s and -t options.

The example uses the just installed Linux Application Node Image of section ―Installing

the Application Node Image‖ i.e. the source image (specified by -s option) is an image

delivered by Fujitsu and therefore has the FSC_ prefix. This is easier to explain in this

example. But this is not a constraint: you can use any image from an earlier Maintenance

Cycle i.e. you can ―maintain a maintained image‖.

11.4.5 Step #2: Assigning the Maintenance Image, booting and maintaining

As stated in the ―Schematic overview‖ we now need to create a Root Image since we

want to boot a real Linux Application Node in order to have a system for the software

maintenance:

Software and Hardware Update and Maintenance

326 Administration and Operation

1

`

os/Linux/FSC_50A*/

productive mode

generation 1

/FlexFrame/volFF/

Base Im

ag

es

root_img and

var_img of Node #n

root_img and

var_img of Node #2root_img and

var_img of Node #1

base_img/

root_img/

os/Linux/CST_50A*/

root_img and

var_img of Node #n

root_img and

var_img of Node #2root_img and

var_img of Node #1

base_img/

root_img/

os/Linux/MNT_50A*/

root_img of

Maintenance-Node

base_img/

root_img/

maintenance mode

generation 1’

productive mode

generation 2

Ro

ot

Imag

es

1.

2.

3.

11.4.5.1 Choosing a Node

Therefore you have to choose one of your Application Nodes. Choose one out of a Pool

Group that meets your needs.

Notice that during the Maintenance Cycle this Application Node is not available

for Applications like SAP services, databases etc. It is not visible to FlexFrame

and can therefore not serve as a spare node during that time.

Software and Hardware Update and Maintenance

Administration and Operation 327

In the following example we have the pool pp1 consisting of four Application Nodes

running on the FSC_5.1A00-001.SLES-10.x86_64 image as displayed by the

ff_an_adm.pl command (output trimmed):

cn1:/ # ff_an_adm.pl --op list-all --pool pp1 | egrep

"Path:|^[[:blank:]]*pa-|Pool"

Pool pp1

Pool Group sles10_x64

pa-rx300-1

OS Path: pafiler-pp1-st:/vol/volFF/os/Linux/FSC_5.1A00-

001.SLES-10.x86_64

pa-rx300-2

OS Path: pafiler-pp1-st:/vol/volFF/os/Linux/FSC_5.1A00-

001.SLES-10.x86_64

pa-rx300-3

OS Path: pafiler-pp1-st:/vol/volFF/os/Linux/FSC_5.1A00-

001.SLES-10.x86_64

pa-rx300-4

OS Path: pafiler-pp1-st:/vol/volFF/os/Linux/FSC_5.1A00-

001.SLES-10.x86_64

We choose the Application Node pa-rx300-1 of pool group sles10_x64 for running

the Maintenace Cycle. Move running applications to the remaining Application Nodes and

shut it down.

cn1:/ # # evacuating the node depends on the running applications

cn1:/ # ssh pa-rx300-1 init 0

11.4.5.2 Assigning

Now follows the actual step #2: the assigning of the created ―Maintenance Image‖ of step

#1 to this node by help of the well known ff_an_adm.pl command:

cn1:/ # ff_an_adm.pl --op os --ospath Linux/MNT_5.1A00-001.SLES-

10.x86_64 \

--name pa-rx300-1

Create the image for the chosen Application Node by help of the well known ff_new_an.sh command:

cn1:/ # ff_new_an.sh –n pa-rx300-1

Software and Hardware Update and Maintenance

328 Administration and Operation

11.4.5.3 Booting

Finally boot the Node by help of the well known ff_ponpoff.pl:

cn1:/ # ff_ponpoff.pl --op power-on --name pa-rx300-1

After a short time the Application Node pa-rx300-1 will be up and running. During the

boot process you certainly have seen the following message:

###########################

RUNNING IN MAINTENANCE MODE

###########################

After login you will get the information again:

cn1:/ # ssh pa-rx300-1

Last login: Mon Mar 22 15:23:45 2010

Authorized uses only. All activity may be monitored and reported.

Welcome to FlexFrame(TM) for SAP(R)

(c) 2010 Fujitsu Technology Solutions

This is Application Node pa-rx300-1

Running in MAINTENANCE mode

Have fun.....

pa-rx300-1:~ #

You see: without manipulating anything the system booted into the ―Maintenance Mode‖.

Checking the filesystems will reveal:

pa-rx300-1:~ # mount -t nfs | grep " / "

10.10.11.14:/vol/volFF/os/Linux/MNT_5.1A00-001.SLES-10.x86_64/root_img

on / type nfs (rw,v3,…)

pa-rx300-1:~ # mount -t nfs | grep " /var "

pa-rx300-1:~ #

pa-rx300-1:~ # pgrep -lf AppAgt

pa-rx300-1:~ #

Software and Hardware Update and Maintenance

Administration and Operation 329

Looking closer to this command output shows that

the root file system is mounted read-write

In contrast to the normal operation the root file system is now writable and allows

installation of software

the /var file system is not mounted

In contrast to the normal operation the /var directory resides on the root file system

and is no separate file system. This means that installation of software affecting the

/var will go into the root file system.

the FA Agents are not running

In contrast to the normal operation the FlexFrame will not recognize the Application

Node pa-rx300-1 due to the non-started FA Agents.

11.4.5.4 Maintaining

This Application Node runnig in Maintenance Mode is prepared to get maintained now.

Before starting please keep in mind the following:

Changing the Application Nodes‗ software is delicate since any change can affect the

functionality of the entire FlexFrame. Therefore it is essential to test carefully all

changes.

Installing/Upgrading software should be done based on RPMs since it is easier to

keep track of the changes you made.

Installing SLES patches may have huge effects on the functionality of FlexFrame

Application Node images. Therefore evaluate security and recommended patches

whether they are applicable within a FlexFrame Application Node image.

Installing of complete SLES Service Packs is not allowed during the Maintenance

Mode. Service Packs contain a lot of changes that can have influences to the

FlexFrame functionality. Service Packs can also contain initial configuration files that

will substitute the FlexFrame configurations.

The only way to obtain a new SLES Service Pack for Application Nodes is to install

new Application Node images delivered from Fujitsu on new ―Application Node –

Images Linux SLES1x‖ DVDs.

The section 11.4.8 Maintaining Use Cases contains detailed use cases of maintaining

work.

Software and Hardware Update and Maintenance

330 Administration and Operation

Finish maintaining of the Application Node‘s software always with intensive tests. The

maintained Application Node Image will become the image for other Application Nodes as

well.

Notice that all Application Nodes you will boot later on from that maintained

image will inherit all the changes you made. Therefore keep an eye on unwished

remnants i.e. unwished temporary files, notes, directories etc. Clean them up.

Finally shut down the Application node running in Maintenance mode i.e. in the example

do:

pa-rx300-1:~ # init 0

11.4.6 Step #3: Reverting the Maintenance Image

As stated in the ―Schematic overview‖ the last step in the Maintenance Cycle is to revert

the Maintenance Image:

Software and Hardware Update and Maintenance

Administration and Operation 331

1

`

os/Linux/FSC_50A*/

productive mode

generation 1

/FlexFrame/volFF/

Base Im

ag

es

root_img and

var_img of Node #n

root_img and

var_img of Node #2root_img and

var_img of Node #1

base_img/

root_img/

os/Linux/CST_50A*/

root_img and

var_img of Node #n

root_img and

var_img of Node #2root_img and

var_img of Node #1

base_img/

root_img/

os/Linux/MNT_50A*/

root_img of

Maintenance-Node

base_img/

root_img/

maintenance mode

generation 1’

productive mode

generation 2

Ro

ot

Imag

es

1.

2.

3.

The ―Maintenance Image‖ resulting of step #2 needs to be transformed into a ―Base

Image‖. This - i.e. creating a Base Image - can be done by help of the already known

ff_install_an_linux_images.sh command. Since a revert is needed the –r option

has to be used.

The reverting is required since only a ―Base Image‖ can be used to create a ―Root Image‖

(as known from the ―Understanding the Application Node Image‖ section) which is

needed to run productive Application Nodes on it.

Software and Hardware Update and Maintenance

332 Administration and Operation

Based on the maintained Linux Application Node Image (specified by -s option) the

ff_install_an_linux_images.sh command will spawn another Linux Application

Node Image (specified by -t option). You can assign systems to this Linux Application

Node Image by using ff_an_adm.pl and create Nodes by using the ff_new_an.sh

command. All of the Nodes running on it will benefit from the software maintenance you

made. This launch of ff_install_an_linux_images.sh command will copy the

entire ―Root Image‖ to a ―Base Image‖ as known.

Synopsis

ff_install_an_linux_images.sh –r -s <srcdir> -t <trg> \

[-v] [-d] [-i]

Example

(Remember the step #2 asked to shut down the used Application Node)

cn1:/ # cd /FlexFrame/volFF/os/Linux

cn1:/FlexFrame/volFF/os/Linux # ff_install_an_linux_images.sh –r \

-s MNT_5.1A00-001.SLES-10.x86_64/ -t CST_5.0A00-001.SLES-10.x86_64/

Notice that the ff_install_an_linux_images.sh command accepts

relative and absolute path names as arguments to the –s and –t option. You

may save some typing when you first change to the

/FlexFrame/volFF/os/Linux/ directory as shown in the example above.

It is worthwhile using a meaningful prefix for the directory name to keep track of

the images. A good start for choosing the name is to replace the original prefix

(here: MNT_) by CST_ indicating a ―Customer‖ image.

Consult the ff_image_types(8) manual page for a suggestive naming of the

arguments to the -s and -t options.

As mentioned earlier the SLES operating systems, which are used within a FlexFrame

environment, have been changed in a number of places to implement the concepts,

which make FlexFrame unique. In some situations the changes made during

Maintenance (Step #2) may conflict with FlexFrame‘s concepts. In order to keep

FlexFrame going smoothly the FlexFrame implementations have higher priority than the

changes made during Maintenance mode. This means: these changes made during

Software and Hardware Update and Maintenance

Administration and Operation 333

Maintenance mode will be replaced during reverting5. If possible the

ff_install_an_linux_images.sh command will point out to these conflicts:

cn1:/FlexFrame/volFF/os/Linux # ff_install_an_linux_images.sh –r \

-s MNT_5.1A00-001.SLES-10.x86_64/ -t CST_5.0A00-001.SLES-10.x86_64/

...

The following files were modified during maintenance. To ensure

FlexFrame's functionality they will be replaced now by their

FlexFrame specific versions. See "Administration and Operation

Guide" for details:

/etc/issue

/root/.ssh/authorized_keys

/etc/fstab

A copy of the maintenance version is available as

"<file>.maintenance".

...

Here you must compare for example /etc/fstab.maintenance (this is the version of

this file after leaving maintenance and before reverting) and the /etc/fstab (this is the

needed FlexFrame version of this file, installed after revert). You may have to merge

entries from the /etc/fstab.maintenance file into the /etc/fstab file.

