sg246261
Front cover
Implementing Linux
with IBM Disk Storage
Your guide to Linux implementations on
IBM eServers
Use Linux efficiently with FAStT
and ESS
Explore options relating
to Linux in SANs
Bertrand Dufrasne
Ronald Annuss
James Goodwin
Paul McWatt
Arwed Tschoeke
ibm.com/redbooks
International Technical Support Organization
Implementing Linux with IBM Disk Storage
June 2003
SG24-6261-01
Note: Before using this information and the product it supports, read the information in
“Notices” on page xiii.
Second Edition (June 2003)
This edition applies to the IBM Enterprise Storage Server, and the IBM FAStT Storage Server, for
use with the Linux operating system (Red Hat Enterprise Advanced Server 2.1 and SuSE Linux
Enterprise Server Edition 8) on IBM eServers.
© Copyright International Business Machines Corporation 2001 2003. All rights reserved.
Note to U.S. Government Users Restricted Rights -- Use, duplication or disclosure restricted by GSA ADP
Schedule Contract with IBM Corp.
Contents
Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Preface . . . . . . . . . . . . . . . . . . . . . . . . . .
Support considerations. . . . . . . . . . . . . . .
The team that wrote this Redbook . . . . . .
Become a published author . . . . . . . . . . .
Comments welcome . . . . . . . . . . . . . . . . .
.......
.......
.......
.......
.......
......
......
......
......
......
.......
.......
.......
.......
.......
......
......
......
......
......
. . xv
. . xv
. . xvi
. xviii
. xviii
Summary of changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix
May 2003, Second Edition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix
Chapter 1. Introduction to Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Historical perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 UNIX and the culture of collaboration . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 GNU and free software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.3 Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Open source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Linux distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3.1 Red Hat Enterprise Linux AS (formerly Advanced Server) . . . . . . . . . 6
1.3.2 SuSE Linux Enterprise Edition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.3 UnitedLinux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 IBM’s commitment to Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Chapter 2. Introduction to IBM TotalStorage . . . . . . . . . . . . . . . . . . . . . . . 11
2.1 Enterprise Storage Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.1 ESS Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.2 ESS features and benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 IBM Fibre Array Storage Technology (FAStT) . . . . . . . . . . . . . . . . . . . . . 18
2.2.1 FAStT models and expansion enclosure . . . . . . . . . . . . . . . . . . . . . 18
2.2.2 FAStT700 features and benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Chapter 3. SuSE Linux on zSeries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1 Introduction to zSeries and Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.1.1 zSeries specific storage attachments . . . . . . . . . . . . . . . . . . . . . . . . 27
3.1.2 Distributed storage attachments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
© Copyright IBM Corp. 2001 2003. All rights reserved.
iii
3.1.3 zSeries-specific versus distributed storage controllers . . . . . . . . . . . 32
3.1.4 zSeries and storage: Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 System requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2.1 Hardware requirements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.2.2 Software requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.3 Connection requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3 Configuration used for this redbook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.4 Implementing Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.4.1 Initial steps for installation as a z/VM Guest . . . . . . . . . . . . . . . . . . . 40
3.4.2 Initial steps for LPAR installation. . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.4.3 Installation with YaST2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.5 zSeries and disk storage under Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.5.1 ESS using FCP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3.5.2 Multipath support. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Chapter 4. SuSE Linux on pSeries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.1 Introduction to pSeries and Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.2 Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.2.1 Hardware requirements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.2.2 Software requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.2.3 Connection requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.3 Configuration used for this redbook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.4 Implementing SuSE Linux Enterprise Server 8 . . . . . . . . . . . . . . . . . . . . . 77
4.4.1 Installation steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.4.2 Linux installation in an LPAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.4.3 Post-installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Chapter 5. Linux on xSeries and BladeCenter . . . . . . . . . . . . . . . . . . . . . . 97
5.1 Introduction to xSeries, BladeCenter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.2 Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.2.1 Hardware requirements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.2.2 Software requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.2.3 Connection requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.3 Configurations used for this redbook. . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.4 Implementing Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.4.1 Installation steps for Red Hat on xSeries . . . . . . . . . . . . . . . . . . . . 102
5.4.2 Installation steps for Red Hat on BladeCenter . . . . . . . . . . . . . . . . 115
5.4.3 Installation steps for SLES8 on xSeries . . . . . . . . . . . . . . . . . . . . . 117
5.4.4 Installing SuSE Linux on BladeCenter . . . . . . . . . . . . . . . . . . . . . . 121
5.5 Post-installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.5.1 Configuring for the Summit kernel (Red Hat) . . . . . . . . . . . . . . . . . 121
5.5.2 Updating the kernel on BladeCenter (Red Hat) . . . . . . . . . . . . . . . 122
5.5.3 Issues on BladeCenter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
iv
Implementing Linux with IBM Disk Storage
5.6 Setup for Fibre Channel attachment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
5.6.1 Install the adapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
5.6.2 Prepare the Fibre Channel Host Bus Adapter cards . . . . . . . . . . . . 126
5.6.3 Installing the Fibre Channel Host Bus Adapter driver . . . . . . . . . . . 127
5.6.4 Loading the FAStT HBA driver . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.6.5 Attaching to ESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.7 BladeCenter Fibre Switch Module configuration . . . . . . . . . . . . . . . . . . . 137
5.8 Configuring FAStT. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.9 Configuring ESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Chapter 6. Red Hat Cluster Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
6.1 Preparing to install Cluster Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
6.1.1 Quorum partitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
6.1.2 Software watchdog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6.1.3 NMI watchdog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
6.1.4 Rawdevices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
6.1.5 Log file for Cluster Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
6.2 Installing Cluster Manager. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
6.2.1 Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
6.2.2 Basic setup of Cluster Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
6.2.3 Setting up an NFS clustered share . . . . . . . . . . . . . . . . . . . . . . . . . 149
Chapter 7. Configuring ESS for Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
7.1 IBM TotalStorage ESS Specialist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
7.2 Using the ESS Specialist. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
7.2.1 ESS attachment configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
7.2.2 Storage allocation procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
7.2.3 Defining Linux hosts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
7.2.4 Configure the host adapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
7.2.5 Configure RAID arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
7.2.6 Allocating storage for Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Chapter 8. BladeCenter SAN Utility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
8.1 Installing the BladeCenter SAN Utility . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Chapter 9. FAStT Storage Manager . . . . . . . . . . . . . . . . . . . . . . .
9.1 FAStT Storage Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9.1.1 Storage Manager concepts . . . . . . . . . . . . . . . . . . . . . . . .
9.1.2 FAStT Storage Manager client: Getting started . . . . . . . .
9.1.3 Updating the FAStT firmware . . . . . . . . . . . . . . . . . . . . . .
9.1.4 Setting up arrays, LUNs, and Storage Partitioning . . . . . .
. . . . . . . 187
. . . . . . . 188
. . . . . . . 188
. . . . . . . 191
. . . . . . . 195
. . . . . . . 198
Chapter 10. FAStT MSJ (Management Suite for Java) . . . . . . . . . . . . . . . 211
10.1 FAStT MSJ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
Contents
v
10.1.1 Configuring the LUN attachment . . . . . . . . . . . . . . . . . . . . . . . . . . 214
10.1.2 Use FAStT MSJ to configure Fibre Channel paths . . . . . . . . . . . . 216
10.1.3 Using FAStT as storage: A summary . . . . . . . . . . . . . . . . . . . . . . 222
Chapter 11. Implementing VMware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
11.1.1 VMware ESX architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
11.1.2 IBM and VMware. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
11.1.3 VMware system requirements. . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
11.2 Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
11.3 Implementing VMware ESX Server 1.5.2 . . . . . . . . . . . . . . . . . . . . . . . 230
11.3.1 Installation steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
11.3.2 Configuration of the ESX server . . . . . . . . . . . . . . . . . . . . . . . . . . 233
11.3.3 Setup disk space for VMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
11.3.4 Define a VM using the wizard . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
11.3.5 Install VMware remote console . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
11.3.6 Monitoring a VM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
Appendix A. Storage from the OS view. . . . . . . . . . . . . . . . . . . . . . . . . . . 253
Disk storage in the Windows environment . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
Disk letters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
Disk partitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
Disk storage in the Linux environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
Disk partitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
File hierarchy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
File systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
Identifying your SCSI storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Using your storage on Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
The file system table /etc/fstab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
Related publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Other publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
Online resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
How to get IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
vi
Implementing Linux with IBM Disk Storage
Figures
2-1
2-2
2-3
3-1
3-2
3-3
3-4
3-5
3-6
3-7
3-8
3-9
3-10
3-11
3-12
3-13
3-14
3-15
3-16
3-17
3-18
3-19
3-20
3-21
3-22
3-23
3-24
3-25
3-26
3-27
3-28
3-29
3-30
4-1
4-2
4-3
4-4
4-5
Photograph of ESS Model 800 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Storage area network (SAN) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
FAStT700 - Rear view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Running Linux on zSeries in an LPAR or as z/VM guest . . . . . . . . . . . . 27
Typical zSeries storage attachment concept . . . . . . . . . . . . . . . . . . . . . 27
SCSI addressing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
World Wide Names . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
FCP to Linux device name mapping . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
DASD addressing scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
DASD access translations FICON verses FCP . . . . . . . . . . . . . . . . . . . 33
IBM eServer zSeries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Hardware configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
HMC panel to load from FTP source . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Select the source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
HMC operating system messages interface . . . . . . . . . . . . . . . . . . . . . 49
Enter commands on the HMC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Network settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Installation server settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Language selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Load dasd module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Select installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Installation settings without partitioning settings . . . . . . . . . . . . . . . . . . 54
Expert partitioner without dasda . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Expert partitioner with dasda . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Expert partitioner with partition settings . . . . . . . . . . . . . . . . . . . . . . . . . 59
Installation settings with partitioning settings . . . . . . . . . . . . . . . . . . . . . 60
Warning before proceed with installation . . . . . . . . . . . . . . . . . . . . . . . . 60
Installation progress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Ready to IPL from dasd for first time . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Saving the settings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Add the Linux host . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Add volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
View volume assignments to display LUNs . . . . . . . . . . . . . . . . . . . . . . 67
pSeries models supporting Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Installation using serial console . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Installation using VNC client . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Installation using Web browser acting as VNC client by JAVA . . . . . . . 79
Main Menu, choose 2 to proceed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
© Copyright IBM Corp. 2001 2003. All rights reserved.
vii
4-6
4-7
4-8
4-9
4-10
4-11
4-12
4-13
4-14
4-15
4-16
4-17
4-18
4-19
4-20
4-21
4-22
4-23
4-24
5-1
5-2
5-3
5-4
5-5
5-6
5-7
5-8
5-9
5-10
5-11
5-12
5-13
5-14
5-15
5-16
5-17
5-18
5-19
5-20
5-21
5-22
6-1
6-2
viii
Power Control Menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
End of system boot, press 1/F1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Select boot device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Choose CD drive for installation CD boot . . . . . . . . . . . . . . . . . . . . . . . 82
Choose 1 to trigger SLES install . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
yaboot “boot”-prompt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Loading installation system from CD . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Type of terminal selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Start of VNC server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Enter IP address as specified in boot parameter . . . . . . . . . . . . . . . . . . 85
Enter the password you have defined . . . . . . . . . . . . . . . . . . . . . . . . . . 85
YaST2 in VNC window. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Installation settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Start of Linux installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Package installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Installation settings of the hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
HMC partition setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
LPAR boot mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Installation of the Emulex driver packages . . . . . . . . . . . . . . . . . . . . . . 94
IBM BladeCenter chassis and Blade . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Initial boot screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Welcome screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Install Options screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Initialize disk message . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Manual Partitioning screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Partitioning with Disk Druid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Adding a new partition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Boot Loader installation screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Network Configuration screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Authentication Configuration screen . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Selecting Package Group screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Individual Package selection screen . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Copying files screen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
End of installation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Boot prompt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
SLES 8 Installation Settings screen . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Network and peripherals setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Apply network configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Loaded SDD module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Output of cfgvpath . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Configure SDD vpath devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Configuration of HA cluster . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
grub.conf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Implementing Linux with IBM Disk Storage
6-3
6-4
6-5
6-6
6-7
6-8
6-9
7-1
7-2
7-3
7-4
7-5
7-6
7-7
7-8
7-9
7-10
8-1
8-2
8-3
8-4
8-5
8-6
8-7
8-8
8-9
8-10
8-11
8-12
9-1
9-2
9-3
9-4
9-5
9-6
9-7
9-8
9-9
9-10
9-11
9-12
9-13
9-14
/proc/interrupts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
/etc/sysconfig/rawdevices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
/etc/syslog.conf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Cluster status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Clustered NFS service added . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
Example of fstab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Clustat during failover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
ESS welcome screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
ESS Specialist main screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Storage Allocation panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Open System Storage panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Modify Host Systems panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Configure Host Adapter Ports panel . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Fixed Block Storage panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Add Volumes panel (1 of 2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Add Volumes panel (2 of 2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Modify volume assignments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
SAN Utility Installation Introduction screen . . . . . . . . . . . . . . . . . . . . . 178
SAN Utility Installation Folder screen. . . . . . . . . . . . . . . . . . . . . . . . . . 179
SAN Utility Create Links screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
SAN Utility Log Location screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
SAN Utility Browser Select screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
SAN Utility Pre-installation Summary screen . . . . . . . . . . . . . . . . . . . . 181
SAN Utility Installation screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
SAN Utility Install Complete screen . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
SAN Utility screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Add New Fabric screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
SAN Utility Topology View screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
SAN Utility Faceplate view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Password prompt for shell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
netCfgShow screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
Automatic Discovery screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Storage Manager Enterprise Management window . . . . . . . . . . . . . . . 194
Add Device screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
Subsystem Management screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Firmware Download Location screen. . . . . . . . . . . . . . . . . . . . . . . . . . 196
Firmware update screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
NVSRAM Download Location screen . . . . . . . . . . . . . . . . . . . . . . . . . 197
NVSRAM Update screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Default Host Type screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
Logical Drive Wizard screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Create Array screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
Logical Drive Parameters screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Figures
ix
9-15
9-16
9-17
9-18
9-19
9-20
9-21
9-22
9-23
9-24
9-25
9-26
10-1
10-2
10-3
10-4
10-5
10-6
10-7
10-8
10-9
10-10
10-11
10-12
10-13
10-14
11-1
11-2
11-3
11-4
11-5
11-6
11-7
11-8
11-9
11-10
11-11
11-12
11-13
11-14
11-15
11-16
11-17
x
Advanced Logical Drive Parameters screen . . . . . . . . . . . . . . . . . . . . 202
Create a New Logical Drive screen . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Arrays and logical drives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Mappings Startup Help screen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Host Group added . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Hosts added . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Define Host Port screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Host Ports added . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Storage Partitioning Wizard screen . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Select Host Group or Host screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Select Logical Drives/LUNs screen . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Logical drives added . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Installation Splash screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
FAStT_MSJ Introduction screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Product Features screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
Building a ramdisk image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Example of qlremote running . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
FAStT_MSJ screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
FAStT_MSJ connected . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Invalid Configuration screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Port Configuration screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Paths before balancing LUNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Configuration Saved screen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Balanced paths before rebooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Balanced paths after rebooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Example of modules.conf. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
VMWare ESX server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Begin installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Driver disk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Partitioning of console OS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Network setup of console OS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Define root password . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Security certificate of VMware ESX server . . . . . . . . . . . . . . . . . . . . . 234
Login on VMware management console . . . . . . . . . . . . . . . . . . . . . . . 234
VMware ESX server configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
VMware ESX server management. . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
Boot configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
Initial device allocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Extension required to use fail-over driver . . . . . . . . . . . . . . . . . . . . . . 239
Final assignment of devices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
Settings for FCal storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Update VMkernel flag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Volumes are visible . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Implementing Linux with IBM Disk Storage
11-18
11-19
11-20
11-21
11-22
11-23
11-24
11-25
11-26
11-27
11-28
11-29
11-30
11-31
11-32
11-33
11-34
A-1
11-35
11-36
11-37
Create new partition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Expert mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Swap for VMware (more logical than physical memory) . . . . . . . . . . . 244
Define VM basic settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Define VM SCSI disks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Define VM networking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
Define VM CD-ROM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
Define VM floppy drive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
Define VM miscellaneous options and create VM . . . . . . . . . . . . . . . . 247
Installation of remote console and launch dialog . . . . . . . . . . . . . . . . . 248
Linux Remote Console: connecting to VMware server . . . . . . . . . . . . 248
Linux Remote Console: select configuration . . . . . . . . . . . . . . . . . . . . 249
Linux Remote Console. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Linux Remote Console: Start guest OS installation . . . . . . . . . . . . . . . 250
Connect to different ISO-file part 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Connect to different ISO-file part 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Active VM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
/proc/partitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Content of /proc/scsi/scsi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
sg_scan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
sg_inq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Figures
xi
xii
Implementing Linux with IBM Disk Storage
Notices
This information was developed for products and services offered in the U.S.A.
IBM may not offer the products, services, or features discussed in this document in other countries. Consult
your local IBM representative for information on the products and services currently available in your area.
Any reference to an IBM product, program, or service is not intended to state or imply that only that IBM
product, program, or service may be used. Any functionally equivalent product, program, or service that
does not infringe any IBM intellectual property right may be used instead. However, it is the user's
responsibility to evaluate and verify the operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter described in this document.
The furnishing of this document does not give you any license to these patents. You can send license
inquiries, in writing, to:
IBM Director of Licensing, IBM Corporation, North Castle Drive Armonk, NY 10504-1785 U.S.A.
The following paragraph does not apply to the United Kingdom or any other country where such provisions
are inconsistent with local law: INTERNATIONAL BUSINESS MACHINES CORPORATION PROVIDES
THIS PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED,
INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT,
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer
of express or implied warranties in certain transactions, therefore, this statement may not apply to you.
This information could include technical inaccuracies or typographical errors. Changes are periodically made
to the information herein; these changes will be incorporated in new editions of the publication. IBM may
make improvements and/or changes in the product(s) and/or the program(s) described in this publication at
any time without notice.
Any references in this information to non-IBM Web sites are provided for convenience only and do not in any
manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of the
materials for this IBM product and use of those Web sites is at your own risk.
IBM may use or distribute any of the information you supply in any way it believes appropriate without
incurring any obligation to you.
Information concerning non-IBM products was obtained from the suppliers of those products, their published
announcements or other publicly available sources. IBM has not tested those products and cannot confirm
the accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on
the capabilities of non-IBM products should be addressed to the suppliers of those products.
This information contains examples of data and reports used in daily business operations. To illustrate them
as completely as possible, the examples include the names of individuals, companies, brands, and products.
All of these names are fictitious and any similarity to the names and addresses used by an actual business
enterprise is entirely coincidental.