Note: it is important to double-check the reverted image whether it contains the expected

changes: This is possible for example by simply comparing the just created ―Base Image‖

with the ―Root Image‖. If there are changes which were replaced by FlexFrame settings it

is needed to decide whether they can be inserted again. Please take notice of section

11.4.8 Maintaining Use Cases.

After reverting the source of the operation the ―Maintenance Image‖ seems to be

needless: it was the source for the revert and can be removed now. But in section 11.4.8

Re-Using the Maintenance Image on page 335 you will see how experienced

administrators might use it again.

11.4.7 Migrating remaining Application Nodes

Now, after testing and reverting the ―Maintenance Image‖ it is time to go productive with

all the Application Nodes.

The created ―Base Image‖ acts as a normal ―Base Image‖ i.e. it can now serve as a

source for other Application Node Images. All Application Nodes inherit the changes from

Maintenance Mode.

5 FlexFrame for SAP Infrastructure Consultants might seek advice in /opt/FlexFrame/lib/an.

Software and Hardware Update and Maintenance

334 Administration and Operation

To do this it is only necessary to use the well known administration commands. Since this

is not ―Maintenance Mode‖ specific – the procedure and all the commands are already

known - now only the example follows:

At first assign the created image to the Application Nodes by help of the ff_an_adm.pl

command and then create an image by help of the ff_new_an.sh command and boot.

Remember we had four Application Nodes in group sles10_x64 of the pp1 pool.

The first one – named pa-rx300-1 – was used for the Maintenance Cycle and is

currently down. Let us start with that node:

cn1:/ # ff_an_adm.pl --op os --ospath Linux/CST_5.1A00-001.SLES-

10.x86_64 \

--name pa-rx300-1

The ff_an_adm.pl command asked for it so let us launch the ff_new_an.sh

command:

cn1:/ # ff_new_an -n pa-rx300-1

Finally boot the Application Node:

cn1:/ # ff_ponpoff.pl --op power-on --name pa-rx300-1

All the other nodes of the pool group – named pa-rx300-2, pa-rx300-3 and pa-

rx300-4 – which are still running on the old image can be handled in a similar way.

Instead of powering them on they just need to be rebooted:

cn1:/ # ff_an_adm.pl --op os --ospath Linux/CST_5.1A00-001.SLES-

10.x86_64 \

--name pa-rx300-2

cn1:/ # ff_new_an -n pa-rx300-2

cn1:/ # ssh pa-rx300-2 init 6

cn1:/ # ff_an_adm.pl --op os --ospath Linux/CST_5.1A00-001.SLES-

10.x86_64 \

--name pa-rx300-3

cn1:/ # ff_new_an -n pa-rx300-3

cn1:/ # ssh pa-rx300-3 init 6

cn1:/ # ff_an_adm.pl --op os --ospath Linux/CST_5.1A00-001.SLES-

10.x86_64 \

--name pa-rx300-4

cn1:/ # ff_new_an -n pa-rx300-4

cn1:/ # ssh pa-rx300-4 init 6

Software and Hardware Update and Maintenance

Administration and Operation 335

You can do this of course in a more efficient way – the code above tries to show the basic

strategy.

The switch to the maintained image costs only a few minutes! In case of problems you

can switch back in the same way.

Please bear in mind: It is very important that all the Linux Application Nodes of

one pool group have the same configuration (software, patches) since each one

must be able to take over if another one fails. That is why all of the Linux

Application Nodes of a pool group must run on the same image.

In the course of time you might get a lot of Application Node Images in

/FlexFrame/volFF/os/Linux/. In order to keep track of their history you might use

the ―.itoc‖ file of the ―Base Image‖.

11.4.8 Re-Using the Maintenance Image

This section describes a time-saving extension to the ―Maintenance Cycle‖. You should

use it not before you have got some experience with the ―Maintenance Cycle‖ itself.

Reverting the ―Maintenance Image‖ as described in section 11.4.6 Step #3: Reverting the

Maintenance Image on page 330 ended with an apparently needless ―Maintenance

Image‖: it was the source of the operation. If you are satisfied with the results of the

Maintenance Cycle i.e. the migrated Application Nodes are running fine you can remove

it indeed.

But why it is not removed automatically?

Because it is possible to configure the Application Node (used to perform the

maintenance tasks) and to boot it from the ―Maintenance Image‖ again: this allows fixing

errors you found while testing the just generated images without starting a new

―Maintenance Cycle‖.

Be aware that only the same Application Node can be used to boot the ―Maintenance Image‖ again. You can use the ff_an_adm.pl command to get the needed information.

Launch it with no option (output trimmed):

Software and Hardware Update and Maintenance

336 Administration and Operation

cn1:/ # ff_an_adm.pl

--ospath <path to os image>

The path of the os image to be used.

Currently available os image pathes are:

Linux/CST_5.1A00-001.SLES-10.x86_64

Linux/FSC_5.1A00-001.SLES-11.x86_64

Linux/MNT_5.1A00-001.SLES-10.x86_64 (maintenance only (pa-rx300-

1))

Linux/TST_5.1A00-002.SLES-10.x86_64

It reveals the name of the only Application Node that can use this ―Maintenance Image‖:

the pa-rx300-1 node.

To assign the ―Maintenance Image‖ again to this Node use the well known ff_an_adm.pl command:

cn1:/ # ff_an_adm.pl --op os --ospath Linux/MNT_5.1A00-001.SLES-

10.x86_64 \

--name pa-rx300-1

After that create the image for the Application Node:

cn1:/ # ff_new_an.sh –B -n pa-rx300-1

Notice that ff_new_an.sh needs to be called with the ―-B‖ option to only configure boot

information for this node because the image is already configured for this node and

cannot be reconfigured.

Finally, reboot the maintenance node:

cn1:/ # ssh pa-rx300-1 init 6

or switch it on if it is powered off:

cn1:/ # ff_ponpoff.pl --op power-on --name pa-rx300-1

After some time the node will come up running the ―Maintenance Image" and you can

perform further maintenance tasks and fix errors you found while testing.

As soon as you have finished maintaining this image, shut down the node and revert the

―Maintenance Image‖ to a new ―Base Image‖ again as described in section 11.4.6 Step

#3: Reverting the Maintenance Image on page 330. Note that you can not use the same

target directory name for the revert operation (it already exists.) Use a different directory

name or remove the old one if you want to use the same target directory name as before.

Software and Hardware Update and Maintenance

Administration and Operation 337

Now it is possible to have one node for maintaining the Maintanance Image and

another node to test the reverted image in production mode because the

Maintenance and Base Images reside inside two different directory trees.

You will not have to switch one Application Node forth and back between

Maintenance Image and Base Image again and again. You still need to shut

down the maintenance Application Node before reverting to ensure a consistent

image state.

In theory it is also possible to run one Application Node always in ―Maintenance

Mode‖ in order to keep it up to date with security updates and create a Base

Image from it once in a while as needed.

However, it is strongly recommended to remove the Maintance Image once the

maintained Image has been tested intensively in production mode and perform

further maintenance tasks by starting over with a complete new Maintance Cycle

based on the Maintained Images, or in other words: perform the whole 1-2-3

procedure again on the already customized Images.

Following this procedure closely allows keeping all maintained image versions

and rolling back to working images if something failed during a Maintenance

Cycle.

You can keep track of all changes by just looking into the mentioned ―.itoc‖ file

of the ―Base Images‖.

11.4.9 Maintaining Use Cases

We would like to point out that changes due to 3rd party software or patches or

customer modifications may have heavy (negative) side effects to the FlexFrame

functionality.

Installing third party software on Application Nodes may cause functional

restrictions and other malfunctions of the FlexFrame system software. Fujitsu

shall have no liability whatsoever whether of a direct, indirect or consequential

nature with regard to damages caused by the third party software or its

erroneous installation.

Installation of software or other maintaining work as described in this section has

to be done by help of a Maintenance Cycle as described in section 11.4

Maintenance of Application Nodes - Software on page 318. Not following this

procedure i.e. maintain software when running in normal operation may harm the

entire Application Node group. Resulting malfunctions are not an issue of

FlexFrame.

Software and Hardware Update and Maintenance

338 Administration and Operation

11.4.9.1 Service Packs

It is not allowed to install any Service Packs on Application Node images manually. The

only way to obtain a new service pack for Application Nodes is to install new Application

Node images delivered from Fujitsu. Service Packs contain a lot of changes that can

have influences to the FlexFrame functionality. Service Packs can also contain initial

configuration files that will substitute the FlexFrame configurations.

11.4.9.2 Updating/Installing a New Linux Kernel

This section describes how to install an additional (alternative) Linux kernel into an

existing Application Node image.

The netboot configuration /tftpboot/pxelinux.cfg/<hw class>_<pool name>_<image tree name>

will determine which Linux kernel and initial ramdisk need to be used for a subset of

Application Nodes. The image tree name of original images delivered by Fujitsu begins with FSC_ and follows the scheme

FSC_<FF version>.<OS version>.<OS architecture>.

11.4.9.2.1 Software Stage

To install and update any software packages in FlexFrame, it is useful to mount your NAS

system or jukebox or any other software stage on a local mount point at your Control

Node. The same software stage must be mounted on an Application Node which is

appointed to install the new kernel-RPMs during the maintenance cycle.

For example, your software stage on the Control Node is /FlexFrame/volFF/FlexFrame/stage and the new kernel to install is

2.6.16.60-0.63.1.

Create a subdirectory for the delivered software:

control1: # cd /FlexFrame/volFF/FlexFrame/stage

control1:/FlexFrame/volFF/FlexFrame/stage # mkdir -p SLES10/kernel

Copy the delivered software packages to the software stage.

Kernel RPM:

control1: # cp <somewhere>/kernel-smp-2.6.16.60-0.63.1.x86_64.rpm

/FlexFrame/volFF/FlexFrame/stage/SLES10/kernel

Kernel kdump:

control1: # cp <somewhere>/kernel-kdump-2.6.16.60-0.63.1.x86_64.rpm

/FlexFrame/volFF/FlexFrame/stage/SLES10/kernel

Software and Hardware Update and Maintenance

Administration and Operation 339

Kernel source:

control1: # cp <somewhere>/kernel-source-2.6.16.60-0.63.1.x86_64.rpm

/FlexFrame/volFF/FlexFrame/stage/SLES10/kernel

With SLES 10 (Service Pack 1 and later) Novell has introduced the Partner

Linux Driver Process (PLDP). There is no need to download drivers for the

update kernel. The existing drivers are reused for the new kernel.

11.4.9.2.2 Installing a New Linux Kernel

Logon to the appointed Application Node running the maintenance cycle.

Check the currently installed Linux kernels:

pa-rx300-1:/ # rpm -qa | grep kernel

kernel-kdump-2.6.16.60-0.59.1

kernel-smp-2.6.16.60-0.59.1

kernel-source-2.6.16.60-0.59.1

Mount the software stage:

pa-rx300-1: # mkdir /mnt/stage

pa-rx300-1: # mount -t nfs <filer>:/vol/volFF/FlexFrame/stage /mnt/stage

To install the new kernel, execute the following command:

pa-rx300-1: # cd /mnt/stage/SLES10/kernel

pa-rx300-1:/mnt/stage/SLES10/kernel # export YAST_IS_RUNNING=instsys

pa-rx300-1:/mnt/stage/SLES10/kernel # rpm -ihv kernel-smp-2.6.16.60-

0.63.1.x86_64.rpm

pa-rx300-1:/mnt/stage/SLES10/kernel # rpm –ivh kernel-kdump-2.6.16.60-

63.1.x86_64.rpm

pa-rx300-1:/mnt/stage/SLES10/kernel # rpm –ivh kernel-source-2.6.16.60-

63.1.x86_64.rpm

‗YAST_IS_RUNNING=instsys‘ is used to skip updating the bootloader

menu.

11.4.9.2.3 New initrd, Kernel and Netboot Configuration

Before booting the maintained Application Node image, it is necessary to create a new

initrd, install the initrd and kernel in /tftpboot and update the PXE configuration file.

All these steps will be automatically performed on the Control Node by the command:

control1:~ # ff_new_an.sh -B -n pa-rx300-1

Software and Hardware Update and Maintenance

340 Administration and Operation

Internally, this will call the ff_mkinitrd.sh utility, update the storage export

configuration and create a new PXE configuration file.

The latest kernel will be automatically used in the future. If you like to revert to a previous

kernel, follow the instructions described below in Chapter 11.4.9.2.5.

Now reboot the Application Node and verify that it is working properly:

pa-rx300-1:~ # reboot

11.4.9.2.4 Partner Linux Driver Process

The Partner Linux Driver Process (PLDP) replaces the traditional method of driver

provisioning for SLES 10 (Service Pack 1 and later). The most benefit of this process is

the support for reusing existing drivers for errata kernels. There is no need to download

drivers for those kernels.

Nevertheless if updated driver packages are released (e.g. a new version) these

packages can be installed on Control Node or Application Nodes like an ―installable

patch‖.

Depending on a special configuration the document ―Novell Partner Linux Driver Process

for PRIMERGY servers‖ describes the update procedure in detail. This document is

available at URL:

http://ts.fujitsu.com/support/downloads.html

11.4.9.2.5 Switching Between Different Kernels

If you have multiple kernel releases installed in the Application Node image and like to

switch to a different kernel, change the the symlink …/root_img/boot/vmlinuz and

run ff_new_an.sh -B -n <node name> again.

This will change the PXE configuration files for this image and hardware class and will

affect all Application Nodes of the same hardware class using this image.

The initrd does not need to be re-created nor does the initrd-symlink need to be

changed because it always contains all kernel releases found in an image.

11.4.9.3 ServerView Update

The ServerView update comprises the components ―Agents‖ and ―RAID Manager‖ on the

Application Node.

The components for the Application Nodes are the same as for the Control

Nodes.

Software and Hardware Update and Maintenance

Administration and Operation 341

README files or other installation instructions are available in the download

area.

Always use the instructions that come with the downloaded software!

Mount your software stage:

pa-rx300-1:/mnt # mkdir /mnt/stage

pa-rx300-1:/mnt # mount -t nfs <filer>:/vol/volFF/FlexFrame/stage

/mnt/stage

11.4.9.3.1 ServerView Agents

Go to your software stage directory. There you will find the following RPMs: srvmagt-agents-*.i386.rpm

srvmagt-mods_src-*.i386.rpm

srvmagt-eecd-*.i386.rpm

ServerViewConnectorService-*.386.rpm

And the associated installation script: srvmagt.sh.

Upgrade the ServerView Agents by using srvmagt.sh:

pa-rx300-1:~ # cd /mnt/stage

pa-rx300-1:/mnt/stage # sh srvmagt.sh install

11.4.9.3.2 ServerView RAID Manager

Go to your software stage directory. There you will find the

ServerView_RAID-*.i386.rpm RPM.

Upgrade the ServerView RAID Manager by:

pa-rx300-1:~ # cd /mnt/stage

pa-rx300-1:/mnt/stage # rpm -U ServerView_RAID-*.i386.rpm

The ServerView update is now finished.

11.4.9.4 Upgrading the Application Software

The rules for installation of application software (SAP/Oracle) apply here as well.

An Application Node should be available exclusively for upgrading the application

software, so that the physical host name can be set to be identical with the virtual host

name of the application.

Software and Hardware Update and Maintenance

342 Administration and Operation

Provided this has been done, upgrading can be carried out in accordance with the

software vendor's standard guides.

All parts of the application software must reside on the network files system mounted

from the NAS system.

For more details see the SAP Installation Guides.

11.4.9.5 Updating RPM Packages on an Application Node

To update the SLES RPM packages you can use the original Novell tools to perform an

online update.

11.4.9.5.1 Setup a patch server

It is not recommended to connect your FlexFrame environment directly to the Internet to

download patches or updates. Therefore it is advisable to set up an own patch server,

which provides the necessary update packages. See: http://www.novell.com/de-

de/linux/smt/ and http://www.novell.com/communities/files/appnote-2009-12-10-v1_7.pdf

for hints.

Provided that this patch server is accessible for the Application Node in maintenance

mode, a normal online update with YaST can be performed.

11.4.9.5.2 Connect the patch server to the FlexFrame environment

There are several possibilities to connect your local patch server to your FlexFrame

environment. It‘s up to your internal network and security policy, which way is the right

way for you.

Some of them are described in http://www.novell.com/communities/files/appnote-2009-

12-10-v1_7.pdf

It is possible to setup an ‗External Connectivity‘ on the Control Node to connect your

patch server to the FlexFrame environment. On the Application Node in maintenance

mode you have to setup a ‗SSH tunnel‘. Or you can copy the data from your patch server

to an ‗USB disk‘ and connect this disk directly to the Application Node in maintenance

mode. Or copy the data to the local disk of the Control Node or to a directory on the filer

and share (NFS, SMB/CIFS, HTTP(S) or FTP) this directory for the Application Node in

maintenance mode.

11.4.9.5.3 Prepare the Online Update

To prepare the online update you have to define the installation source. That‘s the way,

how you provide the update packages, i.e. your patch server

On an SLES10 Application Node:

YaST Software Installation Source Add Media Type

Software and Hardware Update and Maintenance

Administration and Operation 343

See section 8.3.3 ―Selecting the Installation Source‖ from

http://www.novell.com/documentation/sles10/sles_admin/data/sec_yast2_sw.html#sec_y

ast2_sysconfig_medium for more information.

On an SLES11 Application Node:

YaST Software Software Repositories Add Media Type

See Section 9.3 ―Managing Software Repositories and Services‖

http://www.novell.com/documentation/sles11/book_sle_deployment/data/sec_y2_sw_inst

source.html#sec_y2_sw_instsource for more information.

The Media Type strongly depends on the way you provide the update packages (see

section 11.4.9.5.2 ‖Connect the patch server to the FlexFrame environment‖ on page

342).

11.4.9.5.4 Perform the Online Update

Now you can perform the online update:

YaST Software Online Update

For SLES 10 see section 8.3.5 ―YaST Online Update‖ from

http://www.novell.com/documentation/sles10/sles_admin/data/sec_yast2_sw.html#sec_y

ast2_sysconfig_onupdate for more information.

For SLES11 see section 1.0 ―YaST Online Update‖ from

http://www.novell.com/documentation/sles11/book_sle_admin/data/cha_onlineupdate_yo

u.html#cha_onlineupdate_you for more information.

If the online update has installed a new kernel, it is necessary to create a new

initrd and update the netboot configuration, before you can reboot the

maintained Application Node.

All these steps will be automatically performed on the Control Node by the command:

control1:~ # ff_new_an.sh -B -n pa-rx300-1

Internally, this will call the ff_mkinitrd.sh utility, update the storage export

configuration and create a new PXE configuration file.

11.4.9.6 Updating vmware-open-vm-tools

If an update of the vmware-open-vm-tools is necessary, the needed packages will be part

of the current FlexFrame Service DVD.

Software and Hardware Update and Maintenance

344 Administration and Operation

On this Service DVD in directory ‗vmware/SLES1<x>-SP<y>‘ you will find the following

RPMs:

vmware-open-vm-tools-common-*.x86_64.rpm

vmware-open-vm-tools-nox-*.x86_64.rpm

primergy-vmware-kmp-*.x86_64.rpm

The following example will show the update for an Application Node running SLES10.

On the Control Node create a subdirectory for the delivered software:

control1: # cd /FlexFrame/volFF/FlexFrame/stage

control1:/FlexFrame/volFF/FlexFrame/stage # mkdir -p SLES10/vmware

Copy the delivered software packages to the software stage.

control1: # cp /media/dvd/vmware/SLES10-SP3/*

/FlexFrame/volFF/FlexFrame/stage/SLES10/vmware

Log on to the Application Node in Maintenance and perform an rpm update:

pa-rx300-1:~ # cd /mnt/stage/SLES10/vmware/

pa-rx300-1:/mnt/stage/SLES10/vmware # rpm –Uhv primergy-vmware-kmp-

*.x86_64.rpm vmware-open-vm-tools-common-*.x86_64.rpm vmware-open-vm-

tools-nox-*.x86_64.rpm

After installing the vmware KMP package, it is necessary to create a new initrd

before you can use the image for a virtual Application Node.

All these steps will be automatically performed on the Control Node by the command:

control1:~ # ff_new_an.sh -B -n pa-rx300-1

Internally, this will call the ff_mkinitrd.sh utility, to build a new initrd file with correct

linkage for the vmware drivers.

11.4.9.7 Upgrading the Application Software

The rules for installation of application software (SAP/Oracle) apply here as well.

An Application Node should be available exclusively for upgrading the application

software, so that the physical host name can be set to be identical with the virtual host

name of the application.

Software and Hardware Update and Maintenance

Administration and Operation 345

Provided this has been done, upgrading can be carried out in accordance with the

software vendor's standard guides.

All parts of the application software must reside on the network files system mounted

from the NAS system.

For more details see the SAP Installation Guides.

11.4.9.8 Updating RPM Packages on an Application Node

To update the SLES RPM packages you can use the original Novell tools to perform an

online update.

11.4.9.8.1 Setup a patch server

It is not recommended to connect your FlexFrame environment directly to the Internet to

download patches or updates. Therefore it is advisable to set up an own patch server,

which provides the necessary update packages. See: http://www.novell.com/de-

de/linux/smt/ and http://www.novell.com/communities/files/appnote-2009-12-10-v1_7.pdf

for hints.

Provided that this patch server is accessible for the Application Node in maintenance

mode, a normal online update with YaST can be performed.

11.4.9.8.2 Connect the patch server to the FlexFrame environment

There are several possibilities to connect your local patch server to your FlexFrame

environment. It‘s up to your internal network and security policy, which way is the right

way for you.

Some of them are described in http://www.novell.com/communities/files/appnote-2009-

12-10-v1_7.pdf

It is possible to setup an ‗External Connectivity‘ on the Control Node to connect your

patch server to the FlexFrame environment. On the Application Node in maintenance

mode you have to setup a ‗SSH tunnel‘. Or you can copy the data from your patch server

to an ‗USB disk‘ and connect this disk directly to the Application Node in maintenance

mode. Or copy the data to the local disk of the Control Node or to a directory on the filer

and share (NFS, SMB/CIFS, HTTP(S) or FTP) this directory for the Application Node in

maintenance mode.