COPYRIGHT LICENSE:
This information contains sample application programs in source language, which illustrates programming
techniques on various operating platforms. You may copy, modify, and distribute these sample programs in
any form without payment to IBM, for the purposes of developing, using, marketing or distributing application
programs conforming to the application programming interface for the operating platform for which the
sample programs are written. These examples have not been thoroughly tested under all conditions. IBM,
therefore, cannot guarantee or imply reliability, serviceability, or function of these programs. You may copy,
modify, and distribute these sample programs in any form without payment to IBM for the purposes of
developing, using, marketing, or distributing application programs conforming to IBM's application
programming interfaces.
© Copyright IBM Corp. 2001 2003. All rights reserved.
xiii
Trademarks
The following terms are trademarks of the International Business Machines Corporation in the United States,
other countries, or both:
AIX 5L
AIX®
AS/400®
Balance®
DB2®
developerWorks™
ECKD™
Enterprise Storage Server™
ESCON®
Everyplace™
FICON™
FlashCopy®
IBM.COM™
IBM®
iSeries™
Lotus®
Notes®
NUMA-Q®
OS/2®
OS/390®
Parallel Sysplex®
Perform™
pSeries™
Redbooks (logo)
™
Redbooks™
RS/6000®
S/390 Parallel Enterprise
Server™
S/390 Parallel Enterprise
Server™
S/390®
Seascape®
ServeRAID™
ServerProven®
SP1®
Tivoli®
TotalStorage™
™
VM/ESA®
X-Architecture™
xSeries™
z/Architecture™
z/OS™
z/VM™
zSeries™
The following terms are trademarks of other companies:
Intel, Intel Inside (logos), MMX, and Pentium are trademarks of Intel Corporation in the United States, other
countries, or both.
Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the
United States, other countries, or both.
Java and all Java-based trademarks and logos are trademarks or registered trademarks of Sun
Microsystems, Inc. in the United States, other countries, or both.
UNIX is a registered trademark of The Open Group in the United States and other countries.
SET, SET Secure Electronic Transaction, and the SET Logo are trademarks owned by SET Secure
Electronic Transaction LLC.
Other company, product, and service names may be trademarks or service marks of others.
xiv
Implementing Linux with IBM Disk Storage
Preface
This IBM® Redbook explains the considerations, requirements, pitfalls, and
possibilities when implementing Linux with IBM disk storage products.
This redbook presents the reader with a practical overview of the tasks involved
in installing Linux on a variety of IBM Eserver platforms.
It also introduces the people who are already familiar with Linux to the IBM disk
storage products.
Specifically, we worked with the Enterprise Storage Server™ (ESS) and the
FAStT Storage Server, covering their installation and configuration with Linux
distributions on the IBM eserver xSeries™, zSeries™, and pSeries™ models.
Configurations and practical experiments conducted by a team of professionals
at the ITSO, San Jose Center, who implemented these Linux environments are
documented in this book. We discuss the implementation of Linux from a storage
perspective. It is expected that people who wish to work with Linux and storage
subsystems will first become familiar with Linux and its installation
documentation and support sources. We provide the basic steps required to get
the storage subsystems in a state ready for Linux host attachment, and storage
specific steps on preparing the Linux host for attachment to the storage
subsystem. We also provide pointers and references to the documents and Web
sites that will be of interest at the time you will be doing your implementation.
IT specialists in the field will find this book helpful as a starting point and
reference when implementing Linux using the IBM disk storage servers.
Support considerations
Note: Before starting your Linux implementation activities using this redbook,
you should check the latest availability status and documentation in respect to
Linux, for the products and functions presented in this book.
You can consult your IBM Field Technical Support Specialist for the general
support available. You can also find support information at the following Web site:
: http://www.storage.ibm.com/hardsoft/products/
© Copyright IBM Corp. 2001 2003. All rights reserved.
xv
The team that wrote this Redbook
This redbook was produced by a team of specialists from around the world
working at the International Technical Support Organization, San Jose Center.
Bertrand Dufrasne is a Certified Consulting I/T Specialist and
Project Leader for Disk Storage Systems at the International
Technical Support Organization, San Jose Center. He has
worked at IBM for 21 years in many IT areas. Before joining the
ITSO he worked for IBM Global Services in the US as an IT
Architect. He holds a degree in Electrical Engineering.
Ronald Annuss works as an IT Specialist for IBM in Germany.
He holds a diploma in Geodesy from the Technical University
Berlin. He joined IBM in 1999 and has 8 years of experience
with Linux. He is a RHCE and is working on Linux projects in
Germany with zSeries and xSeries customers. He also is the
co-author of a redbook and a Redpaper on Linux for zSeries.
James Goodwin is a Senior IT Specialist with Americas’
Advanced Technical Support for Storage. He regularly
develops and presents training materials for IBM Storage
products. He has nearly 20 years of experience with open
systems. He holds a degree in Mechanical Engineering from
the University of New Mexico and has expertise in UNIX-like
systems and storage. Prior to joining the ATS group, he worked
in technical support for IBM NUMA-Q®.
Paul McWatt is an EMEA xSeries Advanced Technical Support
Engineer based in Scotland. He has worked at IBM for 6 years
and has 10 years experience of IBM Intel based servers. He is
an RHCE and Advisory IT Specialist. His areas of expertise
include Linux, storage, and High Availability Clustering. This is
Paul’s first redbook, but he has written two papers on similar
subjects
Arwed Tschoeke is an IBM zSeries Systems Engineer,
located in Hamburg, Germany. He worked for four years in the
xSeries presales support team specializing in Linux and
MSCS. He currently focuses on zOS and cross platform
solutions. He holds a degree in Physics from the University of
Kaiserslautern, Germany.
xvi
Implementing Linux with IBM Disk Storage
Thanks to the following people for their contributions to this project:
Bob Haimowitz
Roy Costa
International Technical Support Organization, Poughkeepsie Center
Chuck Grimm
IBM Technical Support Marketing Lead
Richard Heffel
IBM Storage Open Systems Validation Lab
Franck Excoffier
IBM Storage Open Systems Validation Lab
Rainer Wolafka
IBM Storage Open Systems Validation Lab
Timothy Pepper
IBM Storage Development
Mark S Fleming
IBM Storage Systems Level Test Lab
Nick Davis
IBM EMEA Linux Solutions Sales Manager, xSeries
Shawn Andrews
IBM x-Series World Wide Level 3 Support
Christopher M McCann
IBM xSeries World Wide Level 2 Support
Silvio Erdenberger
IBM xSeries Presales Technical Support Germany
Wendy Hung
IBM NOS technology support
Mic Watkins
IBM FAStT Storage Products Development
Marco Ferretti
IBM EMEA xSeries ATS - Education
James Gallagher
IBM pSeries Development Lab
Preface
xvii
Tan Truong
IBM pSeries Development Lab
Felica A Turner
IBM Raleigh - NOS Technology Support
Become a published author
Join us for a two- to six-week residency program! Help write an IBM Redbook
dealing with specific products or solutions, while getting hands-on experience
with leading-edge technologies. You'll team with IBM technical professionals,
Business Partners and/or customers.
Your efforts will help increase product acceptance and customer satisfaction. As
a bonus, you'll develop a network of contacts in IBM development labs, and
increase your productivity and marketability.
Find out more about the residency program, browse the residency index, and
apply online at:
ibm.com/redbooks/residencies.html
Comments welcome
Your comments are important to us!
We want our Redbooks™ to be as helpful as possible. Send us your comments
about this or other Redbooks in one of the following ways:
򐂰 Use the online Contact us review redbook form found at:
ibm.com/redbooks
򐂰 Send your comments in an Internet note to:
[email protected]
򐂰 Mail your comments to:
IBM Corporation, International Technical Support Organization
Dept. QXXE Building 80-E2
650 Harry Road
San Jose, California 95120-6099
xviii
Implementing Linux with IBM Disk Storage
Summary of changes
This section describes the technical changes made in this edition of the book and
in previous editions. This edition may also include minor corrections and editorial
changes that are not identified.
Summary of changes
for SG24-6261-01
for Implementing Linux with IBM Disk Storage
as created or updated on June 27, 2003.
May 2003, Second Edition
This revision reflects the addition, deletion, or modification of new and changed
information described below.
New information
򐂰 Linux LPAR installation for IBM
ESS
򐂰 Linux implementation on IBM
򐂰 Linux implementation on IBM
򐂰 VMware ESX
򐂰 BladeCenter SAN utility
zSeries and FCP attachment to the
r BladeCenter
pSeries
Changed information
򐂰 Covers Red Hat Enterprise Linux Advanced Server 2.1 and Suse Linux
Enterprise Server (SLES) 8.
򐂰 FAStT Storage Manager 8.3 and ESS Specialist
© Copyright IBM Corp. 2001 2003. All rights reserved.
xix
xx
Implementing Linux with IBM Disk Storage
1
Chapter 1.
Introduction to Linux
This chapter provides a brief introduction to Linux:
򐂰
򐂰
򐂰
򐂰
򐂰
Historical perspective
Open source
Linux distributions
IBM’s commitment to Linux
Summary
© Copyright IBM Corp. 2001 2003. All rights reserved.
1
1.1 Historical perspective
Years ago when the time came to install an operating system, a good plan was to
have some activity to occupy one’s time during the tedious phases. With the
advent of fast processors, high-bandwidth networks, and high-speed drives for
installation media, the luxury of “popping in a tape and going out to lunch” has
been reduced to “kickstart the installation and read the splash screens for ten
minutes.” Readers may choose to save this chapter as a possible diversion
during these ever-shorter installation periods.
Nearly everything one reads about Linux nowadays begins with the ritual
invocation of “... an operating system created in 1991 as a hobby by Linus
Torvalds at the University of Helsinki...” While this is true, it does not do justice to
the significance of the work and its broad implications for today’s implementers.
We know that Linux refers to a UNIX-like operating system, so let us begin with a
brief overview of the development of portable operating systems and open
source, to see the context in which Linux evolved.
1.1.1 UNIX and the culture of collaboration
In 1969, several AT&T Bell Labs employees1 began work on an operating system
(OS) that would come to be called UNIX. A significant novelty of their
development was that the OS was portable across hardware platforms. Prior to
this, it was more typical to emulate the (usually) older hardware on a new system,
so the OS would run unchanged, or to rewrite the operating system completely
for the alternate hardware. Such portability was achievable only through writing
most of the OS in a higher level language “above” the hardware. Unlike an OS
written in assembly language for a particular architecture, this abstraction2 and
the language (“C”) they developed to implement it permitted the study of their OS
without much regard to hardware specifics, and by 1976 UNIX was being taught
in classes on operating systems at the university3 level.
At the time, AT&T, for various legal reasons, was permitting free academic
access to the source code to UNIX while charging over $20,000 (in 1976 dollars!)
for commercial or government access. AT&T halted publication of the source
code in university texts as this revealed proprietary Bell Labs code. The era of
collaborative programming had arrived.
1
Ken Thompson, Dennis Ritchie, and J.F. Ossanna
For an alternative view of this principle, see The Linux Edge, Linus Torvalds, in Open Sources:
Voices from the Open Source Revolution, O’Reilly, 1999 ISBN 1-56592-582-3
3
Prof. John Lions, University of New South Wales, Australia
2
2
Implementing Linux with IBM Disk Storage
1.1.2 GNU and free software
In the US, the academic variant of interest became the Berkeley Systems
Distribution (BSD),4 where virtual memory and networking were added. These
advancements permitted large collaborative projects with contributors being
scattered throughout the world. Lawsuits eventually ensued among AT&T, the
Regents of the University of California, and other parties over access to and
distribution of the OS source code. Such constraints on intellectual property
rights to code, along with commercialization of academic artificial intelligence
projects in the late 1970s and early 1980s, provided strong motivation for one
researcher5 from the Artificial Intelligence Laboratories at the Massachusetts
Institute of Technology to write an operating system that was both portable and
also would be licensed in a manner that would prevent its eventual constraint by
intellectual property claims.
The new OS was to be named GNU, a recursive acronym for “Gnu’s Not UNIX.”
This work would be “copylifted” instead of copyrighted; licensed under the GNU
General Public License6 (GPL), which stipulates that all programs run under or
derived from GNU must have their source code published freely, relinquishing
rights of control while retaining rights of ownership. This was the birth of free (as
in freedom) software, in contrast to software in the public domain.
By this time, vendors such as Sun, Hewlett-Packard, and IBM had proprietary
commercial offerings derived from licensed AT&T UNIX that were gaining
popularity with corporate customers. The nascent GNU development effort
began by making tools such as editors, compilers and file utilities available in
source form that could be compiled and executed on any platform, standardizing
and improving upon those offered by commercial vendors. Around 1990,
programmers had contributed a nearly complete operating environment to GNU,
with the exception of a kernel. The GNU kernel was to be based on a microkernel
architecture for improved portability. This approach required an (arguably better)
architecture completely different from that of a monolithic kernel. The GNU kernel
project is known as the Hurd. In the words of its principal architect:
“The Hurd project was begun before Linux was a twinkle in Linus Torvalds'
eye, but because it is a more challenging task, and because we were less
adept at mobilizing large-scale volunteer excitement, the Linux kernel was
developed and deployed much sooner.”7
4
Twenty Years of Berkeley UNIX: From AT&T-Owned to Freely Redistributable, Marshall Kirk
McKusick, in Open Sources: Voices from the Open Source Revolution, O’reilly, 1999 1-56592-582-3
5 The GNU Operating System and the Free Software Movement, Richard M. Stallman, ibid.
6 Details at the Free Software Foundation, http://www.fsf.org/licenses/licenses.html
7
Thomas Bushnell, BSG, in a letter to Technology Review, March/April 1999
Chapter 1. Introduction to Linux
3
1.1.3 Linux
With the withdrawal of AT&T source code from the university environment, and
the BSD operating system mired in legal challenges by AT&T to its claim to be
unencumbered by proprietary code, a small scale, UNIX-like skeleton of an
operating system8 called Minix was published in a text to be used as a teaching
tool. It is here that Linus Torvalds enters the story. He decided to write a
UNIX-like OS with improved functionality over that of Minix to run on readily
available personal computers. His purpose was to teach himself the C language
and improve his understanding of operating system design. He and colleague
Lars Wirzenius published their source code under the GPL on the Internet for
public comment, and Linux was born.
Linux was a kernel without utilities, GNU was an operating environment lacking a
finished kernel, and unencumbered non-kernel BSD pieces were available to
complete the picture. In short order the components were combined with
installation and maintenance tools and made available by distributors; the first of
serious note being Slackware9 in 1993, followed by many others, making the
GNU/Linux (or simply Linux, as it has since come to be known) combination
readily available. In only a few years, a worldwide Linux community10 has
evolved, comprised of programmers and users attracted by the reliability and
flexibility of this “free” operating system.
The term “open source” began to replace the term “free software” as the
commercial adoption of GNU/Linux grew. There is in fact a difference, upon
which hinges the fate of commercial ventures in this arena.11
1.2 Open source
There is a difference between open source and the GNU General Public License
as noted on the GNU Web site (http://www.gnu.org). Linux was developed
under GNU, which has articulated a philosophy that defines “free code” – the
user’s right to use the code – rather than defining what they cannot do, which is
the case with proprietary software. This license allows the user to alter, distribute,
and even sell the code covered under its license as long as they allow those to
whom they distribute the code to do the same.
8
Dr. Andrew S. Tannenbaum, Vrije Universiteit, Amsterdam, The Netherlands
compiled by Patrick Volkerding
10 http://counter.li.org estimates 18 million Linux users as of April 2003
11 e.g., X386 vs. XFree86 - http://www.xfree86.org/pipermail/forum/2003-March/000191.html or
the evolution of Cygnus Support
9
4
Implementing Linux with IBM Disk Storage
The General Public License promotes free code on the GNU Web page. It also
provides protection for the developer and prevents a user from altering the code
and then asserting proprietorship over the code. This does not mean the code
cannot be sold. According to the GNU Web site,12 “free software” allows a user
to run, copy, distribute, study, change, and improve the software. It must also be
available for commercial use.
Standards enable communication among heterogeneous software and systems.
Open source can be viewed as a manifestation of this process, and the process
itself as a necessity for extending the development of inter-networking. When
there is a need for new software or hardware support, or a defect is discovered
and reported, the software (creation or correction) can be swiftly done by the
user who required the changes (or by the original author), with no need for a
design review, cost analysis, or other impositions of a centralized development
structure. This is made possible by open source code. IBM has recognized the
efficacy of this community and sees the benefit of the rapid and innovative
development of robust and stable code to provide the enabling layer for
e-business applications.
As a result of the evolutionary development of Linux, pieces of the code are
located on various Web sites. Without some integration, it is difficult to install and
upgrade the product, keep track of module dependencies, and acquire drivers for
the hardware. Distributions provide coherent bundles of software in one location
that furnish the capabilities needed for a usable server or desktop. Generally, the
products of different distributors have a lot of things in common, but there may be
particular features or functions that you require that are not available from a given
distributor.
1.3 Linux distributions
Linux is available in many configurations from a variety of distributors. Linux
advocates have strong preferences for one distribution over the other.
Distributions from, for example, Mandrake,13 Debian,14 and gentoo15 are
presently available and offer their own advantages in package management,
installation, development environment, and ease of use of various tools.
The distributions we used in our investigations were the so-called enterprise
distributions, one from Red Hat Software Inc.,16 and the other from SuSE
Holding AG.17 Both of these distributors also offer workstation distributions,
12
13
14
15
16
http://www.gnu.org/philosophy/free-sw.html
http://www.mandrake.com/
http://www.debian.org/
http://www.gentoo.org/
http://ww.redhat.com/
Chapter 1. Introduction to Linux
5
whose relative merits or disadvantages (along with those of many other
distributors) are discussed at length on the Internet.
1.3.1 Red Hat Enterprise Linux AS (formerly Advanced Server)
Enterprise Linux AS (formerly Red Hat Linux Advanced Server) is the core
operating system and infrastructure enterprise Linux solution from Red Hat,
supporting up to 8 CPUs and 16 GB of main memory, and is certified by DISA
(US Defense Information Systems Agency) as COE (Common Operating
Environment) compliant. It features High Availability Clustering and IP load
balancing capabilities, asynchronous I/O support, Linux Standard Base interface
conformity, improved SMP integration and reduced I/O-memory copy overhead.
Red Hat offers support and maintenance services for their enterprise
distributions.
1.3.2 SuSE Linux Enterprise Edition
SuSE Linux Enterprise Server (SLES) is a package distribution of UnitedLinux
(see 1.3.3, “UnitedLinux) intended for server application. SuSE distributions tend
towards standardized core content as defined by the Linux Standards Base
(LSB).18 SuSE also offers support and maintenance services for their enterprise
distributions.
1.3.3 UnitedLinux
UnitedLinux19 is a partnership of industry-leading Linux companies combining
their intellectual property, sales, support, and marketing expertise to produce a
uniform distribution of Linux designed for business.
Key elements of the UnitedLinux distribution include POSIX standard
asynchronous I/O (AIO), raw I/O enhancements that provide high-bandwidth,
low-overhead SCSI disk I/O, and direct I/O that moves data directly between the
userspace buffer and the device performing the I/O, avoiding expensive copy
operations, and bypassing the operating system's page cache.