11.4.9.8.3 Prepare the Online Update

To prepare the online update you have to define the installation source. That‘s the way,

how you provide the update packages, i.e. your patch server

On an SLES10 Application Node:

YaST Software Installation Source Add Media Type

Software and Hardware Update and Maintenance

346 Administration and Operation

See section 8.3.3 ―Selecting the Installation Source‖ from

http://www.novell.com/documentation/sles10/sles_admin/data/sec_yast2_sw.html#sec_y

ast2_sysconfig_medium for more information.

On an SLES11 Application Node:

YaST Software Software Repositories Add Media Type

See Section 9.3 ―Managing Software Repositories and Services‖

http://www.novell.com/documentation/sles11/book_sle_deployment/data/sec_y2_sw_inst

source.html#sec_y2_sw_instsource for more information.

The Media Type strongly depends on the way you provide the update packages (see

section 11.4.9.8.2 ‖Connect the patch server to the FlexFrame environment‖ on page 1).

11.4.9.8.4 Perform the Online Update

Now you can perform the online update:

YaST Software Online Update

For SLES 10 see section 8.3.5 ―YaST Online Update‖ from

http://www.novell.com/documentation/sles10/sles_admin/data/sec_yast2_sw.html#sec_y

ast2_sysconfig_onupdate for more information.

For SLES11 see section 1.0 ―YaST Online Update‖ from

http://www.novell.com/documentation/sles11/book_sle_admin/data/cha_onlineupdate_yo

u.html#cha_onlineupdate_you for more information.

If the online update has installed a new kernel, it is necessary to create a new

initrd and update the netboot configuration, before you can reboot the

maintained Application Node.

All these steps will be automatically performed on the Control Node by the command:

control1:~ # ff_new_an.sh -B -n pa-rx300-1

Internally, this will call the ff_mkinitrd.sh utility, update the storage export

configuration and create a new PXE configuration file.

11.4.9.6 Updating vmware-tools

To update the vmware-tools to a newer version, we recommend to use our provided

Application Node Images containing this new version of vmware-tools.

To install a newer version of vmware-tools during a FlexFrame maintenance cycle please

refer to the OSP documentation of VMware, which can be found at

http://packages.vmware.com

Software and Hardware Update and Maintenance

Administration and Operation 347

Installation of vmware-tools from an ESXi host is not permitted.

Log on to the Application Node in Maintenance, copy the downloaded vmware-tools

packages to this Application Node and follow the instructions to install the vmware-tools

on SLES11 or SLES10 as described in the VMware OSP documentation mentioned

above.

11.5 Maintenance of the Application Nodes - Hardware

11.5.1 Changing BIOS Settings for Netboot

For detailed information on setting BIOS parameters refer to the hardware

documentation. Look at some of the configuration examples below. The exact procedure

may differ depending on the BIOS version of your hardware. For newer server models

refer to the corresponding documentation called HW Characteristics (HW-Steckbrief) – in

these documents (delivered with the Service CD of a FlexFrame for SAP version and also

on the FTS Extranet on https://partners.ts.fujitsu.com/com/service/ps/Software/flexframe)

the server specific properties relevant for installing the server within a FlexFrame for SAP

environment are collected.

11.5.2 Replacing a Network Card – Application Node

There are two cases for replacing a network card on an Application Node:

The card has failed and you want to replace it by a card of same type.

The network card is replaced by a card of another technology, e.g. 1GBit to 10GBit

migration or vice versa.

If a network card on an Application Node fails and is replaced, further actions are

necessary to ensure proper operation.

During their boot processes, PRIMERGY nodes need to retrieve some information

through lower level protocols which are based on hardware addresses (MAC addresses).

The PRIMERGY's BIOS utilizes DHCP for that purpose.

The MAC address is the built-in, physical address which is unique to that network card.

Since there are two network cards configured for booting you may not even notice the

MAC address change of a single one, because the other card will be used for booting.

Once the server is up, operation is normal. For full redundancy – even for booting – we

recommend to adjust the configuration accordingly.

Software and Hardware Update and Maintenance

348 Administration and Operation

On PRIMERGY blade systems, use the management blade to determine the server

blade MAC addresses.

On PRIMERGY non-blade systems, use BIOS or network card labels to determine

the MAC addresses.

To change the MAC addresses for booting, use the command below. Replace the node

name and MAC addresses of the sample with your Application Node's name and noted

addresses:

control:~ # ff_an_adm.pl --op mac --name appnode5

--mac 00:e0:00:c5:19:41,00:e0:00:c5:19:41

MAC addresses changed.

The program changed the MAC addresses of the node within DHCP configuration and

LDAP. No further intervention is necessary.

Some software licenses are derived from information based on server specific

numbers, such as MAC addresses of network cards. On Linux, double-check

your SAP licenses after you have replaced a network card.

If a network card on an Application Node is replaced for technology change reasons the

following steps are necessary:

1. Remove the Application Node from the FlexFrame landscape with ff_an_adm.pl --op rem

2. Remove the 1GB network cables of the Application Node

3. Remove the 1GB NIC cards from the Application Node

4. Plug in the 10GB NIC cards into the Application Node

5. (see corresponding HW-Characteristics Quickguide)

6. Install the Application Node with ff_an_adm.pl

(for details see the corresponding HW-Characteristic Quickguide)

Example:

ff_an_adm.pl --op add --name sno2rx1 --type RX600S4 --pool

pool2 --group default ---swgroup 1 --ospath Linux/FSC_5.1A00-

005.SLES-11.x86_64 --host 25 --mac

00:1b:21:24:4f:5d,00:1b:21:24:4f:5c -–10gbit

7. Plug in the network cables according to the wiring output given by ff_an_adm.pl

8. Call ff_new_an.sh to generate an image for the new Application Node

9. Boot the Application Node

Software and Hardware Update and Maintenance

Administration and Operation 349

11.5.3 Replacing a Switch Blade

There are two cases for replacing a switch blade:

In case of an HW error you have to replace one switch blade.

You can change the type of the used switch blade. In this case you have to replace

both switch blades.

In the first case (HW error) perform the following actions:

1. You have to replace the switch blade as described in the HW description of the

BX Server.

2. Create the configuration file for the switch blade with the function ff_bx_cabinet.pl –-op swb-config. See chapter 7.7.7 Getting Switch Blade

Configuration for a detailed description of the function.

3. Upload configuration to switch as displayed by ff_bx_cabinet.pl –-op swb-

config.

In the second case (change the type) perform the following actions:

1. Deactivate the usage of each server blade.

2. Remove the network cables from the switches.

3. Replace both switches.

4. Plug in the network cables.

5. Call ff_bx_cabinet_adm.pl --op swb-change to update LDAP Database.

6. Create the configuration file for the switch blade with the function

ff_bx_cabinet.pl –-op swb-config. See chapter 7.7.7 Getting Switch Blade

Configuration for a detailed description of the function.

7. Upload configuration to switch as displayed by ff_bx_cabinet.pl –-op swb-

config.

8. Now you can use the server blades again.

11.5.4 Replacing Power Control Hardware

This section describes what to do after the replacement of power control hardware,

depending on your Application Nodes and Control Nodes hardware.

For instructions on replacing the power control hardware, please refer to the

documentation delivered with your hardware.

Software and Hardware Update and Maintenance

350 Administration and Operation

11.6 Maintenance of ESXi Servers

11.6.1 BIOS Updates

To update the BIOS of a PRIMERGY server that is used as an ESXi server, use the

standard BIOS update procedure for PRIMERGY servers based on a bootable USB

memory stick. Please note that the method using ASPs (Autonomous Primergy Support

Packages) is not applicable in this case.

11.6.2 Replacing a Network Card - ESXi Server

If a network card of an ESXi server fails and is replaced, further actions are necessary to

ensure proper operation.

PRIMERGY servers used as ESXi servers in FlexFrame get the address used for ESXi

management from a DHCP server running on the FlexFrame Control Center. The DHCP

server determines the correct IP address based on the requester's MAC address.

The MAC address is the built-in, physical address which is unique to that network card.

As the MAC address is changed after replacing a network card, we recommend adjusting

the configuration accordingly,

● On PRIMERGY blade systems, use the management blade to determine the server

blade MAC addresses.

● On PRIMERGY non-blade systems, use BIOS or network card labels to determine

the MAC addresses.

To change the MAC addresses for DHCP, use the command below. Replace the node

name and MAC addresses of the sample with your ESXi server's name and noted

addresses:

control:~ # ff_esx_adm.pl --op mac --name esxnode5

--mac 00:e0:00:c5:19:41,00:e0:00:c5:19:42

The program changes the MAC addresses of the node within DHCP configuration and

LDAP. No further intervention is necessary.

11.6.3 ESXi Updates and Patches

VMware provides patches and updates for ESXi servers. While updates are for

maintenance releases, patches address critical security fixes or urgent bug fixes. An

update or patch can include a new build of firmware, an update of VMware Tools, or an

update of the vSphere Client.

Please note that patches and updates of VMware Tools cannot be applied using the

standard VMware procedures. The FlexFrame Application Node images contain the

Software and Hardware Update and Maintenance

Administration and Operation 351

VMware Tools operating system specific packages (OSPs). For a description on how to

upgrade these tools refer to section 11.4.9.6 on page 343.

VMware provides several tools for installing updates and patches to ESXi hosts:

● esxcli - command-line utility

● vCenter Update Manager - automates patching and updates

The esxcli utility is part of the vSphere CLI, which is available on the FlexFrame Control

Nodes.

For more information on these tools as well as further usage hints and recommendations

please refer to the appropriate VMware documentation.

11.7 Maintenance of Other FlexFrame Components

11.7.1 NetApp Storage

ONTAP Patches

Before applying a new patch to the Filer's ONTAP make sure that it is supported for the

FlexFrame concept. In doubt contact the Fujitsu support.

Download the tar file(s) for the new ONTAP onto one of your Control Nodes.

Create a temporary directory e.g. in /FlexFrame/volFF/FlexFrame/stage/ontap

(create the directory if it does not exists) and extract the tar file there.

Proceed as described in the manual or README information of that patch.

Before you activate the new patch by rebooting the Filer, make sure that no SAP system

or Database is busy. You do not need to reboot the Application Nodes, however we

recommend double-checking the mount points after the Filer is up again. If an Application

Node should get stuck, reboot it.

11.7.2 EMC Storage

DART Patches for Celerra Data Movers

Download the patch (default name nas.exe) into the directory /nas/dos/bin on the

Celerra. If you choose the default name nas.exe for your patch, you have to do nothing.

If you take a different name for it, you will have to modify the files

/nas/dos/slot_<ID>/boot.bat with this name, where ID is the data mover ID,

e.g., 2 for server_2 or 3 for server_3.

Software and Hardware Update and Maintenance

352 Administration and Operation

Then you have to reboot all affected data movers by the reboot command on the

Celerra control station:

server_cpu ALL –reboot now

server_version ALL (to see, if the new patched OS version is

active).

Before activating the new patch by rebooting the data movers, make sure that no SAP

system or database is busy. You do not need to reboot the Application Nodes, however

we recommend double-checking the mount points after the data movers are up again. If

an Application Node should get stuck, reboot it.

11.7.3 Cisco Switches and Switch Blades

11.7.3.1 Firmware Update

It may be possible that an update of the switch firmware is necessary. For Cisco Switches

you find a detailed description in the corresponding Cisco software configuration guides

at http://www.cisco.com. For Switch Blades have a look at the corresponding user

manual.