Other functionality in focus includes hyper-threading to enable multi-threaded
server software applications to execute threads in parallel within each individual
server processor; large memory support to take advantage of the Intel Physical
Address Extension to support up to 64 GB of physical RAM, and the full 4 GB of
virtual addressing space per process; Internet Protocol Version 6 (IPv6), the next
generation protocol designed by the IETF to replace the current version Internet
17
6
18
http://www.suse.com/
http://www.opengroup.org/lsb/cert/
19
http://www.unitedlinux.com
Implementing Linux with IBM Disk Storage
protocol; and LDAPv3, the latest available directory services protocol for better
user base management and application integration for mail servers and
authentication servers, for instance.
1.4 IBM’s commitment to Linux
The history of IBM’s involvement with Linux is so large as to be outside the scope
of this work. IBM is not a distributor of Linux. IBM is a contributor to Linux, a
supplier of servers that run Linux, and a provider of support for customers who
choose to use Linux. IBM’s commitment to Linux may best be illustrated with
some of the many notable highlights over the last several years.
As early as March 1999, IBM had announced key alliances, products, and global
support for Linux, including distributors Caldera Systems Inc., (now the SCO
Group20), Pacific HiTech Inc., (now Turbolinux, Inc.21), SuSE Holding AG (all of
which later founded UnitedLinux along with distributor Conectiva S.A.22), and
Red Hat Software Inc.23 The IBM Thinkpad 600 became the first notebook
computer listed as supported hardware by Red Hat.
Linux made its official debut on S/390® servers in May 2000, with commercial
deployment by a major ISP for server consolidation following shortly thereafter. In
June, SuSE announced Enterprise Linux for the RS6000.
In 2001, IBM won top honors at LinuxWorld for zSeries and iSeries™ systems
running Linux, completing the availability of Linux across the entire IBM eServer
product line. Complementing this, was the announcement of a broad range of
storage options for Linux servers, including the Enterprise Storage Server (ESS)
and Fibre Array Storage System (FAStT) external storage arrays. IBM’s Linux
supercomputer systems were deployed in research and also in oil and gas
exploration, and prepackaged Linux clusters for e-business made their debut.
In 2002, continued emphasis on robust computing for small and medium
businesses, server consolidation, retail, network infrastructure, advertising,
telecom, and life sciences strengthened IBM’s position as the most committed
Linux player across all market segments. By December 2002 IBM had made the
IBM eServer p630 available; the first pSeries system dedicated to Linux support.
Internally, IBM began migrating its own business-critical services to Linux,
demonstrating its indisputable commitment, and also demonstrating the
scalability of Linux e-business solutions.
20
21
22
23
http://www.sco.com/
http://ww.turbolinux.com/
http://www.conectiva.com/
See press releases at http://www-916.ibm.com/press/prnews.nsf/homepage
Chapter 1. Introduction to Linux
7
Opening in 2003 were the announcement of the new Linux efforts, ranging from
Linux supercomputing on demand, to bringing Linux to consumer electronics
such as PDAs and handheld computers through the definition of a reference
installation and IBM software for Linux (Websphere Micro Environment, Power
Manager, DB2® Everyplace™, Tivoli® Device Manager, and IBM’s Service
Manager Framework).
In February 2003 IBM announced its commitment to “work with the Linux
community to enter the Common Criteria (CC) certification process for the Linux
operating system early this year, and proceed with a progressive plan for
certifying Linux at increasing security levels through 2003 and 2004.” With its
delivery of enterprise middleware including DB2, Websphere, Tivoli and Lotus®,
along with robust, secure, and scalable platforms, IBM also demonstrates its
commitment to Linux in government, thus completely spanning the marketplace
with pervasive support for Linux at every level.
1.5 Summary
Linux is an efficient multi-tasking operating system that can run on a minimal
amount of hardware by comparison to today’s standards. It can run very well on a
486 processor, for example. It is a multi-user system, and as such, offers various
levels of built-in security. It is also a stable system that is capable of sustained
up-time. For additional information, some further reading is noted:
The Evolution of the UNIX Time-sharing System, Dennis Ritchie, ca. 1979
(http://cm.bell-labs.com/cm/cs/who/dmr/hist.html)
Overview of the GNU Project, Free Software Foundation, ca. 1998
(http://www.gnu.org/gnu/gnu-history.html)
A Brief History of Hackerdom, Eric S. Raymond, ca. 1998
http://www.tuxedo.org/~esr/writings/hacker-history/
For more details on Linux in general, see:
򐂰 Linux Online offers a large amount of information on available applications,
distributions, and locations for downloading the code (for free),
documentation, education (including online courses), and information
regarding hardware as well as a variety of other information:
http://www.linux.org/apps/index.html
򐂰 The Linux Kernel Archives site is where the kernel sources can be
downloaded from (http://www.kernel.org/) but it also contains links to
various other related locations.
8
Implementing Linux with IBM Disk Storage
򐂰 The Linux Installation “HOWTO” site is sponsored by the “Linux
Documentation Project” (LDP). LDP has free documentation and HOWTO
documents that are short guides to do many of the tasks involved with Linux.
This particular link is how to install Linux:
http://www.linuxdoc.org/HOWTO/Installation-HOWTO/
򐂰 Linux International offers a nice description of the technical merits of the
operating system at: http://www.li.org
Chapter 1. Introduction to Linux
9
10
Implementing Linux with IBM Disk Storage
2
Chapter 2.
Introduction to IBM
TotalStorage
This chapter provides a brief overview and positioning of IBM TotalStorage™
disk products that can be used and managed in a Linux environment:
򐂰 IBM TotalStorage Enterprise Storage Server
򐂰 IBM TotalStorage FAStT Storage Server
We provide additional details for the models that we used during our
experimentations, which are the ESS 800 and the FAStT 700, respectively.
© Copyright IBM Corp. 2001 2003. All rights reserved.
11
2.1 Enterprise Storage Server
This section introduces the IBM TotalStorage Enterprise Storage Server and
discusses some of the benefits that can be achieved when using it. For more
detailed information, please refer to the redbook: IBM TotalStorage Enterprise
Storage Server Model 800, SG24-6424.
2.1.1 ESS Overview
The IBM TotalStorage Enterprise Storage Server (ESS) is IBM’s most powerful
disk storage server, developed using IBM Seascape® architecture. The ESS
provides unmatchable functions for all the server family of e-business servers,
and also for the non-IBM (that is, Intel-based and UNIX-based) families of
servers. Across all of these environments, the ESS features unique capabilities
that allow it to meet the most demanding requirements of performance, capacity,
and data availability that the computing business may require.
The Seascape architecture is the key to the development of IBM’s storage
products. Seascape allows IBM to take the best of the technologies developed
by the many IBM laboratories and integrate them, producing flexible and
upgradeable storage solutions. This Seascape architecture design has allowed
the IBM TotalStorage Enterprise Storage Server to evolve from the initial E
models to the succeeding F models, and to the more recent 800 models, each
featuring new, more powerful hardware and functional enhancements.
To meet the unique requirements of e-business, where massive swings in the
demands placed on systems are common, and continuous operation is
imperative, demands very high-performance, intelligent storage technologies and
systems that can support any server application. The IBM TotalStorage
Enterprise Storage Server has set new standards in function, performance, and
scalability in these most challenging environments.
Figure 2-1 on page 13 shows a photograph of an ESS Model 800 with the front
covers removed. At the top of the frame are the disk drives, and immediately
under them are the processor drawers that hold the cluster SMP processors. Just
below the processor drawers are the I/O drawers that hold the SSA device
adapters that connect to the SSA loops. Just below the I/O drawers are the host
adapter bays that hold the host adapters. At the bottom of the frame are the AC
power supplies and batteries.
The ESS in this photo has two cages holding the disk drives (DDMs). If the
capacity of this ESS was 64 or fewer disk drives, then the top right side of this
ESS would have an empty cage in it. The photo clearly shows the two clusters,
one on each side of the frame.
12
Implementing Linux with IBM Disk Storage
Figure 2-1 Photograph of ESS Model 800
Between the two cages of DDMs is an operator panel that includes an
emergency power switch, local power switch, power indicator lights, and
message/error indicator lights.
For larger configurations, the ESS base enclosure attaches to an expansion
enclosure rack that is the same size as the base ESS, and stands next to the
ESS base frame.
2.1.2 ESS features and benefits
The ESS set a new standard for storage servers back in 1999 when it was first
available, and since then it has evolved into the F models, and the recently
announced third-generation ESS Model 800.
The IBM TotalStorage Enterprise Storage Server Model 800 introduces
important changes that dramatically improve the overall value of ESS, and
provides a strong base for strategic Storage Area Network (SAN) initiatives.
Chapter 2. Introduction to IBM TotalStorage
13
Storage consolidation
The ESS attachment versatility —and large capacity— enable the data from
different platforms to be consolidated onto a single high-performance,
high-availability box. Storage consolidation can be the first step towards server
consolidation, reducing the number of boxes you have to manage, and allowing
you to flexibly add or assign capacity when needed. The IBM TotalStorage
Enterprise Storage Server supports all the major operating systems platforms,
from the complete set of IBM server series of e-business servers and IBM
NUMA-Q, to the non-IBM Intel-based servers, and the different variations of
UNIX based servers.
With a total capacity of more than 27 TB, and a diversified host attachment
capability —SCSI, ESCON®, and Fibre Channel/FICON™— the IBM
TotalStorage Enterprise Storage Server Model 800 provides outstanding
performance while consolidating the storage demands of the heterogeneous set
of server platforms that must be dealt with nowadays.
Performance
The IBM TotalStorage Enterprise Storage Server Model 800 integrates a new
generation of hardware from top to bottom, allowing it to deliver unprecedented
levels of performance and throughput. Key features that characterize the
performance enhancements of the ESS Model 800 are:
򐂰 New more powerful SSA device adapters
򐂰 Double CPI (Common Platform Interconnect) bandwidth
򐂰 Larger cache option (64 GB)
򐂰 Larger NVS (2 GB non-volatile storage) and bandwidth
򐂰 New, more powerful SMP dual active controller processors, with a Turbo
feature option
򐂰 2 Gb Fibre Channel/FICON server connectivity, doubling the bandwidth and
instantaneous data rate of previous host adapters
Efficient cache management and powerful back-end
The ESS is designed to provide the highest performance for different types of
workloads, even when mixing dissimilar workload demands. For example,
zSeries servers and open systems put very different workload demands on the
storage subsystem. A server like the zSeries typically has an I/O profile that is
very cache-friendly, and takes advantage of the cache efficiency. On the other
hand, an open system server does an I/O that can be very cache-unfriendly,
because most of the hits are solved in the host server buffers. For the zSeries
type of workload, the ESS has the option of a large cache (up to 64 GB) and
—most important — it has efficient cache algorithms. For the cache unfriendly
workloads, the ESS has a powerful back-end, with the SSA high performance
14
Implementing Linux with IBM Disk Storage
disk adapters providing high I/O parallelism and throughput for the ever-evolving
high-performance hard disk drives.
Data protection and remote copy functions
Many design characteristics and advanced functions of the IBM TotalStorage
Enterprise Storage Server Model 800 contribute to protect the data in an
effective manner.
Fault-tolerant design
The IBM TotalStorage Enterprise Storage Server is designed with no single point
of failure. It is a fault-tolerant storage subsystem, which can be maintained and
upgraded concurrently with user operation.
RAID 5 or RAID 10 data protection
With the IBM TotalStorage Enterprise Storage Server Model 800, there now is
the additional option of configuring the disk arrays in a RAID 10 disposition
(mirroring plus striping) in addition to the RAID 5 arrangement, which gives more
flexibility when selecting the redundancy technique for protecting the users’ data.
Peer-to-Peer Remote Copy (PPRC)
The Peer-to-Peer Remote Copy (PPRC) function is a hardware-based solution
for mirroring logical volumes from a primary site (the application site) onto the
volumes of a secondary site (the recovery site). PPRC is a remote copy solution
for the open systems servers and for the zSeries servers.
Two modes of PPRC are available with the IBM TotalStorage Enterprise Storage
Server Model 800:
򐂰 PPRC synchronous mode, for real-time mirroring between ESSs located up to
103 km apart
򐂰 PPRC Extended DIstance (PPRC-XD) mode, for non-synchronous data copy
over continental distances
Extended Remote Copy (XRC)
Extended Remote Copy (XRC) is a combined hardware and software remote
copy solution for the z/OS™ and OS/390® environments. The asynchronous
characteristics of XRC make it suitable for continental distance implementations.
Point-in-Time Copy function
Users still need to take backups to protect data from logical errors and disasters.
For all environments, taking backups of user data traditionally takes a
considerable amount of time. Usually, backups are taken outside prime shift
because of their duration and the consequent impact to normal operations.
Chapter 2. Introduction to IBM TotalStorage
15
Databases must be closed to create consistency and data integrity, and online
systems are normally shut down.
With the IBM TotalStorage Enterprise Storage Server Model 800, the backup
time has been reduced to a minimal amount of time when using the FlashCopy®
function. FlashCopy creates an instant point-in-time copy of data, and makes it
possible to access both the source and target copies immediately, thus allowing
the applications to resume with minimal disruption.
Storage Area Network (SAN)
The SAN strategy is to connect any server to any storage as shown in Figure 2-2.
As SANs migrate to 2 Gb technology, storage subsystems must exploit this more
powerful bandwidth. Keeping pace with the evolution of SAN technology, the IBM
TotalStorage Enterprise Storage Server Model 800 is introducing new 2 Gb Fibre
Channel/FICON host adapters for native server connectivity and SAN integration
These new 2 Gb Fibre Channel/FICON host adapters, which double the
bandwidth and instantaneous data rate of the previous adapters available with
the F Model, have one port with an LC connector for full-duplex data transfer over
long-wave or short-wave fiber links. These adapters support the SCSI-FCP
(Fibre Channel Protocol) and the FICON upper-level protocols.
16
Implementing Linux with IBM Disk Storage
LAN
LAN
Fibre Channel
SAN
ESCON
FICON
switches & directors
Any Server to
Any Storage
Tot alStorage
ESS
Figure 2-2 Storage area network (SAN)
Fabric support now includes the following equipment:
򐂰 IBM TotalStorage SAN switches IBM 3534 Model F08 and IBM 2109 Models
F16, F32, S08 and S16
򐂰 McDATA “Enterprise to Edge” Directors (IBM 2032 Model 064) for 2 Gb
FICON and FCP attachment (up to 64 ports)
򐂰 McDATA 16 and 12 port switches for FCP attachment (IBM 2031)
򐂰 INRANGE FC/9000 Director for FCP attachment (up to 64 and 128 ports) and
FICON attachment (up to 256 ports) — IBM 2042 Models 001 and 128
The ESS supports the Fibre Channel/FICON intermix on the INRANGE FC/9000
Fibre Channel Director and the McDATA ED-6064 Enterprise Fibre Channel
Director. With Fibre Channel/FICON intermix, both FCP and FICON upper-level
protocols can be supported within the same director on a port-by-port basis. This
new operational flexibility can help users to reduce costs with simplified asset
management and improved asset utilization.
Chapter 2. Introduction to IBM TotalStorage
17
The extensive connectivity capabilities make the ESS the unquestionable choice
when planning the SAN solution. For the complete list of the ESS fabric support,
please refer to:
http://www.storage.ibm.com/hardsoft/products/ess/supserver.htm
For a description of the IBM TotalStorage SAN products, please refer to:
http://www.storage.ibm.com/ibmsan/products/sanfabric.html
2.2 IBM Fibre Array Storage Technology (FAStT)
IBM Fibre Array Storage Technology (FAStT) solutions are designed to support
the large and growing data storage requirements of business-critical
applications.
The FAStT Storage Server is a RAID controller device that contains Fibre
Channel (FC) interfaces to connect the host systems and the disk drive
enclosures. The Storage Server provides high system availability through the use
of hot-swappable and redundant components. The Storage Server features two
RAID controller units, redundant power supplies, and fans. All these components
are hot-swappable, which assures excellent system availability. A fan or power
supply failure will not cause downtime, and such faults can be fixed while the
system remains operational. The same is true for a disk failure if fault-tolerant
RAID levels are used. With two RAID controller units and proper cabling, a RAID
controller or path failure will not cause loss of access to data. The disk
enclosures can be connected in a fully redundant manner, which provides a very
high level of availability. On the host side FC connections, you can use up to four
mini-hubs.
The FAStT Storage Sever can support high-end configurations with massive
storage capacities (up to 33 Tb per FAStT controller) and a large number of
heterogeneous host systems. It offers a high level of availability, performance,
and expandability.
2.2.1 FAStT models and expansion enclosure
At the time of writing this book, the FAStT900 and FAStT600 represented the
newest addition to the FAStT family, complementing and joining the range of the
earlier FAStT200, FAStT500, and FAStT700.
򐂰 The FAStT200 and FAStT600 are ideally placed for low-end mid-range setups
such as workgroup and small department consolidations.
18
Implementing Linux with IBM Disk Storage
򐂰 The FAStT500 and FAStT700 are considered mid-range and are aimed at
department storage consolidations, server clusters, and small to mid-range
SANs.
򐂰 The FAStT900 is considered an entry-level enterprise product positioned for
enterprise storage consolidation, mission critical databases, large SAN’s, and
high performance I/O.
The following tables show some of the differences between the FAStT storage
server range.
Table 2-1 Machine matrix
FAStT200
FAStT500
FAStT600
FAStT700
FAStT900
Machine Type
3542
3552
1722-60U/X
1742
1742-90U/X
Max Logical
Drives
512
512
1024
2048
2048
Host Port Link
Rate (Gb/sec)
1
1
2
1,2
1,2
Drive Port Link
Rate (Gb/sec)
1
1
2
1,2
1,2
Max Physical
Drives
66
100/220
42
224
224
Table 2-2 Performance comparisons
FAStT200
FAStT500
FAStT600
FAStT700
FAStT900
Burst I/O rate
cache reads
(512 bytes)
11,800 IOPS
60,000
55,000
110,000
148,000
Sustained I/O
rate disk reads
(4k)
4,610 IOPS
20,000
21,500
31,300 1 Gb
38,000 2 Gb
53,200
Sustained I/O
rate disk writes
(4k)
1,290 IOPS
5,200
6,010
5,980 1 Gb
8,500 2 Gb
10,900
Sustained
throughput
cache read
(512k)
190MB/s
400
--
424
800
Chapter 2. Introduction to IBM TotalStorage
19
FAStT200
FAStT500
FAStT600
FAStT700
FAStT900
Sustained
throughput
disk read
(512k)
170MB/s
370
380
380
772
Sustained
throughput
disk write
(512k)
105MB/s
240
320
240
530
Table 2-3 FAStT700 verses FAStT900
FAStT700
FAStT900
Processor
Celeron 566MHz
Pentium III 850MHz
Processor Memory
--
128MB
NVSRAM
32KB
128KB
Primary PCI Bus
32bit @ 33MHz
64bit @ 66MHz
Buffer Bus
0.5GB/s
1.6GB/s
2.2.2 FAStT700 features and benefits
For illustration and test purposes, during the writing of this book we used a
FAStT700 and EXP700.
For more detailed information about these models, see the following redbook:
IBM TotalStorage FAStT700 and Copy Services, SG24-6808.
The FAStT700 Storage Server has controllers which use the 2 Gbps Fibre
Channel standard on the host side, as well as on the drive side. It connects via
mini-hubs to the newer FAStT FC2-133 Host Bus Adapters (HBA) and the 2109
F16 Fibre Channel switch to give a full 2 Gbps fabric.
The FAStT700 attaches up to 220 FC disks via 22 EXP500 expansion units or up
to 224 FC disks via 16 EXP700 expansion units to provide scalability for easy
growth. To avoid single points of failure, it also supports high availability features
such as hot-swappable RAID controllers, two dual redundant FC disk loops, write
cache mirroring, redundant hot-swappable power supplies, fans, and dual AC line
cords.
Figure 2-3 shows a rear view of the FAStT700, illustrating some of its
components.
20
Implementing Linux with IBM Disk Storage
Host connections
Fan and communication
module
Power supply unit
Drive connections
Power supply unit
Figure 2-3 FAStT700 - Rear view
Using the latest Storage Manager software, the FAStT700 supports FlashCopy,
Dynamic Volume Expansion, and remote mirroring with controller based support
for up to 64 storage partitions. RAID levels 0,1,3,5, and 10 are supported, and for
performance, it includes a 2 GB battery backed cache (1 GB per controller).
Storage Manager
To manage the FAStT700 storage server, use the IBM FAStT Storage Manager
software. Refer to Chapter 9., “FAStT Storage Manager” on page 187 for details.
At a glance, this software allows you to:
򐂰 Configure arrays and logical drives
򐂰 Assign your logical drives into storage partitions
򐂰 Replace and rebuild failed disk drives
򐂰 Expand the size of arrays and logical volumes
򐂰 Convert from one RAID level to another
򐂰 Perform troubleshooting and management tasks, like checking the status of
FAStT Storage Server components, update the firmware or RAID controllers,
and similar actions
򐂰 Configure and manage FlashCopy Volumes and Remote Volume Mirroring
(FlashCopy and RVM are premium features that must be purchased)
Chapter 2. Introduction to IBM TotalStorage
21
It is also possible to use the serial interface and a terminal emulation utility.
However, this is only meant for advanced troubleshooting and management. It
should only be used when other management methods fail, and must be done
under the supervision of IBM level 2 support.
The FAStT Storage Manager software supports two premium features that can
be enabled by purchasing a premium feature key, as well as several new
standard features. Some of these features include:
򐂰 FlashCopy: A premium feature that supports the creation and management
of FlashCopy logical drives. A FlashCopy logical drive is a logical point-in-time
image of another logical drive, called a base logical drive, in the storage
subsystem. A FlashCopy is the logical equivalent of a complete physical copy,
but you create it much more quickly, and it requires less disk space. Because
FlashCopy is a host addressable logical drive, you can perform backups using
FlashCopy while the base drive remains online and user accessible. In
addition, you can write to the FlashCopy logical drive to perform application
testing or scenario development and analysis. The maximum FlashCopy
logical drives allowed is one half of the total logical drives supported by your
controller model.
򐂰 Remote Volume Mirroring: A premium feature that is used for online,
real-time replication of data between storage subsystems over a remote
distance. In the event of a disaster or unrecoverable error on one storage
subsystem, the Remote Volume Mirror option enables you to promote a
second storage subsystem to take responsibility for normal input/output (I/O)
operations. When RVM is enabled, the maximum number of logical drives per
storage subsystem is reduced.
򐂰 Dynamic logical drive expansion: Enables you to increase the capacity of
an existing logical drive
򐂰 2048 logical drive support: Enables you to increase the number of defined
logical drives up to 2048 for each storage subsystem
򐂰 Storage partitioning: Supports up to 64 storage partitions
Storage Partitioning
Storage Partitioning allows you to connect multiple host systems to the same
storage server. It is a way of assigning logical drives to specific host systems or
groups of hosts. This is known as LUN masking.
Logical drives in a storage partition are only visible and accessible by their
assigned group or individual hosts. Heterogeneous host support means that the
host systems can run different operating systems. However, be aware that all the
host systems within a particular storage partition must run the same operating
system because they will have unlimited access to all logical drives in this
22
Implementing Linux with IBM Disk Storage
partition. Therefore, the file systems on these logical drives must be compatible
with the host systems.
Chapter 2. Introduction to IBM TotalStorage
23
24
Implementing Linux with IBM Disk Storage
3
Chapter 3.
SuSE Linux on zSeries
This chapter describes the implementation of SuSE Linux Enterprise Server for
IBM zSeries Version 8 (SLES 8) and reviews disk storage attachment options
supported under Linux. Special focus is on the newly available Fibre Channel
Protocol (FCP) support for Linux on zSeries and the attachment of the IBM
TotalStorage Enterprise Storage Server (ESS) using Fibre Channel.
This chapter discusses the following:
򐂰
򐂰
򐂰
򐂰
򐂰
Introduction to zSeries, Linux and storage options
Hardware and software requirements
The test environment
Linux installation in a logical partition and as a z/VM™ guest
Attachment of ESS through Fibre Channel
© Copyright IBM Corp. 2001 2003. All rights reserved.
25
3.1 Introduction to zSeries and Linux
The IBM
zSeries is the successor of the S/390 series and IBM’s premier
server for e-business transaction workloads and has a long history in running
commercial applications with unmatched reliability and availability. In 1999 Linux
was ported to S/390 and became available in early 2000. Since then, many new
features and device support have been added and more, and more customers
are using Linux for S/390 and zSeries either for running new workloads or
consolidating servers.
While Linux for zSeries supports the Extended Count Key Data (ECKD™) format
on Direct Access Storage Devices (DASD) from the beginning, the support of
Fibre Channel attached devices was added recently in February 2003. The aim
of this book is to show the usage of IBM disk storage with Linux among the
different IBM eServer platforms; in this chapter we will focus on the new zSeries
Fibre Channel Protocol (FCP) support.
For more detailed information on zSeries and Linux in general, the reader should
refer to the following redbooks: Linux for S/390, SG24-4987, and Linux for IBM
zSeries and S/390: Distributions, SG24-6264, as well as to the SuSE installation
manual.
Linux on zSeries can run in three different modes:
򐂰 Native using the whole machine
– A native installation is rarely used since only one Linux system can run at
any one time on the machine.
򐂰 In a logical partition (LPAR)
– The zSeries processors support the partitioning of the native hardware
into logical partitions. Logical partitions are an allocation of the available
processor resource (either shared or dedicated), devices that are
dedicated but can be serially switched between partitions and an
allocation of memory. In LPAR mode, up to 15 operating system instances
can concurrently run on one machine without interfering with each other.
򐂰 As a guest system under the hypervisor z/VM
– When using z/VM as hypervisor to run operating systems as guests, only
the hardware resources limit the amount of concurrently running
instances.
The last two options are depicted in Figure 3-1
26
Implementing Linux with IBM Disk Storage
:
Linux for zSeries
z/VM Guests
Partitions
LPAR Hypervisor
PU
PU
PU
PU
PU
PU
MEMORY
^ zSeries z900/z800 hardware
Figure 3-1 Running Linux on zSeries in an LPAR or as z/VM guest
3.1.1 zSeries specific storage attachments
The I/O component of z/Architecture™, inherited from its predecessors such as
the ESA/390 architecture, is based on channels, control units, and devices.
Channels provide a well-defined interface between a server and the attached
control units. Originally implemented on parallel copper media, today's channels
such as ESCON® (Enterprise System Connection) and FICON™ (Fibre
Connection) use optical serial cables. Also, the parallel channel was a multi-drop
interface, while both ESCON and FICON are switched point-to-point
connections, extended to complex connection infrastructures via ESCON or
FICON switches or directors, as depicted in Figure 3-2.
ESCON
ESCON//
FICON
FICON
Storage
Controller
Switch
(Director)
zSeries
Channel
Adapter
ESCON
ESCON//
FICON
FICON
Storage Devices
Figure 3-2 Typical zSeries storage attachment concept
Chapter 3. SuSE Linux on zSeries
27
ESCON channels use a unique physical interface and transmission protocol.
FICON, on the other hand, is based on industry-standard Fibre Channel
lower-level protocols.
In both cases, however, the higher-level I/O protocol used by software, based on
channel programs consisting of channel command words (CCWs), is unique to
mainframes adhering to z/Architecture or its predecessors. For access to disk
storage, a specific set of CCWs is used, defined by the ECKD protocol. This
protocol describes not only the commands and responses exchanged across the
channel, but also the format of the data as it is recorded on the disk storage
medium. For tape, similar CCW-based command protocols exist, but there are no
associated access protocols for media such as DVDs or scanners, because the
necessary software device drivers and control units have never been provided.
zSeries and storage device addressing
With z/Architecture, software addresses a storage device using a 16-bit device
number, which uniquely identifies the device. Such a device number is mapped
to one or more physical paths to the device. This path is typically described by an
address quadruple consisting of:
1. Channel path identifier: (CHPID)
Identifies the channel that provides a path to the device
2. Physical link address:
Identifies a route through a switch or director
3. Control unit address:
Specifies the control unit
4. Unit address (UA):
Identifies the device
Please refer to the left side of Figure 3-6 on page 32.
In the case of a disk, this is often a logical device partitioned out of the disk space
provided by an array of physical disks. In order to provide redundancy and/or
increased I/O bandwidth, or to allow load balancing, there is typically more than
one physical path to a device. While these paths are specified using different
address quadruples, software still uses a single device number to address the
device, independent of the path that is chosen for any particular I/O request.
Another characteristic of the z/Architecture I/O scheme and the ECKD
architecture for disk, is the sophisticated support for sharing channels, control
units, and devices among multiple operating systems, which may run on the
same or different zSeries systems.
28
Implementing Linux with IBM Disk Storage
3.1.2 Distributed storage attachments
The attachment of storage controllers in the distributed world is predominantly
based on the SCSI standard, using a control-block-based command and status
protocol. Today, there is a clear distinction between the physical level and the
command/status level of the SCSI protocol.
FCP advantages
Based on this higher-level SCSI command/status protocol, a new standard has
been defined called FCP, which stands for Fibre Channel Protocol for SCSI.
FCP employs the SCSI command/status protocol on top of an underlying Fibre
Channel transmission protocol. Due to the superiority of the optical Fibre
Channel connection regarding speed, distance, and reliability, storage
attachments via FCP have the highest growth rate in today's distributed storage
world.
Traditionally, distributed storage controllers have been attached via parallel SCSI
cabling, which allowed only very limited distances between the server and the
controller. And although the SCSI architecture has provisions for physical
controller and device sharing, the length constraints of parallel SCSI cables
impose natural limits on this sharing capability. Also, there has been little need for
a capability to share controllers and devices among multiple operating systems
running concurrently on the same server, because such server virtualization
techniques are just starting to develop in the distributed world.
FCP storage addressing
Traditional SCSI devices have a simple addressing scheme: originally, the device
address, known as the target address was simply a number in the range 0-7
(extended to 0-15 in newer SCSI architecture). As shown in Figure 3-3 on
page 30, SCSI also supports the definition of Logical Unit Numbers (LUNs); each
LUN allows for a sub-addressing of the physical device.
Chapter 3. SuSE Linux on zSeries
29
C o n tro lle r
C o n tro lle r
H o s t A d a p te r
ID 2
ID 1
ID 7
D is k D riv e
D is k D riv e
D is k D riv e
D is k D riv e
LUN 0
LUN 1
LUN 0
LUN 1
Figure 3-3 SCSI addressing
For FCP, a new addressing scheme, based on World Wide Names (WWN) was
developed. Each node and each port (as an FCP port) on a node is assigned a
WWN, respectively known as World Wide Node Name (WWMN) or World Wide
Port Name (WWPN). This is illustrated in Figure 3-4.
End
End Node
Node
WWNN
WWNN
End
End Node
Node
WWNN
WWNN
Port,
W WPN , ID
Port,
WWPN
Sw
Switch
itch
WWNN
WWNN
Port,
WWPN
Port,
WWPN, ID
FCP LUN
FCP LUN
Port,
W WPN , ID
Port,
WWPN
Port,
WWPN
Port,
WW PN, ID
FCP LUN
Figure 3-4 World Wide Names
In addition to WWNN and WWPN names, any addressable unit within a node
(such as a disk drive) can be assigned a unit name, also know as LUN.
30
Implementing Linux with IBM Disk Storage
Linux FCP mapping
The format of a map entry is as follows:
0x6000 0x00000001: 0x5005076300c38550 0x00000000:0x524900000000000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
+-- FCP_LUN
|
|
|
+----------- LUN nr for Linux
|
|
|
(user asssigned)
|
|
+--------------------------- WWPN
|
+---------------------------------------- SCSI ID for Linux
|
(user assigned)
+--------------------------------------------------- Device nr for FCP
channnel
򐂰 The first element is the zSeries device number that must be defined in the
IOCDS, and be attached the FCP channel that is attached to the FCP switch.
򐂰 The second element is the SCSI target number; it is user assigned. Normal
usage is to start with address 1 and increment by 1 for each new WWPN
used.
򐂰 The third element is the WWPN of the device containing the LUN, as seen by
the FC switch.
򐂰 The fourth element is the LUN number to be used by Linux, and is also user
assigned.
򐂰 The fifth element is the LUN number assigned by the node controller.
Figure 3-5 illustrates how these elements are related to actual Linux device
names.
0x6000 0x00000001: 0x5005076300c38550 0x00000000:0x524900000000000
scsi/ host0/ bus0/ target1/ lun0/ disc
Figure 3-5 FCP to Linux device name mapping
Chapter 3. SuSE Linux on zSeries
31
3.1.3 zSeries-specific versus distributed storage controllers
Two types of storage controllers with Fibre Channel interfaces are currently
available for zSeries:
򐂰 Those supporting the z/Architecture specific FICON interface based on the
CCW architecture, with the ECKD command set for disk storage, and similar
sets of commands for tapes.
򐂰 Those supporting the SCSI-based FCP protocol, with SCSI command sets
specific to the device type, such as disk or tape.
Both types of controllers are normally based on the same hardware building
blocks. Their individual characteristics are achieved by different firmware
handling the host interface. The way the DASD addressing or device mapping is
handled, is illustrated in Figure 3-6 and is summarized in our discussion in 3.1.1,
“zSeries specific storage attachments” and 3.1.2, “Distributed storage
attachments”
Classical zSeries I/O
devnum
devnum
zSeries FCP
FCP (SCSI)
(SCSI) I/O
Host
Host
Program
Program
devnum
devnum
WWPN
WWPN
LUN
LUN
IOCP
IOCP
CHPID
Host
Host
Program
Program
IOCP
IOCP
CHPID
CU
UA
QDIO
Queues
Ficon/Escon
CHANNEL
FCP
CHANNEL
Ficon/Escon
SAN
Ficon/Escon
SAN
devnum
identifies
I/O device
CU
UA
devnum identifies
communication path
to FCP channel
WWPN (8 Byte) World Wide Port Name:
Identifies Controller /
Port (or SCSI Bridge).
LUN (8 Byte) Logical Unit Number:
Identifies I/O device.
Control
Unit
Control
Unit
Device
Figure 3-6 DASD addressing scheme
32
Implementing Linux with IBM Disk Storage
Device
3.1.4 zSeries and storage: Summary
To summarize, classical zSeries operating systems such as z/OS™ and z/VM™
were designed for use only with storage controllers that support the I/O protocols
defined by z/Architecture. This changed with the advent of Linux for zSeries,
since its storage I/O component is oriented toward SCSI protocols. Lacking the
capability to access SCSI-based storage devices on a zSeries server system, it
was necessary to add specific support to Linux for zSeries to enable it to function
in a CCW-based zSeries I/O environment. However, this additional layer in Linux
for zSeries is unique to zSeries, and does not allow it to exploit storage
subsystems and applications that are dependent on a SCSI attachment. For this
reason, an FCP attachment capability has been added to the z800 and z900
systems, allowing the attachment of SCSI-based storage controllers and
enabling Linux for zSeries to access these controllers in the Linux-standard
manner. The Figure 3-7 shows the differences/format translation between the
two access methods.
D
D A S D a c c e s s tra n s la tio n s
L in u x b u ffe r
ca che
L in u x b u ffe r
c ac he
D A S D d rive r
S C S I d riv e r
F IC O N
O pen FC P
ESS
ESS
DASD
DASD
b lo c k fo rm a t
tra n s la tio n
b lo c k < -> E C K D
E C K D fo rm a t
Figure 3-7 DASD access translations FICON verses FCP
3.2 System requirements
Linux for zSeries is supported on the z800 and z900 models and on the
predecessor S/390 models G5, G6, and MP3000, which are shown in Figure 3-8.
Chapter 3. SuSE Linux on zSeries
33
Figure 3-8 IBM eServer zSeries
3.2.1 Hardware requirements
In this section we review the hardware requirements for running Linux on
zSeries; and we successively look at the processor, memory, storage, and
network specific requirements.
Processor
The heart of zSeries and S/390 hardware is the Multi-Chip Module (MCM), which
contains up to 20 Processing Units (PU), commonly referred to on other
platforms as CPUs or engines. All PUs are identical, but can take on different
roles in the system. Each PU can be one of:
򐂰
򐂰
򐂰
򐂰
A central processor (CP), sometimes called a standard engine
A System Assist Processor (SAP)
An Internal Coupling Facility (ICF) for Parallel Sysplex®
An Integrated Facility for Linux (IFL), sometimes called an IFL engine
The IFL engine is a new type of PU reserved for Linux; it cannot run any other
operating systems. It is priced lower than standard engines, and its presence
does not affect the power ratings, hence, does not increase the cost of other
software installed on that hardware.
Linux needs at least 1 PU configured as CP or IFL, but scales up to 16
processors on zSeries. The PUs can be either shared with other partitions
(provided the LPARs are running Linux or z/VM) or dedicated.
34
Implementing Linux with IBM Disk Storage
Memory
The memory requirements are mostly dependent on the purpose of the server
system and type of applications you are going to deploy on this system. For the
installation, SuSE recommends the following:
򐂰 128+ MB: For text mode installation
򐂰 256+ MB: For graphical (X11) mode from nfs or smb source
򐂰 512+ MB: For installation using VNC (graphical mode remotely displayed in
Java-enabled Web browser) and FTP source
Storage
To hold the operating system, a certain amount of storage has to be in ECKD
format since booting (or Initial Program Load - IPL) from Fibre Channel attached
storage is not supported.
The types of zSeries DASD are referred to by their machine type and model
number. For the 3390s, there are five possible model numbers: 1, 2, 3, 9, and 27.
Models 3 and 9 are seen most commonly and are sometimes called a pack or
volume. In the past, these DASDs were large physical devices, but today they are
emulated by disk arrays with much larger capacities such as the ESS. A 3390
model 3 (or more simply, a 3390-3) has a capacity of approximately 2.8 GB, but
when formatted for Linux, that is reduced to about 2.3 GB. A 3390 model 9 has a
capacity of about 8.4 GB, but when formatted, that is reduced to about 7.0 GB.
For the operating system files, we need approximately 1.6 GB of DASD storage.
Including the requirements for swap and some free space; the equivalent of 1
3390-3 volume is sufficient for a basic installation of SLES 8.