11.7.3.2 Backup of Switch Configurations

The switch configuration may be different to that at the time of installation due to the

operation of a FlexFrame landscape. New Application Nodes and switches may be

added, ports configured and so on. For system backup, there are a lot of well known

products and programs.

To simplify backing up the configuration of all switches used within the FlexFrame landscape the program /opt/FlexFrame/bin/ff_save_switch_config.pl should

be used. It is designed to be run by cron(8) and recommended to call it once a day.

The program is able to backup all LDAP registered switches and switch blades or the

switch or switch blade given by --name. The program connects to a switch or switch

blade and copies the running or startup configuration via TFTP to a Control Node into /tftpboot/<switch name>.config (if --multi is used to

/tftpboot/saved_switchconfigs/<switch host name>.config.YYY-MM-DD).

Synopsis

ff_save_switch_config.pl [--silent] [--running]

[--name <switch name>]

[--ip <control node ip>]

[--multi] [--max_versions <count>]

Software and Hardware Update and Maintenance

Administration and Operation 353

Options

--silent

The silent mode suppresses any message, even error messages.

--running

In normal case the configuration, which the switch reads on booting, is backed up, the so called startup-config. Using the --running option the current running switch

configuration is used instead of startup-config. For a BX600 switch blade of type

1GbE-10/6-Q (has six external ports) the running switch configuration is backed up

only.

--name <switch name>

Backup only the switch with given name. Keep in mind to use the name known at

LDAP database. It may differ from current switch name if it was not synchronized.

--ip <control node ip>

Use control node with given control lan ip address to save configurations to. In

normal case the program is able to detect the control node with running netboot_srv

service itself. If it is not able to detect the active node or cannot determine the

control lan ip of the control node control, the program requests to use the --ip

option.

--multi

Add a date string (ISO 8601 date format) to configuration file names to store more

than one configuration file of a switch. The files are stored in /tftpboot/saved_switchconfigs.

--max_versions <count>

This option is only used with option --multi and restricts the count of saved switch

configuration files to the latest <count> versions. Count defaults to: 10

11.7.3.3 Restore of Switch Configurations

To restore a backed-up switch configuration clear the actual configuration on the switch

then use copy and paste for the switch config commands. Don‘t forget to copy the

restored new running switch configuration on the switch as startup-configuration for the

next reboot.

Be aware, the backed-up switch configuration may not be the last effective

switch configuration because of changes applied after backup.

Software and Hardware Update and Maintenance

354 Administration and Operation

You can get an actual configuration according to LDAP

for switch blades using

ff_bx_cabinet_adm.pl --op swb-config ... (see 7.7.7)

for switch groups using

ff_swgroup_adm.pl --op gen-config ... (see 9.4)

11.7.4 DNS Servers

If the customer wants to change IP addresses of his DNS servers he also has to change

the corresponding FlexFrame configurations:

The DNS servers of each pool can be modified with the command ff_pool_adm.pl (see

4.5).

The DNS servers of the Control Center can be modified by editing the ―forwarders‖ in

the file /etc/named.conf . Originally the DNS Servers of the Control Center came into

FlexFrame by specifying them in the ―Miscellaneous Control Center Settings‖ in the

Management Tool.

Any other DNS server (e.g. explicitly indicated DNS servers in hardware configuration

files), which manually has been entered by the customer, is not under FlexFrame control

and it is the customers responsibility to modify these addresses at the right place.

11.7.5 Third Party Products

If the customer needs to install third party software on the Application Nodes, the maintenance procedures described in this manual must be followed. Installation of

software or patches may have to be performed again after installing a maintenance

image shipped by Fujitsu.

If you have installed additional 3rd

party software to your FlexFrame landscape,

please make sure to back up this software components and their configuration

data for later restore. Please consider: If you install a new FlexFrame image, the

whole file system and directories are overwritten by a complete new root file

system image. Afterwards, any 3rd

party software and configurations have to be

re-installed or restored.

Installing third party software on Control Nodes or Application Nodes may cause

functional restrictions and other malfunctions of the FlexFrame system software.

Fujitsu shall have no liability whatsoever whether of a direct, indirect or

consequential nature with regard to damages caused by the third party software

or its erroneous installation.

Software and Hardware Update and Maintenance

Administration and Operation 355

11.8 MyAMC Agents

For information on the MyAMC Agents please refer to the "myAMC.FA_Agents -

Installation and Administration" manual, chapter 2 "First Steps". Here you will find

detailed information on installation requirements, installation, configuration, the FA web

interface and the domain manager.

11.9 Project specific changes

As mentioned in chapter 2 ―FlexFrame Architecture‖ the FlexFrame solution consists of

both hardware and software. Most of the components are seen as an appliance: they

have well defined purposes and must work together in order to achieve a smoothly

running FlexFrame landscape. Normally only a well defined set of software and hardware

(see section 2.2 ―Hardware and Software‖) is allowed.

However there might be the need to change or modify components, or configuration

details in order to satisfy project specific requirements.

These special changes – project specific changes - mean that your configuration is

outside the FlexFrame standards and lead to a project specific FlexFrame.

These project specific changes may affect the operation of the entire FlexFrame

landscape. An example: Consider you run a SFG (‗Request for Special Release‘) for a

server which is unsupported in standard FlexFrame landscapes. Additionally you

changed some configuration in order to use some nice features of this server. Everything

runs fine. But: You have always to remember these changes. You have to tell the support

team (or other professionals working on this FlexFrame) about this. Without knowing

these project specific modifications a patch installation or upgrade installation may fail or

lead to unsatisfying results.

In order to keep track of these project specific changes a ―Project Activities

Communication Interface‖ has been established. It creates a communication channel

from human beings to human beings (and to FlexFrame commands).

The most important command of this interface is ff_projectinfo.sh which will be

introduced in the following chapter.

11.9.1 The ff_projectinfo.sh command

Name

ff_projectinfo.sh - add entry to projectinfo database file

Synopsis

ff_projectinfo.sh -n <name> \

-t <text> | -r <text file> \

Software and Hardware Update and Maintenance

356 Administration and Operation

[-c <customer specific SFG#> ] \

[-s <severity> ] \

[-C ] \

[-v] [-d]

-n: (name)identify yourself (email address preferred).

Mandatory!

-t: (text) Text to be added (if it is a short one)

-r: (read text file) Text to be added will be read out of

this file

-c: (custom) If a custom specific Special Release

(Sonderfreigabe) has been applied this option has to

name the SFG number

-s: (severity) classification of project specific change

where severity = ‗info‘ or ‗warning‘ or ‗error‘

-C: (CLEAR) change state of projectinfo database file to

READ

-v: (verbose)

-d: (debug)

Description

This script and the projectinfo database file are for use by FlexFrame certified

personnel only.

The objective of the project activities interface is to store and share information about

project specific activities at the FlexFrame landscape.

The ff_projectinfo.sh command is part of the project activities interface for the

FlexFrame solution. It helps to establish a reliable communication between the

professionals. This interface comprises

· this ff_projectinfo.sh command

· the FlexFrame projectinfo database file

· the ff_projectinfo_detector.sh

The ff_projectinfo.sh command serves as the only interface that creates or fills the

FlexFrame projectinfo database file.

Software and Hardware Update and Maintenance

Administration and Operation 357

Options

-n <name>

This option is mandatory: you need to identify yourself. The preferred argument is your

email address. If you want to use a name containing spaces you have to embed it into a

pair of double quotes.

-t <text>

Choose this option if the information you want to add is a short one and can be typed in

the command line. You will have to embed the text into a pair of double quotes (if it is not

only a single word which seems unlikely). For longer texts use the -r option. But: Usage

of either -t or -r option is mandatory.

-r <text file>

Choose this option if you want to add information you prepared previously, or which is

simply too long to be used at command line with the -t option. But: Usage of either -t or

-r option is mandatory. Note: You can add any text you like as long as file(1)

recognizes it as "text". Do not try to add core, dump, tar, files etc.

-c <custom specific Special Release number>

If a custom specific Special Release (Sonderfreigabe) has been applied to the system

this option is mandatory. It has to name the SFG number which you find in the Special

Release documents. It has typically a SFG-20??-??? appearance.

-s <severity>

This option is dedicated to classify the information you add: It helps the professionals

working later on the system to understand the significance resp. severity. The severity

option accepts the following values: info or warning or error. Without giving an explicit

severity a severity of info will be assumed.

-C

This option clears the default unread state of the projectinfo database file i.e. it sets

the state of the projectinfo database file to read.

Explanation: The projectinfo database file can have two states: unread or read.

Each appending of information (i.e. each successful call of the ff_projectinfo.sh

command without the -C option) will set the status to unread. This signals the

professionals the necessity to read it. Note: A FlexFrame with an unread projectinfo

database file is not entitled for an upgrade installation. Use the -C option not until you run

the FlexFrame upgrade. For details see ff_project_detector.sh(8) and the

―FlexFrame for SAP Upgrade Guide‖. This state of read will remain for a limited period

only.

-v

The command will give some more detailed information during its work.

Software and Hardware Update and Maintenance

358 Administration and Operation

-d

The command will write detailed debug information into FlexFrame's default debug files.

More details and comprehensive use cases are available in the online manual pages of ff_projectinfo.sh and ff_project_detector.sh.

Troubleshooting

Administration and Operation 359

12 Troubleshooting

12.1 Locking of Flex Frame Administration Commands

FlexFrame administrative commands stored at /opt/FlexFrame/bin use a resource

locking mechanism to prevent data corruption.

The implementation on the control nodes uses lock files on a shared file system mounted

from the common NAS. Each lock file represents a resource access. Each administrative

command uses typically only a single lock at a time to prevent deadlocks. The lock files

contain information about:

the locking user,

the executing host and the process id,

the timeout set,

the date and time of day when the lock expires,

the name and command line arguments of the locking process.

If a command tries to set a lock which is already set the command will retry to set the lock

for five times with a delay of one second. If the lock could not be set the command will

abort with a notice to user.

To get a robust locking mechanism locks will be removed if a timeout of 300 seconds was

reached AND the locking process no longer exists. If the locking process is still running,

an error message will be displayed by newly called program. The error message and the

exit code define the type or error as shown below.

These locking conditions get detected:

Exit Code Condition Description

0 normal

operation

All fine, locking works as expected. No resource collision with

other processes.

2 normal

operation

There is already a lock and the timeout is not reached.

3 Error IO error setting or unsetting a lock. This should not happen and

points to hanging mounts or filled up filesystems.

Troubleshooting

360 Administration and Operation

Exit Code Condition Description

4 Error Lock set, timeout reached but locking process is still alive. May

be the locking process hangs. Only an administrator is able to

decide if locking process works or hangs. An administrator may

have to clear the lock manually

5 Error Lock set but lock file is empty, therefore no further checks are

possible. This should not happen as FlexFrame locking writes

data to lock files. An administrator has to clear the lock

manually.

Manual administration is necessary for all three error conditions. The administrator has to

check for source of trouble and resolve it. Some examples:

Resource File path

LDAP /FlexFrame/volFF/FlexFrame/locks/\

ldap_lock

ff_new_an.sh /FlexFrame/volFF/FlexFrame/locks/\

ff_new_an.sh_lock

ff_install_an_linux_images.sh /FlexFrame/volFF/FlexFrame/locks/\

ff_install_an_linux_images.sh_lock

The content of the lock files:

User: <login name>

Pid: <id of locking process>

Host: <node name where locking process lives>

Timeout: <time set in seconds> (lock valid till <date and time>)

<command line of locking process>

12.2 Script Logging in FlexFrame

All CN scripts in /opt/FlexFrame/bin log into one file, which is

/var/log/FlexFrame/log_(TIMESTAMP) by default, e.g.

/var/log/FlexFrame/log_2010_05_31. This file is changed daily.

Messages with priority greater than DEBUG (i.e. INFO, WARN and ERROR) are written

into this logfile.

Troubleshooting

Administration and Operation 361

All CN scripts in /opt/FlexFrame/bin debug (if called with option DEBUG) into one

file, which is /var/log/FlexFrame/debug_(TIMESTAMP) by default, e.g.

/var/log/FlexFrame/debug_2010_05_31 . This file is changed daily.

All messages are written into this debug file.

To get further debug information, all Perl scripts can be called with option –debug, as

well as some Shell scripts can be called with option –d .

Exception : The installation scripts ff_init_conf.pl, ff_post_conf.sh and

ff_slapd_init.sh log/debug to /opt/FlexFrame/init_conf/.

ff_nas_ha.pl has its own logging as described in Chap. 8.1.16.

A log entry contains timestamp, priority, username, process ID, hostname, script name

and message text. It looks like this:

2009-04-22 12:29:52 INFO [root,9061,jer1cn1] ff_pool_adm.pl.sh

log example

Two configuration files, log4sh.conf and log4perl.conf reside under

/opt/FlexFrame/etc/.

The destination for Logging/Debugging can be changed in these configuration files by

modifying the corresponding filename lines.

You have to change the destination filenames in both configuration files.

Example:

If you want all scripts (excluding the exceptions) to log into /FlexFrame/volFF/FlexFrame/log_(TIMESTAMP) you have to change

log4perl.appender.LogFileApp.filename=/var/log/FlexFrame/log

to

log4perl.appender.LogFileApp.filename=/FlexFrame/volFF/FlexFrame/l

og

in log4perl.conf

and

log4sh.appender.A1.File = '/var/log/FlexFrame/log'

to

log4sh.appender.A1.File = '/FlexFrame/volFF/FlexFrame/log'

in log4sh.conf

Troubleshooting

362 Administration and Operation

Attention:

The directory of the log-/debugfiles (in the example above

/FlexFrame/volFF/FlexFrame) has to exist.

The user is responsible for the backup/delete of older logfiles.

12.3 Log Files

The default location for logging information of the FlexFrame scripts is

/var/log/FlexFrame on the Control Nodes.

Additional information can be found under /var/log/FlexFrame/tmp .

For easy gathering of log file information on all Application Nodes in FlexFrame, it is not

necessary to login on each Application Node because all log files are directly readable at

the Control Nodes on the mounted NAS system.

To elevate the admin to read these log files it is useful to create symbolic links on the

Control Nodes in the directory /var like this:

control1:/ # cd /var

control1:/var # ls -ld log*

drwxr-xr-x 20 root root 1440 May 2 14:01 log

lrwxrwxrwx 1 root root 58 Apr 29 17:28

log_pool1_klinge1 -> /FlexFrame/volFF/os/Linux/FSC_4.2A00-000.SLES-

9.X86_64/var_img/var-ac10020e/log

lrwxrwxrwx 1 root root 58 Apr 29 17:29

log_pool1_klinge2 -> /FlexFrame/volFF/os/Linux/FSC_4.2A00-000.SLES-

9.X86_64/var_img/var-ac10020f/log

lrwxrwxrwx 1 root root 58 Apr 29 17:30

log_pool1_klinge3 -> /FlexFrame/volFF/os/Linux/FSC_4.2A00-000.SLES-

9.X86_64/var_img/var-ac100210/log

lrwxrwxrwx 1 root root 58 Apr 29 17:30

log_pool1_klinge4 -> /FlexFrame/volFF/os/Linux/FSC_4.2A00-000.SLES-

9.X86_64/var_img/var-ac100211/log

lrwxrwxrwx 1 root root 58 Apr 29 17:30

log_pool1_klinge5 -> /FlexFrame/volFF/os/Linux/FSC_4.2A00-000.SLES-

9.X86_64/var_img/var-ac100212/log

lrwxrwxrwx 1 root root 58 Apr 29 17:31 log_otto_RX300-

01 -> /FlexFrame/volFF/os/Linux/FSC_4.2A00-000.SLES-

9.X86_64/var_img/var-ac100113/log

lrwxrwxrwx 1 root root 58 Apr 29 17:33 log_otto_RX300-

02 -> /FlexFrame/volFF/os/Linux/FSC_4.2A00-000.SLES-

9.X86_64/var_img/var-ac100114/log

control1:/var #

Troubleshooting

Administration and Operation 363

To read the /var/log/messsages from Application Node klinge5 in pool pool1

invoke this:

control1:/ # cd /var

control1:/var # less log_pool1_klinge5/messages

Or enter the log directory from this Application node and look for log files:

control1:/ # cd /var/log_pool1_klinge5

control1:/var/log_pool1_klinge5 # ls -lrt|tail

-rw-r--r-- 1 root root 278 May 4 16:23 log.scagt

-rw-r--r-- 1 root root 174 May 4 16:24 log.vvagt

-rw-r--r-- 1 root root 257 May 4 16:24 log.statusagt

-rw-r--r-- 1 root root 612 May 4 16:28 ntp

-rw-r--r-- 1 root root 5542 May 4 16:35 warn

-rw-r--r-- 1 root root 1696 May 4 16:35 auth.log

-rw-r--r-- 1 root root 484 May 4 18:16 log.busagt

drwxr-xr-x 2 root root 4096 May 4 18:19 sa

-rw-r--r-- 1 root root 27622 May 4 18:19 messages

prw------- 1 root root 0 May 4 18:24 psadfifo

12.4 Network Errors

In most cases, network problems are most times configuration mistakes. All switches are

configured to send SNMP traps and log via syslog to Control Nodes. To see what happened, look at the Control Nodes' /var/log/messages. Any conditions reported by

the switches can be found here. For SNMP traps look at the FA Agents support

database. The FA Agents collect all kinds of SNMP traps for the entire FlexFrame

environment.

12.5 NFS Mount Messages

During start/stop procedures of SAP instances, NFS mount failures for /usr/sap/SYS

and /oracle/db_sw can be seen in /var/log/messages. These messages are not

FlexFrame-specific failures, but may occur on any system that has /usr/sap and

/oracle in its automount maps. These messages are caused by the way SAP is linking

binaries during software development process.

12.6 LDAP Error Codes and Messages

LDAP failures may have various reasons. To determine the reason, first have a look at

the Control Node's /var/log/messages. If slapd does not report anything you may

have to increment the log level configured with /etc/openldap/slapd.conf. Log

levels are additive, and available levels are:

Troubleshooting

364 Administration and Operation

Log level Meaning

1 Trace function calls

2 Debug packet handling

4 Heavy trace debugging

8 Connection management

16 Print out packets sent and received

32 Search filter processing

64 Configuration file processing

128 Access control list processing

256 Stats log connections/operations/results

512 Stats log entries sent

1024 Print communication with shell backends

2048 Entry parsing

Most LDAP problems are related to access control and connection management.

If you have changed the log level send a -HUP to the slapd to advice it to reread

configuration files (use pkill -HUP slapd).

Another reason may be the LDAP client configuration. Check the following issues:

Does the client use the correct server address?

Look at the client messages file to get this information.

If ldapsearch -x on Linux and works properly the problem may be the nsswitch

configuration. These are the top most problems.

Troubleshooting

Administration and Operation 365

When reporting problems with the administration scripts, please provide the

following information:

The error message of the admin tool

A dump of the LDAP database (ldapsearch -x > <file name>)

The configuration files /etc/openldap/slapd.conf and

/FlexFrame/volFF/FlexFrame/ldap/common/slapd.acl.conf

12.7 LDAP and Cache Coherence

While using command scripts to administer your FlexFrame environment, it is possible

that you do not see the changes on Application Nodes immediately. The reason is the

usage of a cache daemon (nscd) which refreshes the cache content in defined time

periods.

There are different caches for different types (e.g. passwd, group, etc.) The configuration of the cache properties is done in /etc/nscd.conf (default). Normally you must not

change the FlexFrame default configuration. Please ask professional service collegues

before doing any change.

You can display your current configuration with nscd –g.

For further information see the corresponding manpages (nscd, nscd.conf).

12.7.1 Linux

To force the refresh of the cache content use nscd –i or nscd restart.

12.8 Start/Stop Script Errors

While executing the SAP service start/stop scripts, several errors may occur for arbitrary

reasons. The messages are grouped by the severity of the message.

The scripts write three types of logs (/FlexFrame/script/logs)

sapservice.log

The log is always available and logs the requests. It contains only the requested

command and the exit state of the request

sap<type>_<SID>_<action>.log

That log contains the messages also written to STDOUT/STDERR

sapservice.<nnn>.debug.<SID>.<type>.<instnr>.action>

That log is only written if option -d is set. It contains a detailed logging of the course

of actions.

The message text below uses ‗%%‘ to indicate there will be a variable output in the

message. There are up to four varying strings in a message.

Troubleshooting

366 Administration and Operation

If you are creating a debug file you are requested to delete debug files by yourself. Otherwise they resist in /FlexFrame/script/log forever.

12.8.1 Severity ‘INFO’

Messages from severity ‗INFO‘ should show the progress of a request and they are always written to STDOUT and to the debug file if option -d is set.

12.8.2 Severity ‘WARN’

Messages from severity ‗WARN should give you a hint to possible configuration errors.

They are always written to STDOUT and to the debug file if option -d is set.

Most of the WARN messages are not critical but you should check if such a message is

displayed.

12.8.3 Severity ‘ERROR’

Messages from severity ‗ERROR‘ are always written to STDOUT and to the debug file if

option -d is set. You have to check it and resolve the issue which leads to the error

message. With opcode ‗cleanup‘ sometimes messages with severity ‗ERROR‘ are shown

but can be ignored.

12.8.4 Severity 'DEBUG'

Debug messages are only printed to the debug file if option -d is used. Those message

are only written to support diagnostics.

12.9 Script Debugging

12.9.1 Shell Scripts

For many FlexFrame scripts, debugging is activated by default. If debugging is not active,

it can be activated with the option –d:

ff_netscan.sh -d

The default location for the debug log file is /var/log/FlexFrame

Further on, shell scripts can be traced using the option -x.

For details, see the man page of the shell in use (man sh; man bash; etc.)

sh -x ff_netscan.sh

Troubleshooting

Administration and Operation 367

12.9.2 Perl Scripts

Before debugging, activate the logging functions of the shell by calling script (For

details, see man script).

The debugger is called using

perl -d <script_name>

Functions:

h

Help

x <expression>

Evaluate an expression (hash/array) and display the result

p <string>

Returns strings

s

Execute next command; follow subroutines

n

Execute next command; skip subroutines

b <line_number>

Set break point at line <line number>

q

Quit.