To attach storage via Fibre Channel you need a zSeries FCP channel. A FCP
channel requires a FICON card (feature 2315 or 2318), or a FICON Express card
(feature 2319 or 2320). This is the same channel hardware used for FICON
channels, however, a different firmware load is required.
The type of firmware to be loaded into the FICON/FICON Express card, turning it
into either an FCP channel or into one of the FICON type channels (FC or FCP),
is controlled by the definition of the channel type for that particular channel in the
IOCP (CHPID statement). Thus, by defining FCP type channels in the IOCP, the
total number of FICON type channels that can be configured is reduced
accordingly.
The two channels residing on a single FICON/FICON Express card can be
configured individually, and each can be a different channel type.
For the LPAR installation of SLES 8 you can use the HMC CD-ROM or an FTP
server, or if that is not possible, you have to prepare a tape to IPL from during the
Chapter 3. SuSE Linux on zSeries
35
installation. Thus, you might need access to a tape unit such as a 3480, 3490, or
3590.
Network
Physical networking is typically done through one or more flavors of the Open
Systems Adapters (OSA). Please refer to 3.2.3, “Connection requirements” on
page 37.
3.2.2 Software requirements
The following MicroCode and software are required for the installation (as
described in the SLES installation manual).
Microcode level and APARs/fixes
For installation under VM, you need at least VM/ESA® 2.4. For the 64-bit zSeries
release you need z/VM 3.1 or higher. If you want to use Hipersockets under VM,
you need z/VM 4.2 or higher and on 2064/z900 you need microcode EC E26949
level 013 or higher.
For the installation of SuSE Linux Enterprise Server 8 on IBM S/390 or zSeries,
the following microcode levels and z/VM APARs are required:
OSA-Express QDIO
zSeries 900 GA3
Driver 3G, OSA microcode level 3.0A
MCLs: J11204.007 and J11204.008 (available May 03, 2002)
zSeries 900 GA2
Driver 3C, OSA microcode level: 2.26
MCLs: J10630.013 and J10630.014 (available May 20, 2002)
zSeries 800 GA1
Driver 3G, OSA microcode level 3.0A
MCLs: J11204.007 and J11204.008 (available May 03, 2002)
S/390 Parallel Enterprise Servers G5 and G6
Driver 26, OSA microcode level: 4.25
MCLs: F99904.032 and F99904.033 (available May 16, 2002)
VM/ESA and z/VM
z/VM 4.3
All necessary fixes and enhancements included.
z/VM 4.2
36
Implementing Linux with IBM Disk Storage
APAR: VM62938, PTF: UM30225
APAR: VM63034, PTF: UM30290
FCP support
FCP and SCSI controllers and devices can be accessed by Linux for zSeries with
the appropriate I/O driver support. Linux may run either natively in a logical
partition, or as a guest operating system under z/VM Version 4, Release 3.
Note: z/VM Version 4, Release 3 is required to support FCP for Linux guests.
However, z/VM itself does not support FCP devices.
3.2.3 Connection requirements
This section describes the connection requirements, such as console and
network access needed to install SuSE Linux for zSeries.
Access to the Hardware Management Console (HMC) or a network connection to
a virtual console in z/VM is needed to perform the installation. The intended use
of the console is solely to launch the Linux installation. After Linux is running, use
a Telnet connection directly to Linux for S/390 to logon to a Linux command shell
and other applications.
For the Linux system a TCP/IP network connection is required in order to get files
from the installation server and to Telnet into the Linux system. The connection
can be one of the following:
򐂰
򐂰
򐂰
򐂰
OSA (OSA-2, OSA Express)
CTC (virtual or real)
ESCON channels
PCI adapter (emulated I/O, only on MP3000)
All network connections require the correct setup on both Linux for zSeries and
the remote system, and a correct routing between both ends.
Since the installation is done via the network, you need an installation server,
which can access the installation CDs either via FTP, NFS, or SMB.
Installing SuSE Linux Enterprise Server via non-Linux based NFS or FTP, can
cause problems with the NFS/FTP server software. Especially Windows
standard FTP server can cause errors, so we generally recommend that you
install via SMB on these machines.
To connect to the SuSE Linux Enterprise Server installation system one of the
following methods is required:
Chapter 3. SuSE Linux on zSeries
37
Telnet or ssh with terminal emulation
ssh or Telnet are standard UNIX tools and should normally be present on
any UNIX or Linux system. For Windows, there is a Telnet and ssh client
called Putty. It is free to use and is included on CD 1 in the directory
/dosutils/putty. More information on Putty can be obtained at:
http://www.chiark.greenend.org.uk/~sgtatham/putty.html
VNC client
For Linux, a VNC client called vncviewer is included in SuSE Linux as part
of the VNC package. For Windows, a VNC client is included in the present
SuSE Linux Enterprise Server. You will find it in
/dosutils/vnc-3.3.3r9_x86_win32.tgz of CD 1. Alternatively, use the
VNC Java client and a Java-enabled Web browser.
X server
You will find a suitable X server implementation on any Linux or UNIX
workstation. There are many commercial X-Window environments for
Windows. Some of them can be downloaded as free trial versions. A trial
version of MI/X (MicroImages X Server) can be obtained at:
http://www.microimages.com/mix
3.3 Configuration used for this redbook
For our tests we used a z800 2066-004 connected to a SAN switch 2032-064
(McData) connected to the ESS 2105-800. The configuration is shown in
Figure 3-9.
The z800 was partitioned in to several LPARs, one of them running z/VM with
Linux as a guest and one of them running Linux natively. The OSA adapter and
FCP card are shared between those LPARs and can be used from either side
with the same settings. The relevant settings (in the IOCDS, generated by the
IOCP) for the FCP device are:
CHPID PATH=(15),SHARED,
PARTITION=((LINUX1,LINUX2),(LINUX1,LINUX2)),TYPE=FCP
CNTLUNIT CUNUMBR=0600,PATH=(15),UNIT=FCP
IODEVICE ADDRESS=(600,064),CUNUMBR=(0600),UNIT=FCP
38
Implementing Linux with IBM Disk Storage
Linux Workstation
for configuration
and control
Intranet
z800 2066-004
Ethernet
ESS 2105-800
OSA-Express GbE
Ethernet
Fixed Block disks for Linux
FCP channel
2032-64
SAN Switch
FICON channel
CKD disks for Linux and z/VM
(multiple)
Figure 3-9 Hardware configuration
3.4 Implementing Linux
This section shows the installation of SuSE Linux Enterprise Server Version 8
(SLES 8) in an LPAR and as a z/VM guest and the attachment of the ESS via
Fibre Channel. The necessary preparations and installation and customization
steps are described in the following manner:
򐂰 Two separate sections, 3.4.1 “Initial steps for installation as a z/VM Guest” on
page 40, and 3.4.2 “Initial steps for LPAR installation” on page 46 cover the
initial steps specific to each installation type;
򐂰 A common section, 3.4.3 “Installation with YaST2” on page 51 describes the
completion of the SuSE Linux installation using YaST2.
򐂰 The remaining section, 3.5 “zSeries and disk storage under Linux” on
page 64 explains the necessary steps to attach and configure the ESS
storage.
Chapter 3. SuSE Linux on zSeries
39
3.4.1 Initial steps for installation as a z/VM Guest
The initial installation of SLES 8 as a z/VM guest mandates the following
sequence of tasks:
1. Create a Linux guest.
2. Upload the kernel image, the initial ramdisk, and the parameter file to the VM
guest.
3. Create an IPL-script and IPL from virtual reader.
4. Perform the initial network setup on the VM virtual console.
After the initial system is up and running, the installation proceeds using the
graphical installation program YaST2. (see 3.4.3, “Installation with YaST2” on
page 51).
Create a Linux guest
To run Linux under z/VM, we have to create a new user (guest) in the z/VM
directory. In our example, we created a guest with the following settings.
Example 3-1 z/VM directory entry for Linux guest
USER LINUXA XXXXX 512M 1G G
ACCOUNT ITS30000
IPL CMS PARM AUTOCR
MACHINE XA 2
DEDICATE 2C04 2C04 (OSA GbE Adapter - 2C04 thru 2C06)
DEDICATE 2C05 2C05
DEDICATE 2C06 2C06
DEDICATE 0600 0600 (FCP - 0600 thru 063F)
......
DEDICATE 063F 063F
CONSOLE 0009 3215
SPOOL 000C 3505 A
SPOOL 000D 3525 A
SPOOL 000E 1403 A
LINK MAINT 0190 0190 RR
LINK MAINT 019E 019E RR
LINK MAINT 019F 019F RR
LINK MAINT 019D 019D RR
MDISK 0191 3390 1959 50 VMLU1R MR
MDISK 0202 3390 0001 200 LX3A43 MR
MDISK 0201 3390 0201 3138 LX3A43 MR
Going through these settings, you can see that we have dedicated an OSA
Express GbE adapter (2C04-2C06) and a FICON FCP adapter (0600-063F). In
40
Implementing Linux with IBM Disk Storage
addition we have three minidisks: 191 as the home disk for the guest and 201,
202 for the Linux system and swap space.
For detailed information on how to create Linux guests under z/VM please refer
to the redbooks Linux for IBM zSeries and S/390: Distributions, SG24-6264, and
Linux on IBM zSeries and S/390: Large Scale Linux Deployment, SG24-6824.
Transfer setup files through FTP
We will now describe how to get the initial installation system up and running
within the z/VM guest.
This is done through an FTP transfer.
For that purpose, you need access to an FTP client program on your z/VM guest.
The minidisk 592 of user TCPMAINT contains such one. To access the disk, log
on to your guest and issue the following commands on your virtual console:
LINK TCPMAINT 592 592 rr
ACC 592 592
Note: The above commands assume that there is no read password set on
TCPMAINTs 592 disk, which is the default. If it is protected, or the settings on
your system are different, you need to contact your z/VM administrator to give
you access.
There are three files you need to transfer from the installation media to your
home disk: the kernel image, the initial ramdisk and the parameter. Make sure
that you transfer the kernel and the ramdisk in binary mode and with a record
length of 80. The following example shows a extract of the procedure while
highlighting the actual input in bold.
Example 3-2 FTP transfer of the three initial files
Ready; T=0.01/0.01 14:11:55
ftp gallium.almaden.ibm.com
VM TCP/IP FTP Level 430
Connecting to GALLIUM.ALMADEN.IBM.COM 9.1.38.184, port 21
USER (identify yourself to the host):
stobdr2
>>>USER stobdr2
331 Please specify the password.
Password:
...
Command:
bin
...
cd /iso/s390/cd1/boot
Chapter 3. SuSE Linux on zSeries
41
...
locsite fix 80
...
get vmrdr.ikr sles8.image
...
get initrd sles8.initrd
...
asc
...
get parmfile sles8.parm
...
quit
Now the files are stored on your home disk, which you can verify using the filel
command.
IPL from the virtual reader
The three files are used to IPL from the virtual reader and have to be put
(punched) in the reader; you then IPL your system from the virtual reader. The
easiest way to do this is to create a small REXX script as shown in Example 3-3.
Example 3-3 Script SLES8 EXEC to IPL from reader
/**/
’close rdr’
’purge rdr all’
’spool punch * rdr’
’PUNCH SLES8 IMAGE A (NOH’
’PUNCH SLES8 PARM A (NOH’
’PUNCH SLES8 INITRD A (NOH’
’change rdr all keep nohold’
’ipl 00c clear’
Execute the this script and you should see the Linux boot messages appear on
the virtual console. Once the initial system has completed the boot process, the
network configuration dialog will start automatically.
Initial network and installation setup
The network configuration dialog starts with asking for the type of the network
device. In our example we used an OSA Express GbE adapter and network
settings as shown in Table 3-1.
42
Implementing Linux with IBM Disk Storage
Table 3-1 Network settings
OSA device addresses
2C04, 2C05, 2C06
OSA portname
OSA2C00
Hostname
vmlinuxa.itso.ibm.com
IP address
9.12.6.73
Netmask
255.255.254.0
Broadcast address
9.12.7.255
Gateway address
9.12.6.92
DNS server
9.12.6.7
DNS search domain
itso.ibm.com
MTU size
1492
These settings have to be supplied in the dialog. The following example shows a
walkthrough (a condensed extract) according to our setup; the actual input is
highlighted in bold.
Example 3-4 Initial network setup
=
===
Welcome to SuSE Linux Enterprise Server 8 for zSeries
=
-==
=
Please select the type of your network device:
0) no network
1) OSA Token Ring
2) OSA Ethernet
3) OSA-Gigabit Ethernet or OSA-Express Fast Ethernet
4) Channel To Channel
5) Escon
6) IUCV
8) Hipersockets
9) Show subchannels and detected devices
Enter your choice (0-9): 3
...
To set up the network, you have to read and confirm the license information
of the network device driver provided by IBM.
Do you want to see the license (Yes/No) ? yes
------------------------------------------------------------------------------International License Agreement for Non-Warranted Programs
...
------------------------------------------------------------------------------'
Do you agree with this license (Yes/No) ? yes
Chapter 3. SuSE Linux on zSeries
43
...
Enter the device addresses for the qeth module, e.g. '0x2c04,0x2c05,0x2c06'
(0x2c04,0x2c05,0x2c06): 0x2c04,0x2c05,0x2c06
Please enter the portname(case sensitive) to use(suselin7): OSA2C00
...
eth0 is available, continuing with network setup.
Please enter your full host name, e.g. 'linux.example.com' (linux.example.com):
vmlinuxa.itso.ibm.com
Please enter your IP address, e.g. '192.168.0.1' (192.168.0.1): 9.12.6.73
Please enter the net mask, e.g. '255.255.255.0' (255.255.255.0): 255.255.254.0
Please enter the broadcast address if different from (9.12.7.255): 9.12.7.255
Please enter the gateway's IP address, e.g. '192.168.0.254' (192.168.0.254):
9.12.6.92
Please enter the IP address of the DNS server or 'none' for no DNS (none):
9.12.6.7
Please enter the DNS search domain, e.g. 'example.com' (itso.ibm.com):
itso.ibm.com
Please enter the MTU (Maximum Transfer Unit), leave blank for default: (1500):
1492
Configuration for eth0 will be:
Full host name
: vmlinuxa.itso.ibm.com
IP address
: 9.12.6.73
Net mask
: 255.255.254.0
Broadcast address: 9.12.7.255
Gateway address : 9.12.6.92
DNS IP address
: 9.12.6.7
DNS search domain: itso.ibm.com
MTU size
: 1492
Is this correct (Yes/No) ? yes
For security reasons you have to set an temporary installation
system password for the user "root".
You'll be asked for it only when you Telnet into this installation
system to limit the access to it and it will be cleared as soon
as you shut down or reset the installation system
Please enter the temporary installation password: XXXXXX
Temporary installation password set.
...
Network Setup finished, running inetd...
You should be able to login via Telnet now, for ssh wait a few seconds,
temporary host keys (only for installation) are being generated now:
Generating /etc/ssh/ssh_host_key.
Generating public/private rsa1 key pair.
44
Implementing Linux with IBM Disk Storage
...
After the network setup is complete, the creation of the ssh keys is started. Even
though it is possible to log in now, you are advised to wait for the system to finish
the key generation because additional information about your installation still
needs to be supplied.
Note: The root password you provide during this step is valid for the initial
installation system only. Since it is only stored in memory, it will be gone after
the first re-IPL from DASD, and you will have to provide a password once
again.
The next dialogs prompt you for the type of server containing the installation
CDs, and for the type of installation (graphical or text based). We used an FTP
server as the installation source, and a Linux system running an X server to
perform an X-based graphical installation.
Please refer to your Linux or UNIX documentation on how to set up X on your
workstation. If you do not have a Linux, or UNIX, or Windows workstation running
an X server, you can still use the VNC client provided on the installation CD for a
graphical interface. Otherwise, just use an ssh client such as PuTTY, for a text
based installation.
Example 3-5 shows the dialog relevant to our installation.
Example 3-5 Choose installation media and type
Please specify the installation Source:
1) NFS
2) SAMBA
3) FTP
0) Abort
Choice: 3
Please enter the IP-Number of the host providing the installation media:
9.1.38.184
...
Please enter the directory of the installation media: /iso/s390/cd1
Is the following correct?
Installation Source: ftp
IP-Address: 9.1.38.184
Directory: /iso/s390/cd1
Yes/No: yes
Chapter 3. SuSE Linux on zSeries
45
Please enter the username for the FTP-access (for anonymous just press enter):
stobdr2
Please enter the password for the FTP-Access (for anonymous just press enter):
XXXXXXX
Is the following correct?
FTP User: stobdr2
FTP Password: XXXXXX
Yes/No: yes
Which terminal do want to use?
1) X-Window
2) VNC (VNC-Client or Java enabled Browser)
3) ssh
Choice: 1
Please enter the IP-Number of the host running the X-Server: 9.43.152.107
...
ramdisk /dev/ram0 freed
>>> SuSE Linux installation program v1.4 (c) 1996-2002 SuSE Linux AG <<<
Starting hardware detection...
Searching for
infofile......
Loading data into ramdisk...............
....
cintegrating the installation system into the ramdisk...
integrating the shared objects of the installation system...
integrating kernel modules of the installation system...
starting yast...
Once the installation system is loaded into memory, YaST2 is started
automatically and is displayed on the X server. Please proceed with 3.4.3,
“Installation with YaST2” on page 51.
3.4.2 Initial steps for LPAR installation
Within LPAR mode you have several options to IPL a system. While for zSeries
operating systems like z/OS and z/VM, the IPL from a tape is preferred, the most
convenient way to IPL a Linux system is to use the option to load from the HMC
CD-ROM drive or from an FTP server. However, you can prepare a Linux IPL
tape as well if you do not have access to the HMC CD-ROM or to a FTP server.
46
Implementing Linux with IBM Disk Storage
Please refer to the documentation of your distribution on how to create an IPL
tape.
In our example we use an FTP server, which is accessible within the HMC
network segment. The Linux installation CDs must be copied on this server. It is
recommended to copy each CD in a separate directory, using the same initial
path, but separating the different CDs in subdirectories named CD1, CD2, CD3.
Also, note that the first CD has the necessary information to allow the IPL: it
contains the suse.ins file, which is a tape description file, used to emulate the CD
as a tape for the zSeries.
To perform the IPL, log on to your HMC and then select the following:
1. Go into Defined CPCs and highlight the CPC you will work on.
2. Change to the CPC Recovery menu and double-click Single Object
Operations.
3. Confirming the action in the next window will bring you on the CPC console.
4. Go into Images and highlight the LPAR to IPL.
5. Change to the CPC Recovery menu and double-click Load from CD-ROM
or Server.
You will get a warning that all jobs will be cancelled. Make sure that no system is
active or that it is safe to re-IPL the system. Clicking Yes opens the window
shown in Figure 3-10. You have to provide the hostname or IP address of the
FTP server, a valid user ID and password, as well as the path indicating the
location of the first CD (i.e: /code/sles8/cd1 in our case).
Chapter 3. SuSE Linux on zSeries
47
Figure 3-10 HMC panel to load from FTP source
When you click Continue, the system verifies the installation source and
provides a list of possible software to load. In the case of our example there is
only one choice as shown in the Figure 3-11.