For further information, see the man page of perldebug.

12.10 Debugging the Linux Kernel

For the Linux Kernel, debugging is disabled by default. There are several approaches to

collect debugging information. Everything is only described briefly; please contact Fujitsu

support for further instructions.

12.10.1 Netconsole

If the Linux kernel detects an internal error, it tries to write some information about the so

called ―Oops‖ into the messages file.

Troubleshooting

368 Administration and Operation

If an error completely crashes the kernel (which is known as a kernel panic), this

information may be generated but cannot be written to the disk or the NFS mount.

To get this information anyway, activate the netconsole feature to redirect kernel logging

over the network.

If the netconsole output does not provide the required information, you should configure

the crash dump functionality.

Information about using the netconsole can be found in the Linux Kernel documentation

in /usr/src/linux/Documentation/networking/netconsole.txt.

Please note that due to the use of bonding interfaces in FlexFrame there may be some

limitations, netconsole may not work properly with the bonding slaves. It is recommended

to use a dedicated interface.

12.10.2 Capturing Crash Dumps

12.10.2.1 Common Restrictions for Taking Crash Dumps

If taking crash dumps is configured and activated, the process of writing the dump

must not be interrupted, so the Linus HA cluster and myAMC agent power shutdown

features must be disabled for this specific node, for example by pulling the IPMI cable or

setting appropriate timeouts. On the Control Nodes, STONITH should be disabled if the

crash dump cannot be written within 45 seconds which is very likely.

This can be done by disabling the ―Enable STONITH‖ checkbox of the ―linux-ha‖ proper-

ties in the Heartbeat GUI interface.

Amongst others that is one reason for not activating crash dump capturing by default.

12.10.2.2 "Kdump" Kernel Crash Dump Capturing

Since SLES10, the implementation how to take crash dumps was changed to the kdump

method. If a kernel panic occurs, a special kdump-kernel will be directly executed by the

running kernel and during the boot process of this kernel, the dump of the whole system

memory will be written to a sparse file in /var/log/dump.

This feature is well integrated in FlexFrame Application Nodes and may be activated by

specifying a kernel command line option in the PXE configuration file:

Open the corresponding file in /tftpboot/pxelinux.cfg

Append crashkernel=128M@16M to the APPEND line

Reboot the corresponding Application Node(s)

Troubleshooting

Administration and Operation 369

12.10.2.3 Forcing a Crash Dump

There may be situations where the kernel does not initiate a crash dump automatically.

You can manually start a crash dump by using the sysrq shortcuts.

Activate sysrq as follows:

cn1:~ # echo 1 >/proc/sys/kernel/sysrq

To make this setting persistent, set ENABLE_SYSRQ="yes" in

/etc/sysconfig/sysctl.

Now a manual's crash dump capture can be manually initiated by pressing

Alt+SysRq+c altogether on the keyboard. On most keyboards, the SysRq key is equal

to the PrintScrn key.

If the graphical desktop is running, it is recommended to switch to a text console by pressing Ctrl+Alt+F10 before starting the dump, otherwise it might not be possible to

see anything while the kdump-kernel is being executed and the dump is being written.

12.11 Activate core dumps on CN or AN

To investigate problems (e.g. bus errors) with Linux commands it is sometimes helpfull to

get a ‗core dump‘. Normally the core file creation is disabled by default.

To activate core dumps in the current running shell use the following command::

In a ‗bash‘ shell use

an_Linux# ulimit –c unlimited

In a ‗csh‘ shell use

rx300-15 /home_sap/sapadm> unlimit coredumpsize

Now run the faulty command within this shell. The core dump can be found in the

directory /var/log/coredumps.

The filename follows the format:

core-<executable filename>-<time of dump in seconds>

Troubleshooting

370 Administration and Operation

If it is necessary to enable core file creation for a complete login session, you have to globally enable core dump writing .Change the file /etc/security/limits.conf on

the CN or /FlexFrame/volFF/os/Linux/FSC*/root_img/etc/security/limits.conf

for an Application Node and add the line

* soft core unlimited

In the next login session the creation of core dump files is enabled. To check use the command ulimit –a for a bash shell environment or the command limit for a csh

shell environment.

To disable this functionality remove the above entry from the limits.conf file.

Abbreviations

Administration and Operation 371

13 Abbreviations

ABAP Advanced Business Application Programming

ACC Adaptive Computing Controller

ACI Adaptive Computing Infrastructure

ACPI Advanced Configuration and Power Interface

APM Advanced Power Management

APOLC Advanced Planner & Optimizer Life Cache

CCU Console Connection Unit

CIFS Common Internet File System

DART Data Access in Real Time

DHCP Dynamic Host Configuration Protocol

DIT Domain Information Tree

ERP Enterprise Resource Planning

EULA End User License Agreement

FAA FlexFrame Autonomous Agent

FC Fiber Channel

FSC FlexFrame Software Container, name part used for

Images and CDs/DVDs

FTP File Transfer Protocol

IP Internet Protocol

LAN Local Area Network

LDAP Lightweight Directory Access Protocol

LUN Logical Unit Number

MAC Media Access Control

MINRA Minimal Read Ahead

NAS Network Attached Storage

NDMP Network Data Management Protocol

NFS Network File System

Abbreviations

372 Administration and Operation

NIC Network Interface Card

NVRAM Non-Volatile Random Access Memory

OLTP On-Line Transaction Processing

ONTAP Open Network Technology for Appliance Products

OSS Open Source Software

PFS Production File System (on Celerra)

PXE Preboot Execution Environment

PY PRIMERGY

QA Quality Assurance

QS Quality of Service

RAID Redundant Array of Independent (or Inexpensive) Disks

RDBMS Relational Database Management System

SAP BW SAP Business Warehouse

SAPGUI SAP Graphical User Interface

SAPOSS SAP Online System Service

SID System Identifier

SLD System Landscape Directory

SLES SUSE Linux Enterprise Server

SMB Server Message Block

SNMP Simple Network Management Protocol

SPOC Single Point Of Control

TELNET Telecommunications Network

TFTP Trivial File Transfer Protocol

UDP User Datagram Protocol

UPS Uninterruptible Power Supply

VLAN Virtual Local Area Network

VTOC Virtual Table Of Contents

WAN Wide Area Network

WAS Web Application Server

Abbreviations

Administration and Operation 373

WAFL Write Anywhere File Layout

Administration and Operation 375

14 Glossary

Adaptive Computing Controller

SAP system for monitoring and controlling SAP environments.

Advanced Business Application Programming

Proprietary programming language of SAP.

Advanced Power Management

Advanced Power Management defines a layer between the hardware and the

operating system that effectively shields the programmer from hardware details.

Application Agent

A software program for monitoring and managing applications.

Application Node

A host for applications (e.g. SAP instances db, ci, agate, wgate, app etc.). This

definition includes Application Servers as well as Database Servers.

Automounter

The automounter is an NFS utility that automatically mounts directories on an NFS

client as they are needed, and unmounts them when they are no longer needed.

Autonomous Agent

Central system management and high availability software component of FlexFrame.

Blade

A special form factor for computer nodes.

BladeRunner

The working title for the solution part of SAP for FlexFrame.

BRBACKUP

SAP backup and restore tools.

Celerra

NAS system of EMC.

Checkpoint Restore

On EMC Celerra a SnapSure feature that restores a PFS to a point in time using

checkpoint information. As a precaution, SnapSure automatically creates a new

checkpoint of the PFS before it performs the restore operation.

Client LAN

Virtual network segment within FlexFrame, used for client-server traffic.

Common Internet File System

A protocol for the sharing of file systems (same as SMB).

Glossary

376 Administration and Operation

Computing Node

From the SAP ACI perspective: A host that is used for applications.

Control Agent

A software program for monitoring and managing nodes within FlexFrame.

Control Center

The two Control Nodes (CN) of FlexFrame for SAP are also named as the FlexFrame

Control Center (CC). In this documentation the notation Control Node (CN) is used

as a synonym for Control Center (CC) and vice versa.

Control LAN

Virtual network segment within FlexFrame, used for system management traffic.

Control Node

A physical computer system, controlling and monitoring the entire FlexFrame

landscape and running shared services in the rack (dhcp, tftp, ldap etc.).

Control Station

A Control Node in an SAP ACI environment.

DART

Operating system of Celerra data movers (Data Access in Real Time).

Dynamic Host Configuration Protocol

DHCP is a protocol for assigning dynamic IP addresses to devices on a network.

Dynamic Host Configuration Protocol server

A DHCP server provides configuration parameters specific to the DHCP client host,

required by the host to participate on the Internet.

EMC NAS

Network attached storage for file systems of EMC.

Enterprise Resource Planning

Enterprise Resource Planning systems are management information systems that

integrate and automate many of the business practices associated with the

operations or production aspects of a company.

Ethernet

A Local Area Network which supports data transfer rates of 10 megabits per second.

Fiber Channel

Fiber Channel is a serial computer bus intended for connecting high-speed storage

devices to computers.

Filer

Network attached storage for file systems of NetApp.

Glossary

Administration and Operation 377

FlexFrame

A joint project in which the main partners are SAP, Network Appliance, Intel and

Fujitsu.

FlexFrameTM

for SAP®

FlexFrame for SAP®

is a radically new architecture for SAP environments. It exploits

the latest business-critical computing technology to deliver major cost savings for

SAP customers.

FlexFrame internal LAN Switch

Network switches which are integral part of the FlexFrame for SAP hardware

configuration and which are automatically configured by the FlexFrame for SAP

software.

Gigabit Ethernet

A Local Area Network which supports data transfer rates of 1 gigabit (1,000

megabits) per second.

Host name

The name of a node (assigned to an interface) that is resolved to a unique IP

address. One node can have multiple host names (cf. node name). In SAP

environments host names are currently limited to 13 alphanumeric characters

including the hyphen (― - ―). The first character must be a letter. In the SAP

environment host names are case-sensitive.

Image

In the FlexFrame documentation, ―Image‖ is used as a synonym for ―Hard Disk

Image‖.

Internet Protocol Address

A unique number used by computers to refer to each other when sending information

through networks using the Internet Protocol.

Lightweight Directory Access Protocol

Protocol for accessing on-line directory services.

Local Area Network

A computer network that spans a relatively small area. Most LANs are confined to a

single building or group of buildings. However, one LAN can be connected to other

LANs over any distance via telephone lines and radio waves. A system of LANs

connected in this way is called a Wide Area Network (WAN).

Local host name

The name of the node (physical computer); it can be displayed and set using the

command /bin/hostname.

Logical Unit Number

An address for a single (SCSI) disk drive.

Glossary

378 Administration and Operation

MAC address

Device identifier number of a Network Interface Card. In full: "media access control

address".

MaxDB

A relational database system from mySQL (formerly ADABAS and SAPDB).

Media Access Control address

An identifier for network devices, usually unique. The MAC address is stored

physically on the device.

NAS system

Network Attached Storage of any vendor (in our context: EMC NAS or NetApp Filer).

NDMPcopy

NDMPcopy transfers data between Filers using the Network Data Management

Protocol (NDMP).

Netboot

A boot procedure for computers where the operating system is provided via a

network instead of local disks.