Figure 3-11 Select the source
48
Implementing Linux with IBM Disk Storage
Do the following:
1. Select the appropriate source and click Continue.
2. Acknowledge the action by clicking Yes in the confirmation window that pops
up.
3. The load process is started; when finished, a dialog window informs you of its
completion. Click OK on this dialog window.
4. Go back to the HMC by ending your CPC Console session (ending the Single
Object Operation).
5. On the HMC go into CPC Images and highlight your LPAR.
6. Change to the CPC Recovery menu and double-click Operating System
Messages.
After a moment you will see the boot messages from Linux and the start of the
network configuration dialog as shown in Figure 3-12.
Figure 3-12 HMC operating system messages interface
You have to click Respond first to be able to enter commands and respond to
questions presented in the Message Text pane. This will give you an input
window where you can type answers or input commands as shown in
Figure 3-13.
Chapter 3. SuSE Linux on zSeries
49
Figure 3-13 Enter commands on the HMC
To send the command to the Linux system click Send. Proceed in this manner
until you have answered all the questions using the same answers as for the
z/VM installation.1 (See “Initial network and installation setup” on page 42) You
can see the completed network dialog in Figure 3-14.
Figure 3-14 Network settings
1
For our example we used the very same network settings for the Linux LPAR
installation as for the installation as a z/VM guest, because the devices such as
OSA FCP adapter were the same and the z/VM Linux guest was not active
during these tests.
50
Implementing Linux with IBM Disk Storage
Once the network settings are complete, you need to provide information
regarding the installation server as shown in Figure 3-15.
Figure 3-15 Installation server settings
Finally, you will have to specify the type of installation, either graphical or text
based, you want to perform. (See “Initial network and installation setup” on
page 42)
The initial steps for installation in an LPAR are completed at this point. For the
second phase of the installation, using YaST, please proceed to the next section.
Note: Once the installation with YaST is started, do not exit the program if you
have to perform additional actions such as formatting DASD. It would end the
entire installation procedure and you would have to do the LPAR IPL all over
again. Instead use a parallel Telnet or ssh session to format DASDs or other
additional tasks.
3.4.3 Installation with YaST2
After completion of the initial installation steps for either a z/VM or LPAR
configuration, the second phase of the installation is performed using SuSEs
installation program YaST, which will start automatically. In our example we used
the graphical X based installation, but the procedures and menus for a VNC or
Chapter 3. SuSE Linux on zSeries
51
text based installation are the same, just the appearance, obviously, is different.
Please refer to the SuSE installation manual for the other types of YaST
installations. The following screen captures were taken from the z/VM
installation, so all the settings shown correspond to “Initial steps for installation as
a z/VM Guest” on page 40.
YaST starts by displaying the End User License for the SuSE Linux Enterprise
Server. You have to click Accept in order to proceed.
The first YaST2 installation screen that follows prompts you for the language
selection as shown in Figure 3-16.
Figure 3-16 Language selection
Select the appropriate language and click Accept.
Now, you need to provide the parameters for the DASD driver to know on what
DASDs to install Linux. Enter your numbers for the dasd volumes (in our case the
z/VM minidisks 201 and 202; note that individual DASD device numbers are
delimited by commas, and ranges of DASD are defined by placing a dash
between the lowest and the highest address of the range.)
52
Implementing Linux with IBM Disk Storage
Click the Load Module button; after a successful load the DASDs will show up in
the lower half of the window. The process is shown in Figure 3-17. Click Accept
to proceed.
Figure 3-17 Load dasd module
In the next window (as shown in Figure 3-18) you have to select if you want to
install a completely new system, or if existent, boot an already installed system.
In our case we reused some old DASDs, so that YaST recognized that there was
a system installed before. Since we want to install a new system, we select New
installation and click OK to continue.
YaST now tries to automatically detect the system settings and set up the
installation settings. Because the DASDs we used were not formatted and
partitioned with this system before, the automatic assignment of mount points
fails and is highlighted (colored red) as an error message that appears under
Partitioning, as shown in Figure 3-19.
Chapter 3. SuSE Linux on zSeries
53
Figure 3-18 Select installation
Figure 3-19 Installation settings without partitioning settings
To fix this you have to assign the volumes manually by clicking Partitioning. This
will bring up the Expert Partitioner as shown in Figure 3-20.
54
Implementing Linux with IBM Disk Storage
Figure 3-20 Expert partitioner without dasda
You will notice, that even if we loaded the DASD module with parameter 201 and
202, only the DASD 202 (which is dasdb in Linux) with one partition shows up
here. This is because the DASD 201 was not formatted and partitioned yet within
Linux. As described in the upper left of the window, you cannot partition DASD
from this graphical dialog; instead, you have to use the command line utilities.
Therefore, we open a separate ssh session to access our Linux system and
perform the formatting and partitioning.
Note: Do not proceed or perform any action within the Expert Partitioner
before the manual formatting and partitioning is completed. You will get error
messages and disturb the setup process within YaST.
After logging into the system with a parallel ssh session, we used the utilities
dasdfmt to format dasda, and fdasd to create one partition on it. The format
during this stage is the low level format and must not be mistaken with the
creation of a file system (e.g. ext2, ext3). Please refer to the SuSE
documentation on how to use the utilities, and the different options and
parameters.
Chapter 3. SuSE Linux on zSeries
55
Example 3-6 shows the process highlighting the input bold.
Example 3-6 Formatting dasd manually
SuSE Instsys vmlinuxa:/root # dasdfmt -b 4096 -d cdl -f /dev/dasda
Drive Geometry: 3138 Cylinders * 15 Heads = 47070 Tracks
I am going to format the device /dev/dasda in the following way:
Device number of device : 0x201
Labelling device
: yes
Disk label
: VOL1
Disk identifier
: 0X0201
Extent start (trk no)
: 0
Extent end (trk no)
: 47069
Compatible Disk Layout : yes
Blocksize
: 4096
--->> ATTENTION! <<--All data of that device will be lost.
Type "yes" to continue, no will leave the disk untouched: yes
Formatting the device. This may take a while (get yourself a coffee).
Finished formatting the device.
Rereading the partition table... ok
SuSE Instsys vmlinuxa:/root #
SuSE Instsys vmlinuxa:/root # fdasd /dev/dasda
reading volume label: VOL1
reading vtoc
: ok
Command action
m
print this menu
p
print the partition table
n
add a new partition
d
delete a partition
v
change volume serial
t
change partition type
r
re-create VTOC and delete all partitions
u
re-create VTOC re-using existing partition sizes
s
show mapping (partition number - data set name)
q
quit without saving changes
w
write table to disk and exit
Command (m for help): p
Disk /dev/dasda:
3138 cylinders,
15 tracks per cylinder,
12 blocks per track
56
Implementing Linux with IBM Disk Storage
4096 bytes per block
volume label: VOL1, volume identifier: 0X0201
maximum partition number: 3
-----------tracks---------Device
start
end
2
47069
length
47068
Id
System
unused
Command (m for help): n
First track (1 track = 48 KByte) ([2]-47069):
Using default value 2
Last track or +size[c|k|M] (2-[47069]):
Using default value 47069
Command (m for help): w
writing VTOC...
rereading partition table...
After the format and partitioning we can close the ssh session and go back to the
graphical Expert partitioner. Select Expert -> Reread the partition table, and
after acknowledging a warning window you should see the newly formatted
dasda in the list as shown in Figure 3-21.
Chapter 3. SuSE Linux on zSeries
57
Figure 3-21 Expert partitioner with dasda
Now we can go ahead and assign mount points and file system formats to the
DASD partitions by selecting the partitions and clicking Edit. First we select the
partition on dasda (/dev/dasda1) and select / as the mount point as well as the
the type of file system desired for that partition. In our case we used the ext3 file
system (please refer to “File systems” on page 258 in Appendix A for further
information on Linux file systems). Click OK to save the settings. Finally, we
select the partition on dasdb (/dev/dasdb1) and assign swap as the mount point.
This automatically omits the file system option. Click OK to save the settings.
Important: In our example we chose to create only one root file system to
simplify the installation process. For a real (production) system, we strongly
recommend that you create more partitions. This is important not only
because of space management, but also for security reasons. By separating
for example /tmp, you prevent ordinary users from filling up your root file
system. Please refer to the documentation of your distribution for specific
recommendations about the partition layout.
58
Implementing Linux with IBM Disk Storage
Now we have correctly assigned our DASD space. The settings are shown in
Figure 3-22. Notice the formatting flag F for the partition dasda1 in the F column.
Figure 3-22 Expert partitioner with partition settings
To go back to the Installation Settings screen click Next. As you can see in
Figure 3-23, we now have the desired action specified for the Partitioning
section.
We can proceed with the installation by clicking Accept.
Note: You can make changes to the software package selection before
clicking Accept, by clicking on Software. In our example we use the default
package selection.
Chapter 3. SuSE Linux on zSeries
59
Figure 3-23 Installation settings with partitioning settings
For the system to proceed, you have to acknowledge the start of the installation
by clicking Yes, install, as shown in Figure 3-24.
Figure 3-24 Warning before proceed with installation
60
Implementing Linux with IBM Disk Storage
YaST will now format the DASDs with the file systems you specified, followed by
the installation of the software packages from the FTP server to DASD. This
process takes a while, but you can watch the progress and the estimated
remaining time as shown in Figure 3-25.
Figure 3-25 Installation progress
When the packages installation has completed, confirm the re-IPL from DASD by
clicking OK on the message window as shown in Figure 3-26.
Chapter 3. SuSE Linux on zSeries
61
Figure 3-26 Ready to IPL from dasd for first time
Note: Do not close your X Server or change any of its settings, because the
YaST installation will resume after the re-IPL.
After a successful re-IPL from DASD you will see the following message on your
system console.
Example 3-7 Message during first IPL from dasd
***
***
***
Please return to your X-Server screen to finish installation
Return to your X server where a dialog asking for the root password should now
display. Fill in a carefully chosen password and click Next to continue. The next
screen allows you to create a normal user account. You can fill in the necessary
information and then proceed with Next. You can also choose to skip this step
and create normal users later, by leaving the fields empty and just clicking Next.
62
Implementing Linux with IBM Disk Storage
At this point, YaST will run several post installation scripts, which will configure
the software packages previously installed.
When the scripts have completed, you can review the network settings or add
printers by clicking the appropriate links in the next screen. Since the network
settings from the installation dialog are carried over automatically and we do not
want to add a printer, we just continue by clicking Next.
Finally, YaST is saving the network settings and setting up the network services
as shown in Figure 3-27.
Figure 3-27 Saving the settings
The installation of the SuSE Linux Enterprise Server is complete. The system is
a up and running and will allow authorized users to log in.
The next section describes how to add Fibre Channel attached storage to the
system.
Chapter 3. SuSE Linux on zSeries
63
3.5 zSeries and disk storage under Linux
In this section we focus on the attachment of disk storage via Fibre Channel as
stated in the introduction.
Important: A large number of SCSI devices exist in the marketplace. IBM
cannot fully test all of these for use in zSeries systems. There will be a
distinction between IBM supported devices and devices that have been used
and appear to work but are not formally supported for zSeries FCP by IBM.
For a complete list of supported devices, consult:
http://www-1.ibm.com/servers/eserver/zseries/connectivity/#fcp
3.5.1 ESS using FCP
This section refers to the physical setup presented in 3.3, “Configuration used for
this redbook” on page 38. Here we discuss the following:
򐂰
򐂰
򐂰
򐂰
Gather prerequisite information about the SAN setup
Set up the ESS and configure LUNs for use with Linux for zSeries
Manually load the required modules and parameters to access the disks
Make the disks permanently accessible
Prerequisite information
From a software configuration standpoint, you need to collect the following
elements of information, as described in 3.1.2 “Distributed storage attachments”
on page 29, in order to prepare a Linux system for accessing the Enterprise
Storage Server through Fibre Channel:
򐂰
򐂰
򐂰
򐂰
򐂰
Hostname of the server hosting the Linux system
Device address (and CHPID) of the FCP port attached to Linux
World Wide Port Name (WWPN) of the FCP port on the zSeries
FC port on the ESS
World Wide Port Name of the FC port on the ESS
You can obtain the data from the HMC, the ESS Specialist, and the McData
switch; in our case, these were:
Table 3-2 World Wide Port Names
64
Linux hostname
vmlinuxa
CHPID of the FCP port on the z800
15
WWPN of the FCP port on the z800
50:05:07:64:01:40:01:7d
FC port on the ESS
Bay 3 Slot 1
Implementing Linux with IBM Disk Storage
Linux hostname
vmlinuxa
WWPN of the FC port on the ESS
50:05:07:63:00:c8:95:89
In order to gather the necessary data for your environment, please refer to the
documentation of your SAN switch management software, and to the Redpaper
Getting Started with zSeries Fibre Channel Protocol, REDP0205.
Set up the ESS
To set up LUNs for use with your Linux system, you first have to define the
system to the ESS. This is done through the ESS Specialist; in this section we
assume that you already are familiar with the ESS Specialist. We enumerated all
the actions, but did not include a picture for all the screens. You can find more
information about the ESS Specialist in Chapter 7. “Configuring ESS for Linux”
on page 155.
Log on to the ESS Specialist and select Storage Allocation -> Open System
Storage -> Modify Host Systems to display the host setup screen. Enter a
nickname for the host, select the host type from the pull-down list, fill in the Linux
hostname and the WWPN of the FCP port on the z800, and select the FC port
on the ESS. The complete settings are shown in Figure 3-28.
Figure 3-28 Add the Linux host
Chapter 3. SuSE Linux on zSeries
65
Click Perform Configuration Update and the settings you just provided are
applied once you confirm.
Go back to the Storage Allocation view, and you now see the Linux system
shown under the nickname you chose.
Click View All Storage to see unassigned storage on the ESS, potentially
available for your Linux system.
Note: In our example, the unassigned loop was already formatted for Open
System Storage (Fixed Block) and a RAID 5 array was defined on that loop.
Go to Open System Storage and select your Linux host system from the list.
Click Add Volumes to create LUNs as shown in Figure 3-29.
Figure 3-29 Add volumes
Again, select your Linux host and the FC port to use. This also highlights the
connection between them; proceed by clicking Next. A second screen appears
where you can choose from the available arrays and add volumes of different
sizes. To add a volume, select the array and the desired size of the volume, then
click Add (in our example, we created one 48 GB volume and two 81.2 GB
volumes on the available RAID 5 array).
66
Implementing Linux with IBM Disk Storage
Click Perform Configuration Update and after confirmation (separate message
window), the volumes are created. A message window informs you that the
actual format of the volumes is being performed in the background. You have to
wait until the formatting is complete before accessing the new volumes. Select
the Linux host on the Open System Storage screen to view the newly created
volumes.
To view the LUNs of the volumes, go to Open System Storage and click Modify
Volume Assignments. In the list shown in Figure 3-30 you can see the LUNs of
the volumes assigned to your Linux system (in the Host Port column).
Figure 3-30 View volume assignments to display LUNs
The LUNs for our Linux system are 5300, 5301, and 5302.
Set up the Linux system
You must have the following modules installed to use the FCP device with the
ESS:
򐂰
򐂰
򐂰
򐂰
qdio - The same module is used for other qdio devices
scsi_mod - SCSI core
zfcp - Provides FCP support for zSeries Linux
sd_mod - SCSI disk support
Chapter 3. SuSE Linux on zSeries
67
The modules should be loaded in the order listed. All modules can be loaded with
modprobe. Further SCSI modules exist for other SCSI devices such as st for tape
support, sr_mod for CD/DVD (read only) support. They have been tested to work
with different devices, but will not be discussed in this book. Please refer to the
Redpaper Getting Started with zSeries Fibre Channel Protocol, REDP-0205.
All modules, except zfcp, can be loaded without parameters. For the zfcp
module, parameters for the mapping of the FCP devices on the ESS are
required. The parameters needed for each device are:
The device number of the FCP port on the zSeries
The SCSI ID starting at 1
WWPN of the ESS FC port
The SCSI LUN within Linux starting at 0
The FCP LUN of the target volume on the ESS
򐂰
򐂰
򐂰
򐂰
򐂰
The format of supplying the mapping parameters to the module is as follows:
map=”devno SCSI_ID:WWPN SCSI_LUN:FCP_LUN” such as:
map="0x0600 1:0x5005076300c89589 0:0x5301000000000000"
To add more then one device to the map, simply add more statements to the
parameter line, separated by semicolons. The next example shows the load of
the modules (omitting the qdio module, because it is loaded and used already by
the OSA Express adapter in our installation) for one volume.
Example 3-8 Load scsi and zfcp modules
vmlinuxa:~ # modprobe scsi_mod
vmlinuxa:~ # modprobe zfcp map="0x0600 1:0x5005076300c89589 \
0:0x5301000000000000"
vmlinuxa:~ # cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 01 Lun: 00
Vendor: IBM
Model: 2105800
Type:
Direct-Access
vmlinuxa:~ # modprobe sd_mod
vmlinuxa:~ # cat /proc/partitions
major minor
8
94
94
94
94
0
0
1
4
5
Rev: .101
ANSI SCSI revision: 03
#blocks
name
rio rmerge rsect ruse wio wmerge wsect wuse running use aveq
79296896
2259360
2259264
144000
2403264
sda 2 6 16 10 0 0 0 0 0 10 10
dasda 24799 32810 460872 116390 34395 47194 654280 738410 0 175040 854800
dasda1 24778 32810 460704 116350 34395 47194 654280 738410 0 175000 854760
dasdb 53 105 1264 80 0 0 0 0 0 80 80
dasdb1 32 105 1096 40 0 0 0 0 0 40 40
The example shows that a new SCSI volume, sda, is accessible by our system.
However, before you can use a new volume, it must be partitioned using fdisk
68
Implementing Linux with IBM Disk Storage
After partitioning the new volume you can create file systems on the partitions
either using the command line utilities or using YaST. In our case, we assigned
the mount point, formatted the partition with the ext3 file system, and applied the
changes. Now the new device is ready for use.
Please refer to Appendix A for details on how to partition disks and create file
systems.
Add more devices and make changes permanent
To add more than one device to your SCSI configuration, it is more convenient to
write a small script with all the parameters included such as shown in
Example 3-9.
Example 3-9 Module load script
vmlinuxa:~ # cat scsi.sh
modprobe scsi_mod
modprobe zfcp map="\
0x0600 1:0x5005076300c89589 0:0x5301000000000000;\
0x0600 1:0x5005076300c89589 1:0x5302000000000000;\
0x0600 1:0x5005076300c89589 2:0x5300000000000000"
modprobe sd_mod
Make sure that the modules are unloaded before executing the script.
Example 3-10 Module load via script
vmlinuxa:~ # sh scsi.sh
vmlinuxa:~ # cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 01 Lun: 00
Vendor: IBM
Model: 2105800
Type:
Direct-Access
Host: scsi0 Channel: 00 Id: 01 Lun: 01
Vendor: IBM
Model: 2105800
Type:
Direct-Access
Host: scsi0 Channel: 00 Id: 01 Lun: 02
Vendor: IBM
Model: 2105800
Type:
Direct-Access
Rev: .101
ANSI SCSI revision: 03
Rev: .101
ANSI SCSI revision: 03
Rev: .101
ANSI SCSI revision: 03
Alternatively, you can also add SCSI devices to an existing configuration via
add_map. Then you have to make the devices known to the SCSI stack manually
as shown in Example 3-11.
Example 3-11 Add devices manually via add_map
vmlinuxa:~ # cat /proc/scsi/scsi
Attached devices:
Chapter 3. SuSE Linux on zSeries
69
Host: scsi0 Channel: 00 Id: 01 Lun: 00
Vendor: IBM
Model: 2105800
Rev: .101
Type:
Direct-Access
ANSI SCSI revision: 03
vmlinuxa:~ # echo "0x0600 0x00000001:0x5005076300c89589 \
0x00000001:0x5302000000000000" > /proc/scsi/zfcp/add_map
vmlinuxa:~ # echo "scsi add-single-device 0 0 1 1" > /proc/scsi/scsi
vmlinuxa:~ # cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 01 Lun: 00
Vendor: IBM
Model: 2105800
Rev: .101
Type:
Direct-Access
ANSI SCSI revision: 03
Host: scsi0 Channel: 00 Id: 01 Lun: 01
Vendor: IBM
Model: 2105800
Rev: .101
Type:
Direct-Access
ANSI SCSI revision: 03
To make the devices available permanently (also after a reboot), you have to
create a new initial ramdisk containing the necessary modules and parameter
information. First, save the module parameters in the configuration file
/etc/zfcp.conf.
Example 3-12 Configuration file for zfcp
vmlinuxa:~ # cat /proc/scsi/zfcp/map
vmlinuxa:~ # cat /etc/zfcp.conf
0x0600 0x00000001:0x5005076300c89589
0x0600 0x00000001:0x5005076300c89589
0x0600 0x00000001:0x5005076300c89589
> /etc/zfcp.conf
0x00000000:0x5301000000000000
0x00000001:0x5302000000000000
0x00000002:0x5300000000000000
Next, create a new ramdisk which is done with the utility mk_initrd and then run
zipl to update the IPL record to point to the new ramdisk. This is shown in
Example 3-13.
Example 3-13 Create new initial ramdisk and run zipl
vmlinuxa:~ # mk_initrd
using "/dev/dasda1" as root device (mounted on "/" as "ext3")
Found ECKD dasd, adding dasd eckd discipline!
Note: If you want to add ECKD dasd support for later mkinitrd
calls where possibly no ECKD dasd is found, add dasd_eckd_mod
to INITRD_MODULES in /etc/sysconfig/kernel
creating initrd "/boot/initrd" for kernel "/boot/kernel/image"
(version 2.4.19-3suse-SMP) (s390)
- insmod scsi_mod
- insmod qdio
70
Implementing Linux with IBM Disk Storage
(kernel/drivers/scsi/scsi_mod.o)
(kernel/drivers/s390/qdio.o)
- insmod zfcp map="
0x0600 0x00000001:0x5005076300c89589 0x00000000:0x5301000000000000
0x0600 0x00000001:0x5005076300c89589 0x00000001:0x5302000000000000
0x0600 0x00000001:0x5005076300c89589 0x00000002:0x5300000000000000"
(kernel/drivers/s390/scsi/zfcp.o)
- insmod jbd
(kernel/fs/jbd/jbd.o)
- insmod ext3
(kernel/fs/ext3/ext3.o)
- insmod dasd_mod dasd=$dasd (kernel/drivers/s390/block/dasd_mod.o)
- insmod dasd_eckd_mod
(kernel/drivers/s390/block/dasd_eckd_mod.o)
- insmod sd_mod
(kernel/drivers/scsi/sd_mod.o)
- zfcp support
Run zipl now to update the IPL record!
vmlinuxa:~ # zipl
building bootmap
adding Kernel Image
adding Ramdisk
adding Parmline
Bootloader for ECKD
Syncing disks....
...done
vmlinuxa:~ #
: /boot/zipl/bootmap
: /boot/kernel/image located at 0x00010000
: /boot/initrd located at 0x00800000
: /boot/zipl/parmfile located at 0x00001000
type devices with z/OS compatible layout installed.
Now, all the necessary parameters and modules are accessible at boot time. The
next time you re-IPL the system the scsi and zfcp modules are loaded
automatically, and the file systems are mounted as defined in /etc/fstab.
3.5.2 Multipath support
Some storage devices allow operating systems to access device storage through
more than one physical path, thus extending availability, and sometimes allowing
better performance by permitting use of all paths in parallel. To make use of
these features, an operating system must be able to detect multipath devices and
manage all access in a way that avoids data corruption.
Currently, Linux for zSeries and S/390 supports only the Enterprise Storage
Server (ESS) through extensions to the Linux Logical Volume Manager (LVM). To
make use of the LVM extensions, any device driver must provide one device
node for every path that can be accessed independently (the device driver
handles the path switching if the device architecture needs anything special for
that purpose).
Block-device managers (such as LVM, MD, or EVMS) take block devices and
provide new logical ones after some transformation. This just requires the lower
Chapter 3. SuSE Linux on zSeries
71
device drivers to provide a separate device for every path so that the manager
knows about the paths.
Please refer to the Device Drivers and Installation Commands documentation on
how to implement multipathing for Linux on zSeries.
72
Implementing Linux with IBM Disk Storage
4
Chapter 4.
SuSE Linux on pSeries
This chapter starts with a brief overview of IBM eServer pSeries and its Linux
support.
It continues with a focus on the requirements and installation steps pertaining to
SuSE Linux Enterprise Server 8 in two ways: first, a native installation on a
pSeries server, then installation in an LPAR (available on the p670 and p690
models).
The chapter ends with a review of the tasks involved in the configuration of an
Emulex Host Bus Adapter, for attachment to FAStT and ESS.
© Copyright IBM Corp. 2001 2003. All rights reserved.
73
4.1 Introduction to pSeries and Linux
The pSeries is one of IBM’s key solutions to fit a variety of customers needs. The
pSeries family presents a full range of servers from entry deskside models to
high-end enterprise models with LPAR capabilities.
Linux for pSeries is a key element of the IBM Linux strategy. IBM is working
closely with the Linux community to increase performance, scalability, reliability,
and serviceability to match the strengths of pSeries servers (see Figure 4-1).
To make it easy to get started with Linux for pSeries, IBM has introduced a
number of pSeries Linux ready Express Configurations. These systems
represent some of the most popular configurations of the p630 and p650
systems. They are provided without an AIX® license and offer great savings with
the ability to add additional features.
High-end
Mid-range
IBM
IBM
IBM
IBM
pSeries
IBM
Entry Deskside
pSeries 690
server
IBM
IBM
pSeries 670
pSer ies
IBM
p Series
pSer ies
pSeries 620
Models
6F0, 6F1
pSeries 630
Model 6E4
pSeries 610
Model 6E1
pSeries 650
pSeries 660
Models 6H0,
6H1, 6M1
pSeries 655
H C R U6
IBM
I BM
IB M
s erve r
p Series
pSeries 610
Model 6C1
pSeries 640
Model B80
pSeries 630
Model 6C4
Entry Rack
Figure 4-1 pSeries models supporting Linux
Linux for pSeries is especially compelling for solutions requiring a 64-bit
architecture or the high-performance floating-point capabilities of the POWER
processor.
In addition, the logical partitioning LPAR capabilities of the pSeries make it
possible to run one or more instances of Linux along with AIX (i.e. zero or more
Linux partitions along with zero or more AIX partitions). This enables the
74
Implementing Linux with IBM Disk Storage
consolidation of workloads from several separate servers onto a single system.
Since the partitioning is controlled by the hypervisor firmware and the Hardware
Management Console, AIX is never required to run Linux. LPAR capabilities also
provide a low-risk way to begin developing and deploying Linux operating
system-ready applications as desired while retaining the enterprise-ready
capabilities of AIX for mission-critical or highly-scalable workloads. Since Linux
does not currently scale to efficiently handle large SMP systems, LPAR also
allows large pSeries systems to be partitioned to run Linux workloads.
It should be noted also that with AIX 5L™, IBM introduced the AIX Toolbox for
Linux applications. This set of tools allows you to recompile Linux source code
under AIX, offering the possibility to run applications developed on Linux while
retaining the more advanced features of AIX.
For an overview of what pSeries models are currently supported by what Linux
distributions, see:
http://www.ibm.com/servers/eserver/pseries/hardware/linux_facts.pdf
In this redbook, our focus is on SuSE SLES8 / UnitedLinux 1.0.
Note: For the latest information on pSeries and Linux in general, refer to:
http://www.ibm.com/eserver/pseries/linux
4.2 Requirements
Attachment to storage systems shown in this book was still under test at the time
of writing. If you plan to use scenarios from this book, please verify first that
these configurations have now been certified by IBM.
4.2.1 Hardware requirements
In general, the SuSE Linux Enterprise Server runs on pSeries systems natively.
You can find additional information about the supported hardware and technical
issues by consulting the following Internet pages:
http://www-i.ibm.com/servers/eserver/pseries/linux
http://www.suse.com/us/business/products/server/sles/i_pseries.html
Chapter 4. SuSE Linux on pSeries
75
4.2.2 Software requirements
Take note: At the time of this writing, there was no multipath driver for the
Emulex Fibre Channel adapters, nor SDD for ESS available yet. This may
have changed since publishing this redbook. Please check the appropriate
Internet pages.
The following software components are necessary for installation:
򐂰 SLES 8 (UnitedLinux 1.0) for pSeries
򐂰 Host adapter device drivers package (provided by SuSE, included in the
Service Pack 1 (SP1®) for SLES 8)
򐂰 Depending on the IBM TotalStorage device chosen, you require for:
a. FAStT:
•
•
Open Build Driver for the Emulex Adapter family (when available)
any system able to run the FAStT Storage Manager Client
b. Enterprise Storage Server ESS:
•
SDD (when announced)
4.2.3 Connection requirements
For the initial setup you will require a network infrastructure to configure the
storage components.
Additionally, you need either a console to do a native installation, or access to the
HMC when doing an LPAR installation. To facilitate the installation tasks, we
suggest that you set up a desktop system with a Telnet client and a VNC client
(available on the UnitedLinux CD #1 in the dosutils folder). However, the VNC
server can be accessed via any Web browser on port 5801.
When setting up your hardware, consider all aspects of required SAN
connections, including the length of your cables.
4.3 Configuration used for this redbook
For the configurations illustrated in this redbook, we used the following hardware:
򐂰 IBM RS/6000® 270
򐂰 IBM pSeries p610 -> config
򐂰 IBM pSeries p690
76
Implementing Linux with IBM Disk Storage
Additionally, we used the following IBM TotalStorage components:
򐂰
򐂰
򐂰
򐂰
FAStT200
FAStT700
ESS 800
IBM Fibre Channel switch 2109-F16
4.4 Implementing SuSE Linux Enterprise Server 8
In principle there are four ways to perform the installation:
򐂰 Local, full graphical install using the system video adapter
򐂰 Serial installation using a terminal console connected to the first serial port of
the pSeries system
Figure 4-2 Installation using serial console
Chapter 4. SuSE Linux on pSeries
77
Figure 4-3 Installation using VNC client
򐂰 Remote installation using the serial console for hardware settings, and a VNC
client for the graphical installation
򐂰 Remote installation using the serial console for hardware settings and a Web
browser supporting JAVA as VNC client for the graphical installation
78
Implementing Linux with IBM Disk Storage
Figure 4-4 Installation using Web browser acting as VNC client by JAVA
The result of the installation is identical whatever mode of installation you use.
Simply select the one that is the most convenient, based on your environment.
4.4.1 Installation steps
For our experiment, we used the remote installation with a serial console for
hardware settings, and a VNC client for the graphical installation.
We set up serial console with the following parameter: VT100, 9600 Baud, 8/1/N
and connect it to the Service Processor. Details on how to install a serial console
using System Management Services is available from the following pSeries
library Web site:
http://www-1.ibm.com/servers/eserver/pseries/library
The paragraphs that follow illustrate the case of a native Linux installation; Linux
installation in an LPAR is not different, except for the initial setup of the LPAR
addressed at the end of this section.
Power on your system. The serial console displays the menu shown in
Figure 4-5.
Chapter 4. SuSE Linux on pSeries
79
Service Processor Firmware
Firmware level: sc020308
Copyright 2000, IBM Corporation
MAIN MENU
1.
2.
3.
4.
5.
6.
99.
Service Processor Setup Menu
System Power Control Menu
System Information Menu
Language Selection Menu
Call-In/Call-Out Setup Menu
Set System Name
Exit from Menus
1> 2
Figure 4-5 Main Menu, choose 2 to proceed
Enter 2 to access the System Power Control Menu shown in Figure 4-6.
SYSTEM POWER CONTROL MENU
1. Enable/Disable Unattended Start Mode:
Currently Enabled
2.
3.
4.
5.
6.
Ring Indicate Power-On Menu
Reboot/Restart Policy Setup Menu
Power-On System
Power-Off System
Enable/Disable Fast System Boot:
Currently Enabled
7. Boot Mode Menu
98. Return to Previous Menu
99. Exit from Menus
1> 4
WARNING: POWERING SYSTEM WILL EXIT MENUS!
Enter "Y" to continue, any other key to abort.y
System Powering On.
Figure 4-6 Power Control Menu
80
Implementing Linux with IBM Disk Storage
Watch your console closely for the line shown in Figure 4-7. As soon as it
displays, and depending on your system architecture, press either F1 (for PPC
with 64bit) or 1 (for PP with 32bit) to enter the System Management Services.
memory
keyboard
network
scsi
speaker
Figure 4-7 End of system boot, press 1/F1
In the System Management Services (Figure 4-8) select Multiboot, by entering 2.
Version NAN02254
(c) Copyright IBM Corp. 2000 All rights reserved.
-----------------------------------------------------------------------------System Management Services
1
2
3
4
Display Configuration
Multiboot
Utilities
Select Language
.------.
|X=Exit|
`------'
===>2
Figure 4-8 Select boot device
This triggers the system to look for boot devices. This operation can last for
several minutes because of time out values in effect (while the operation is in
progress, the bottom corner of the screen shows the device currently being
probed).
When the scan is complete, a menu similar to the one shown in Figure 4-9 is
displayed. Insert your installation CD, and enter the number corresponding to the
CD-ROM device.
Chapter 4. SuSE Linux on pSeries
81
Install Operating System
Device
Number Name
1
SCSI CD-ROM [email protected],0 ( Integrated )
2
Port E2 - 100/10 Ethernet Adapter ( Integrated )
3
Port E1 - 100/10 Ethernet Adapter ( Integrated )
4
None
.------.
|X=Exit|
`------'
===>1
Figure 4-9 Choose CD drive for installation CD boot
The next menu presents a choice of software to install. Select Linux (the ID
string of the CD is displayed) by entering the corresponding number (see
Figure 4-10).
Version NAN02254
(c) Copyright IBM Corp. 2000 All rights reserved.
-----------------------------------------------------------------------------Install Software
->1
SuSE SLES-8 (PPC)<-
.------.
|X=Exit|
`------'
===>1
Figure 4-10 Choose 1 to trigger SLES install
The system starts booting from CD. The companion piece to the boot manager
LILO, called yaboot for the PPC architecture prompts you, as shown in
Figure 4-11.
82
Implementing Linux with IBM Disk Storage
Config file read, 129 bytes
Welcome to SuSE Linux!
Use
Use
"install"
"install32"
to boot the ppc64 kernel
to boot the 32bit RS/6000 kernel
You can pass the option "noinitrd" to skip the installer.
Example: install noinitrd root=/dev/sda4
Welcome to yaboot version 1.3.6.SuSE
Enter "help" to get some basic usage information
boot:
Figure 4-11 yaboot “boot”-prompt
In the boot entry field at the bottom of the screen, enter the command
corresponding to the appropriate kernel type for your platform (install for 64-bit,
install32 for 32-bit).
To perform the installation using the VNC Client you must specify some
additional parameters. In our case we used the following command line (please
adapt the parameters for your environment):
boot: install vnc=1 vnc_password=yourchoice hostip=10.10.10.79 netdevice=eth0
insmod=pcnet32
Table 4-1 lists all commands and possible parameters.
Table 4-1 Standard parameters for boot-prompt
install
install32
selected kernel
vnc=1
start vnc-server
vnc_password=xyz
defines the password xyz that is
mandatory for client access
dhcp=1
network configuration is provided by
existing DHCP server
hostip=a.b.c.d
sets the IP address of the host you are
installing to a.b.c.d
Chapter 4. SuSE Linux on pSeries
83
netmask=a.b.c.d
use a.b.c.d as netmask
gateway=a.b.c.d
use a.b.c.d as gateway
netdevice=eth0
use network device eth0
insmod=pcnet32
install module pcnet32 automatically
As a result of the command, yaboot loads the installation system from a CD to a
ramdisk.
Loading data into ramdisk (40550 kB).............................
integrating the installation system into the ramdisk...
integrating the shared objects of the installation system...
starting syslog (messages are logged to /dev/tty4)...
starting klogd ...
integrating kernel modules of the installation system...
starting yast...
Figure 4-12 Loading installation system from CD
Next, the system prompts you for the type of terminal (Figure 4-13). In our case it
was VT100.
What type of terminal do you have ?
1)
2)
3)
4)
5)
6)
7)
8)
9)
VT100
VT102
VT220
X Terminal Emulator (xterm)
X Terminal Emulator (xterm-vt220)
X Terminal Emulator (xterm-sco)
X Terminal Emulator (xterm-sun)
Linux VGA or Framebuffer Console
Other
Type the number of your choice and press Return: 1
Figure 4-13 Type of terminal selection
Now, the actual SuSE Linux installation program, YaST2, starts. A VNC server
also starts in our case, because we specified the parameter “vnc=1” in the boot
prompt (Figure 4-11 on page 83). The VNC server displays the information you
need to access your host, either through a Web browser or using a VNC client
(see Figure 4-14 on page 85).
84
Implementing Linux with IBM Disk Storage
Please wait while YaST2 will be started
OK
starting VNC server...
a log can be found in /tmp/vncserver.log ...
***
***
***
***
You can connect to 10.10.10.78, display :1 now with vncviewer
Or use a Java capable browser on http://10.10.10.78:5801/
(When YaST2 is finished, close your VNC viewer and return to this window.)
Figure 4-14 Start of VNC server
Switch to the VNC client machine, and launch the VNC client application after
entering the values you gathered from the VNC server screen.
Figure 4-15 Enter IP address as specified in boot parameter
Figure 4-16 Enter the password you have defined
If all settings are correct, the VNC client launches a window, which displays the
output of YaST2 (see Figure 4-17).
Chapter 4. SuSE Linux on pSeries
85
Figure 4-17 YaST2 in VNC window
The installation steps that follow are trivial and very similar to other types of
installation (introduced at the beginning of this section) or even comparing to
other platforms, like zSeries and xSeries. Start by selecting a language, and you
will get a window as shown in Figure 4-18 where YaST2 allows you to override
default settings with your own (if needed). To change the default settings simply
click the blue underlined items (Mode, Keyboard Layout, Mouse, Partitioning or
Software). Each item leads to a new submenu where you can specify and apply
the changes.