Netweaver

SAP NetWeaver is the technical foundation of SAP solutions.

Network Appliance Filer

See ―Filer‖.

Network Attached Storage

A data storage device that is connected via a network to one or multiple computers.

Network File System

A network protocol for network-based storage access.

Network Interface Card

A hardware device that allows computer communication via networks.

Node

A physical computer system controlled by an OS.

Node name

The name of a physical node as returned by the command uname -n. Each node

name within a FlexFrame environment must be unique.

Non-Volatile Random Access Memory

A type of memory that retains its contents when the power is turned off.

On-Line Transaction Processing

Transaction processing via computer networks.

OpenLDAP

An Open Source LDAP Service Implementation.

Glossary

Administration and Operation 379

Open Network Technology for Appliance Products

The operating system of Network Appliance Filers.

Open Source Software

Software that is distributed free of charge under an open source license, such as the

GNU Public License.

Oracle RAC

A cluster database by Oracle Corporation.

Physical host

Name of a physical computer system (node).

Power-On Self Test

Part of a computer's boot process; automatic testing of diverse hardware

components.

Preboot Execution Environment

An environment that allows a computer to boot from a network resource without

having a local operating system installed.

PRIMERGY

Fujitsu's i386-based server product line.

SAP Service

In FlexFrame: SAP Service and DB Services.

SAP service script

An administration script for starting and stopping an SAP application on a virtual host.

SAP Solution Manager

Service portal for the implementation, operation and optimization of an SAP solution.

SAPLogon

Front-end software for SAPGUI.

SAPRouter

Router for SAP services like SAPGUI or SAPTELNET.

SavVol

A Celerra volume to which SnapSure copies original point-in-time data blocks from

the PFS before the blocks are altered by a PFS transaction.

Server

A physical host (hardware), same as node.

Service

A software program providing functions to clients.

Service type

The type of an application or service (db, ci, app, agate, wgate etc.).

Glossary

380 Administration and Operation

Single Point of Control

In FlexFrame: One user interface to control a whole FlexFrame environment.

Storage LAN

A virtual LAN segment within a FlexFrame environment, carrying the traffic to NAS

systems.

SUSE Linux Enterprise Server

A Linux distribution by Novell, specializing in server installations.

Telecommunications Network

A terminal emulation program for TCP/IP networks such as the Internet.

Trivial File Transfer Protocol

A simple form of the File Transfer Protocol (FTP). TFTP uses the User Datagram

Protocol (UDP) and provides no security features. It is often used by servers to boot

diskless workstations, X-terminals, and routers.

TFTP server

A simple FTP implementation.

Virtual host

The name of the virtual host on which an application runs; it is assigned to a physical

node when an application is started.

Virtual Local Area Network

A VLAN is a logically segmented network mapped over physical hardware according

to the IEEE 802.1q standard.

Virtualization

Virtualization means the separation of hardware and processes. In a virtualized

environment (FlexFrame), a process can be moved between hardware nodes while

staying transparent to the user and application.

.

Administration and Operation 381

15 Index

A

abbreviations 371

Application

removing from monitoring by FA

Agents 305

stopping ans starting for upgades

using r3up 307

Application Node

hardware maintenance 347

software maintenance 318

Application Node Images

installation 321

Application Nodes 10

adding 96

administrating 91

and SAN 104

displaying information on a

specific 91

displaying information on all 95

listing 91

reactivating after power shutdown by

FA Agents 48

removing 101, 239

renaming 101

state of 52

updating RPM packages 342, 345

application services

starting and stopping 297

application software

upgrading 341, 344

automounter concept 31

B

Blade Server Cabinets

adding 107

administrating 104

listing 105

removing 109

C

Celerra SRDF NAS

high availability 154

cloning of SIDs 269, 276

cluster

configuring 24

starting 25

status information 25

cluster file system, built 35

Cluster Status 61

common network configuration

parameters

display and change 218

constraints 14

Control Node

defect hardware 318

exchanging 317

failed hardware 317

hardware maintenance 317

OS damaged 318

382 Administration and Operation

replacing network card 318

software updates 311

Control Nodes 10

backup 315

backup/restore 315

restore 315

core dump 369

crash dump 368

create new initrd 339

D

DART patches

installing 351

Debugging

Linux kernel 367

script 366

document history 2

dynamic LUN masking 172

E

EMC NAS 36

cluster architecture 37

control station 36

data mover 37

snapshots 38

two side mirror architecture 38

ESXi server maintenance 350

BIOS update 350

replace network card 350

update and patches 350

F

FA Autonomous Agents 51

ff_change_id.pl 269

ff_clone_sid.pl 254, 255, 268, 269

ff_hosts.sh 80

ff_list_services.sh 296

ff_nas_adm.pl 138

ff_relocate_sid_data.pl 287

ff_san_info.sh 199

ff_san_ldap_conf.pl 189

ff_san_luns.pl 204

ff_san_mount.sh 198

ff_san_qlascan.sh 201

ff_san_srdf.pl 201

ff_sap_shadowport.sh 282

ff_sap_upgrade.pl 284

ff_sid_adm.pl 245

ff_swgroup_adm.pl 208, 230

ff_swport_adm.pl 233

ff_user_adm.pl 82

Filer cluster 35, 36

FlexFrame

architecture 5

backup with tape library 62

basic administration 44

general notes 5

network 52

FlexFrame configuration state,

displaying 49

FlexFrame landscape

Index

Administration and Operation 383

accessing 44

power-off 45

power-on 44

FlexFrame Web portal 50

G

glossary 375

H

host

database 79

I

install kernel 339

install new initrd 339

installing new linux kernel 339

J

Jakarta Tomcat 51

K

kernel switching 340

L

LDAP 155

FlexFrame structure in 11

working with 12

LDAP and Cache Coherence 365

LDAP configuration 20

LDAP error codes and messages 363

LINUX Cluster Components /

Functionality 12

Linux kernel

install new kernel 314

mount software stage 313

reboot control node 314

update/install new 313

update/install new 338

linux-ha cli commands 26

Linux-HA cluster 51

Linux-HA cluster on control center 12

linux-ha gui 27

linux-ha logfile 27

locking 359

log files 362

LUN masking, dynamic 172

M

maintenance cycle 180, 318, 319, 321,

323, 325, 326, 330

assigning, booting, maintaining (step

#2) 325

maintaining use cases 337, 347

maintenance base image (step

#1) 323

migrating remaining Application

Nodes (step #3) 333

overview 319

re-using maintenance images 335

reverting (step #3) 330

N

NAS

add new 138

adding a blade 147

changing LinkAggregate ports 151

disk free 152

display all configured 143

384 Administration and Operation

display configuration 144

remove 141

removing a blade 148

NAS support architectual overview 34

NAS systems

multiple 279

NAS systems configuration (EMC and

NetApp) 138

NetApp Filers

configuring SNMP traps 142

netboot change BIOS settings 347

network 27

automounter concept 31

LAN failover 28

segments 28

switches 29

Network Appilance Filer 34

network card

replacing 347

network errors 363

NFS mount messages 363

notational conventions 1

O

ONTAP patches

installing 351

P

Partner Linux Driver Process

(PLDP) 340

PLDP 340

pool

add to a NAS 145

remove from a NAS 146

pool default router

changing 77

pool DNS domain

changing 76

pool groups

changing group and pool assignment

of Application Nodes 79

changing group assignment of

Application Nodes 79

removing 78

pools

adding 64

adding a group 77, 178

listing details 70, 74

removing 69

state of 51

pools and groups 64

power control hardware

replacing 349

R

related documents 2

remote administration 44

replace

ESXi network card 350

network card 347

power control hardware 349

switch blade 349

requirements 1

Index

Administration and Operation 385

resource ff_manage 22

resource groups 17

resource netboot 22

S

SAN

basic layers 41

configuration in a FlexFrame

environment 158

rules and restrictions with

FlexFrame 43

scope of the FlexFrame

integration 42

support architectural overview 40

SAN configuration

FlexFrame 189

setting up 158

SAP kernel

updates and patches 286

SAP service

scripts 299, 303

starting and stopping without root

privileges 298

SAP service scripts

actions 300

return code 304

user exits 303

SAP services

administrating 295

display status 295

list status 296

starting and stopping multiple 304

SAP SIDs

adding instances 247, 253, 254,

259, 263, 265

modifying instances 247, 253, 254,

259, 263, 265

removing instances 247, 253, 254,

259, 263, 265

SAP SIDs

adding 247

listing 245

listing instances 245

modifying 247

removing 247

SAP SIDs

adding 253

SAP SIDs

removing 253

SAP SIDs

modifying 253

SAP SIDs

adding 254

SAP SIDs

removing 254

SAP SIDs

modifying 254

SAP SIDs

adding 259

SAP SIDs

removing 259

SAP SIDs

386 Administration and Operation

modifying 259

SAP SIDs

adding 263

SAP SIDs

removing 263

SAP SIDs

modifying 263

SAP SIDs

adding 263

SAP SIDs

removing 263

SAP SIDs

modifying 263

SAP SIDs

adding 265

SAP SIDs

removing 265

SAP SIDs

modifying 265

SAP SIDs

adding 265

SAP SIDs

removing 265

SAP SIDs

modifying 265

SAP SIDs

adding 265

SAP SIDs

removing 265

SAP SIDs

modifying 265

SAP SIDs

cloning 268

SAP SIDs

listing instances 268

Sap system

handling 244

SAP system

upgrading 282

SAP systems

state of 52

score 14

script logging 360

scripts

ff_list_services.sh 296

ff_san_info.sh 199

ff_san_luns.pl 204

ff_san_mount.sh 198

ff_san_qlascan.sh 201

ff_san_srdf.pl 201

ff_sap_shadowport.sh 282

ServerView

update 340

update agents 341

update RAID manager 341

ServerView Operations Manager 53

ServerView update via RPM 313

service packs 338

Index

Administration and Operation 387

service switchover 307

severity

DEBUG 366

ERROR 366

INFO 366

WARN 366

shared operating system 9

SID instances

state of 52

simple resources 14

snapshot 35

software stage 338

spare nodes

pool-independent 89

specific configuration 17

SRDF support 179

start/stop script errors 365

state

of Applicatrion Nodes 52

of pools 51

of SAP systems 52

of SID instances 52

stickiness 15

stonith 22

storage systems administration 138

switch

add to a switch group 208

remove from a switch group 211

switch administration 208

switch blade

change blade uplink 115

changing name 112

changing password 113

changing type 111

getting initial configuration 114

replacing 349

switch configuration

backup 352

restoring 353

switch group

adding 220

adding an uplink 226

change host name 216

change password 215

deleting an uplink 228

extending an uplink 227

list configuration 212, 214

moving a blade cabinet to another

switch group 116

removing 225

switch port

add configuration 233

display configuration 236, 237

remove configuration 235

T

Third Party products 354

troubleshooting 359

388 Administration and Operation

U

update and maintenance 311

update PXE config file 339

updating RPM packages 342, 345

patch server connection 342, 345

patch server setup 342, 345

perfrom online update 343, 346

prepare online update 342, 345

updating vmware-open-vm-tools 343,

346

upgrade FF landsacpe 311

user and groups administration 82

V

volFF

unloading 287

volume layout 35

volumes

multiple 279