The particularities of your environment and its intended usage dictate specific
settings, like partitions for instance, and are not detailed in this book. However,
since we are using an Emulex adapter for the purpose our experiment, we review
here the specifics of installing the Emulex device driver and the software
packages it requires.
The Emulex device driver comes as so called “open build” which basically means
that you have to compile sources to get a loadable module. An open build driver,
like the Emulex driver, differs from other source code drivers as it consists out of
two parts: the lower level driver component, which is delivered as object code,
86
Implementing Linux with IBM Disk Storage
and the higher level driver, which comes as source code. To make a new device
driver, the high level driver source must be compiled and the resulting object is
linked with the lower level driver object into a single loadable module.
To do this you have to install the following software packages:
򐂰 The kernel source
򐂰 Compiler (remember that you need the cross compiler for 64 bits)
򐂰 All binutils and libraries required by the compiler´
Figure 4-18 Installation settings
After you have made all your changes, click Accept (Figure 4-18) to save the
new settings.
YaST2 asks you for a final confirmation (Figure 4-19). After you accept, the
installation process starts, partitioning the disk(s). Please check your settings
before you confirm; specially disk partitioning, as it will wipe out any former data.
For more information and advice on disk partitioning, refer to Appendix A.,
“Storage from the OS view” on page 253.
Chapter 4. SuSE Linux on pSeries
87
Figure 4-19 Start of Linux installation
Once the disk is ready, the installation of the software packages begins
(Figure 4-20).
During the installation of the selected software packages, the system can
occasionally prompt you to insert another CD.
88
Implementing Linux with IBM Disk Storage
Figure 4-20 Package installation
Following the installation of all software packages, you are ready to finalize your
configuration.
First, make choice of a root password, create users as desired, and wait until the
corresponding configuration scripts are completed.
Finally, you can configure the network interfaces. The default entry is DHCP,
which is generally not the right selection for a server system. To change this
default, select Network Interfaces to get a configuration menu (see
Figure 4-21).
Note: All settings can be changed at anytime by starting YaST2 (either from a
command shell or in a graphical environment). YaST2 offers the dialog to
apply and commit the changes.
Chapter 4. SuSE Linux on pSeries
89
Figure 4-21 Installation settings of the hardware
Note: When you install your system via onboard video, you can configure your
video settings within the same Installation Settings dialog. (Figure 4-21 in that
case shows an additional underlined item for Video.)
Congratulations, your Linux installation is now complete.
4.4.2 Linux installation in an LPAR
We assume that you have the appropriate knowledge of handling high-end
pSeries systems running in a partitioned environment. Otherwise, the redbook
The Complete Partitioning Guide on IBM eServer pSeries Servers, SG24-7039,
and the LPAR section of the IBM eServer pSeries 690 Availability Best Practices
white paper might be a good start. See:
http://www.ibm.com/servers/eserver/pseries/hardware/whitepapers/p690_avail.
html
90
Implementing Linux with IBM Disk Storage
If not configured properly, any LPAR (including any that runs Linux) could impact
the availability of the whole Regatta system transcending the LPAR itself.
There are many different hardware and software levels on the p670 and p690.
These involve actual hardware parts, firmware upgrades, and software on the
HMC. To successfully run Linux in a partition on the p670/p690, the system must
be configured following the high availability configuration guide, and the system’s
firmware and HMC levels must be at GA2 or a higher level.
For additional details, and up-to-date information on supported hardware and
software configurations, please consult the Release Notes® available on the
pSeries Linux Web site:
http://www.ibm.com/servers/eserver/pseries/linux/linux_release_notes.pdf
For our tests, we use a p690 system with eight partitions defined (see
Figure 4-22).
Figure 4-22 HMC partition setup
The Linux partition, LPAR05, has one processor minimum, two as desired, and
four as maximum defined. Similarly, we have 1 GB of memory set as minimum,
Chapter 4. SuSE Linux on pSeries
91
2 GB as desired, and 4 GB as maximum.1 The boot mode is set to SMS (see
Figure 4-23). The SCSI adapter controlling the CD drive is allocated to the Linux
partition.
Figure 4-23 LPAR boot mode
Basically, installation on a LPAR is identical to the native Linux installation via
serial console. The differences are that:
You must not forget activate the partition.
When you reach the screen displayed in Figure 4-13, you must select option 9)
Other, and then VT320 for the terminal type.
Finally, you must update the boot configuration by selecting the appropriate boot
device in SMS Multiboot and changing the LPAR boot mode (see Figure 4-23) to
normal.
1
minimum are the resources needed at least, desired are assigned if available and minimum
requirements are met
92
Implementing Linux with IBM Disk Storage
4.4.3 Post-installation
Note: At the time of writing this book no multipath capabilities were available
yet. Neither did the Emulex device driver support multiple paths (required for
FAStT) nor was a SDD (required for ESS) available. The supported driver
release might be different in terms of installation and distribution.
At this time there is also no PPC 32-bit support.
SLES 8 contains out-of-the-box, the Emulex device driver V4.20. Prior to the
installation of the new version, any previous version of the driver must be
uninstalled; to do this, enter the following command from a commands shell:
rpm -qa | grep emulex
to get the name of the currently installed driver package. You will get a response
like emulex-4.20-63. To remove the package, enter the command:
rpm -e emulex-4.20-63.
After Linux is fully installed you have to install the package containing the Emulex
Open Build device driver.
Install the Emulex driver package you have obtained from SuSE as shown in
Figure 4-24. When released the new Emulex driver package will part of an
update CD or available on the SuSE support pages.
The package consists out of the following components:
򐂰
򐂰
򐂰
򐂰
򐂰
The source code for storage and networking (not covered in this book)
Tools
Library for the tools
Shell scripts for installation
Make files
We are assuming that the kernel sources, compiler, and libraries are already
installed. To compile the code successfully, make sure you meet the following
prerequisites:
򐂰 /usr/src/linux links to the kernel source; if not either establish the link or
edit the makefile by editing the BASEINCLUDE variable.
򐂰 Synchronize the versions of driver and kernel using the command.
򐂰 # cp /boot/vmlinuz.version.h
/lib/modules/<kernel-version>/build/include/linux/version.h
򐂰 Uninstall any existing Emulex driver package.
Chapter 4. SuSE Linux on pSeries
93
򐂰 Update the kernel and source to the required level.
Tip: The pre-release driver we used, required to build a new kernel image.
Otherwise, the driver failed to load because of unresolved symbols. However,
only the build is necessary, not the installation of the new kernel.
linux:/emulex/suse # rpm -Uvh emulex-4.21-0.src.rpm
emulex
##################################################
linux:/emulex/suse # rpm -Uvh km_emulex-4.21-0.ppc.rpm
km_emulex
##################################################
linux:/emulex/suse # rpm -Uvh emulex-4.21-0.ppc.rpm
emulex
##################################################
linux:/emulex/suse # cd /usr/src/kernel-modules/emulex/
linux:/usr/src/kernel-modules/emulex # ls -l
total 1285
drwxr-xr-x
3 root
root
584 Mar 26 07:41
drwxr-xr-x
3 root
root
72 Mar 26 07:41
-rw-r--r-1 root
root
1492 Feb 19 00:05
-rw-r--r-1 root
root
3739 Feb 21 22:48
-rw-r--r-1 root
root
3294 Feb 19 00:05
-rw-r--r-1 root
root
3826 Feb 21 22:48
-rw-r--r-1 root
root
3433 Feb 19 00:05
-rw-r--r-1 root
root
1227 Feb 19 00:05
-rwxr-xr-x
1 root
root
99812 Feb 21 22:48
-rw-r--r-1 root
root
294091 Feb 19 00:05
-rw-r--r-1 root
root
13378 Feb 19 00:05
drwxr-xr-x
2 root
root
368 Mar 26 07:41
-rwxr-xr-x
1 root
root
17716 Feb 21 22:48
-rw-r--r-1 root
root
27778 Feb 19 00:05
-rwxr-xr-x
1 root
root
108273 Feb 21 22:48
-rw-r--r-1 root
root
22686 Feb 19 00:05
-rw-r--r-1 root
root
136007 Feb 19 00:05
-rw-r--r-1 root
root
2362 Feb 19 00:05
-rw-r--r-1 root
root
401333 Feb 19 00:05
-rwxr-xr-x
1 root
root
146062 Feb 21 22:48
.
..
Install.sh
Makefile
Makefile.kernel
Makefile.module
README
Remove.sh
dfc
fcLINUXfcp.c
fcLINUXlan.c
include
libHBAAPI.so
libdfc.a
libemulexhbaapi.so
lpfc.conf.c
lpfc.conf.defs
lpfc_tar.spec
lpfcdriver
lputil
Figure 4-24 Installation of the Emulex driver packages
Change to the directory where the driver sources reside after copy. You can build
the driver with the command make build
To copy the driver to the appropriate directory, you enter make install
94
Implementing Linux with IBM Disk Storage
To test the module, load it using the command insmod lpfcdd
In order to load the module with the modprobe command you have to do the
following:
1. Add the line alias scsi_hostadapter lpfcdd to the file /etc/modules.conf
2. Run the command depmod -a to make the module dependencies.
3. Change the kernel append to append=”max_scsi_luns=128” (this allows the
kernel to support up to 128 LUNs, default is 0!)
To start your system with the driver available at boot time, you have to build an
appropriate initial ramdisk:
1. Add the name of the module to the file /etc/systemconfig/kernel
2. Rebuild your ramdisk with the command mk_initrd
3. Edit the file /etc/lilo.conf and add the configuration including the name of
the initial ramdisk (see Example 4-1). Do not forget to launch lilo to commit
the changes.
Example 4-1 Adding new initial ramdisk to lilo.conf
# Generated by YaST2
default=linux
timeout=100
boot=/dev/sda1
activate
image=/boot/vmlinuz-kernel_version
label=new_label
root=/dev/sda8
initrd=/boot/new_image_filename
read-only
append="max_scsi_luns=128"
After reboot, the driver is loaded. For details on how to make the storage usable
for Linux, please refer to Appendix A, “Storage from the OS view” on page 253.
Chapter 4. SuSE Linux on pSeries
95
96
Implementing Linux with IBM Disk Storage
5
Chapter 5.
Linux on xSeries and
BladeCenter
This chapter covers Linux installation on xSeries and BladeCenter in conjunction
with external storage.
First, it provides the step-by-step procedure for installing Red Hat Enterprise
Linux Advanced Server (AS) on IBM xServers and BladeCenter. Then, following
the same structure, we cover UnitedLinux 1.0 as part of the SuSE Linux
Enterprise Server (SLES 8) distribution.
Following the basic installation of both products, this chapter reviews required
post-installation tasks and caveats for specific server types.
The chapter ends with instructions and special considerations for attaching
external disk storage, and discusses the SDD multipath driver required for ESS,
while differences between the Red Hat and SuSE implementations are
highlighted.
© Copyright IBM Corp. 2001 2003. All rights reserved.
97
5.1 Introduction to xSeries, BladeCenter
IBM xSeries™ servers leverage IBM Enterprise X-Architecture™'s
state-of-the-art, industry-leading technologies that deliver mainframe-class
power, and enterprise scalability and availability at very attractive prices. IBM is
working closely with leading Linux distributors — Red Hat and the founders of
UnitedLinux including the SCO Group, SuSE Linux AG, and Turbolinux, Inc., to
offer tested and validated configurations for the full line of xSeries servers to
ensure maximum performance and functionality across the full line of xSeries
servers.
Blade servers are a relatively new technology that has captured industry focus
because of its modular design, which can reduce cost with a more efficient use of
valuable floor space, and its simplified management, which can help to speed up
such tasks as deploying, repositioning, updating, and troubleshooting hundreds
of Blade servers. All this can be done remotely with one graphical console using
IBM Director systems management tools. In addition, Blade servers provide
improved performance by doubling current rack density. By integrating resources
and sharing key components, not only will costs be reduced, but also, availability
will be increased. The BladeCenter is ideally placed for High Performance
Computing clusters and Web farms. For more details, see the following
Redpaper The Cutting Edge: The IBM eServer BladeCenter, REDP3581.
Figure 5-1 IBM BladeCenter chassis and Blade
Both Red Hat Enterprise Linux and SuSE Linux Enterprise Server provide
enterprise class features that enable Linux-based solutions to be deployed
across the widest range of enterprise IT environments.
98
Implementing Linux with IBM Disk Storage
5.2 Requirements
Prior to installation, we must first consider the system requirements. Especially,
the impact of the attached networks and storage on overall performance has to
be considered. Some recommendations to follow are:
򐂰 Design the storage subsystem carefully (RAID-level, stripe size, number of
hard drives versus capacity). See Tuning IBM eServer xSeries Servers for
Performance, SG24-5287.
򐂰 Check for appropriate cache policy usage on FAStT (see IBM TotalStorage
FAStT700 and Copy Services, SG24-6808).
Prior to installation, all firmware and BIOS levels must be verified and updated if
necessary. To get the latest information and downloads, visit the following site:
http://www.pc.ibm.com/support/
On the xSeries pages, you will find the latest drivers and firmware releases, as
well as documentation about installation for specific models.
Remember to also check compatibility of you disk systems by visiting:
http://www.storage.ibm.com/proven/index.html
5.2.1 Hardware requirements
The xSeries servers currently range from single processor to 16-way processor
systems. Memory ranges from 128 MB to 64 GB. Requirements will be
dependant on workload: for example, a large database would be well suited to a
system with plenty of memory and CPU power, whereas a Web server would be
better with more networking resources. Many options are available and
supported in IBM xSeries servers. You can view the latest list at the following site:
http://www.pc.ibm.com/us/compat/index.html
Red Hat and Suse also have a list of tested and supported hardware with their
distribution of Linux, respectively at:
http://hardware.redhat.com/hcl or
http://www.suse.com/us/business/certifications/certified_hardware
For attaching to external storage, you will need a Host Bus Adapter. You can find
the list of supported adapters by going to:
http://www.storage.ibm.com
After selecting your storage type, look for interoperabilty matrix.
Chapter 5. Linux on xSeries and BladeCenter
99
For the purpose of this book, we used an IBM FAStT FC2-133 (note that
although the name seems to indicate it is for FAStT, it can be used to attach to
other storage systems, e.g., ESS).
5.2.2 Software requirements
The following is a list of software used to set up Fibre Channel storage with the
xSeries and BladeCenter.
򐂰 Red Hat Enterprise Linux (release AS if setting up a high availability cluster),
or SuSE Linux Enterprise Server 8 (SLES8)
򐂰 IBM FAStT Host Bus Adapter drivers
If you attach to FAStT, you will need:
򐂰 Storage Manager
򐂰 IBM FAStT Management Suite for Java (MSJ)
If you attach to ESS, you will need:
򐂰 Subsystem Device Driver (SDD)
򐂰 ESS Specialist
To configure the BladeCenter FC switch modules you will need:
򐂰 BladeCenter SAN utility
5.2.3 Connection requirements
A network switch is required for systems management and to configure the Fibre
Channel storage. The servers may be connected to the storage via fibre switches
or hubs, or if there is only one to four servers they can be connected directly to
the FAStT with the correct amount of mini-hubs.
All connections have to be planned and set up carefully. Cable length
considerations are also important. The topic is beyond the scope of this book and
we assume that you are familiar with SAN and its implications. For an
introduction, please refer to Introduction to Storage Area Networks, SG24-5470.
Additionally, we assume that ESS and FAStT are set up with redundant paths
from the host system.
100
Implementing Linux with IBM Disk Storage
5.3 Configurations used for this redbook
In preparation of this redbook, we used equipment located at two different sites.
򐂰 Site 1: (We want to cover High Availability clustering for Red Hat, see
Chapter 6, “Red Hat Cluster Manager” on page 139, so we included two
servers.)
x440 (Node #1)
–
–
–
–
–
–
2x 1.5GHz CPUs, Hyper Threading disabled
4096MB RAM
2x 36.4GB Ultra3 10 Krpms HDDs
2x Intel 10/100 Ethernet Adapter
2x FC-2 HBA, BIOS v1.29
1x ServeRAID4Lx, BIOS v5.10
x440 (Node #2)
–
–
–
–
–
–
2x 1.4GHz CPUs, Hyper Threading disabled
2048MB RAM
2x 36.4GB Ultra3 10Krpms HDDs
2x Intel 10/100 Ethernet Adapter
2x FC-2 HBA, BIOS v1.29
1x ServeRAID4Lx, BIOS v5.10
BladeCenter
– 1x Management Module
– 2x Ethernet Switch Modules
– 2x Fibre 2 Port Switch Modules
HS20 Blade #1:
•
•
•
•
2.4GHz CPU
512MB RAM
40GB IDE HDD
2312 Fibre Expansion Card
HS20 Blade #2:
•
•
•
•
2.4GHz CPU
512MB RAM
40GB IDE HDD
2312 Fibre Expansion Card
STORAGE
– 1x FAStT700
– 1x EXP700
– 14x 18.2GB HDDs
Chapter 5. Linux on xSeries and BladeCenter
101
We are using direct connect with the x440’s for the storage as we only have
access to one 2109-F16 switch.
MISC
–
–
–
–
–
1x 4port KVM (Keyboard/Video/Mouse) switch
2x PDU's (Power Distribution Units)
1x 16port Netgear Fast Ethernet switch
1x 9513 Flat Panel Monitor
1x Enterprise rack
򐂰 Site2
x440
–
–
–
–
x440 with two INTEL Xeon 1.6 GHz MP
2 GB of memory
2x IBM FC2-133 Fibre Channel Host adapter in slot 5 and 6
2x 18 GB hard disk drives
Storage
–
–
–
–
2x FAStT700
4x EXP700 with 14 36 GB HDDs
2x IBM SAN switch 2109-F16
1x ESS Model 800
5.4 Implementing Linux
In this section, we review the tasks pertaining to the installation of Red Hat
Enterprise Linux on X Series and BladeCenter servers, then SuSE Linux
Enterprise server on X Series and BladeCenter successively.
5.4.1 Installation steps for Red Hat on xSeries
Boot your server from CD #1 of the Red Hat Enterprise Linux distribution, after
making sure that your system is set to boot from CD-ROM (this may require
changes to the BIOS settings). The screen shown in Figure 5-2 is displayed.
102
Implementing Linux with IBM Disk Storage
Figure 5-2 Initial boot screen
Press Enter at the initial boot window to begin the default graphical install.
A series of screens let you successively select the desired language and input
peripherals. If you cannot find an exact match for your mouse, select the generic
type that most closely resembles yours. If you are using a two-button mouse and
you wish to emulate a three-button mouse, check the Emulate three buttons
box. This is suggested for X-windows users, but is also handy in terminal mode
for cutting and pasting as well. The third or middle button is emulated by pressing
both buttons at the same time.
Read the text in the Welcome screen and click Next to continue (Figure 5-3).
Chapter 5. Linux on xSeries and BladeCenter
103
Figure 5-3 Welcome screen
Select a Custom installation and click Next (Figure 5-4).
Figure 5-4 Install Options screen
104
Implementing Linux with IBM Disk Storage
Be cautious if an installation has already been done on the drives. If you have
just created new logical drives on a ServeRAID™ controller, the installer will
prompt you to initialize the disk (Figure 5-5).
Note: Initialization of disks causes 100% data loss!
Figure 5-5 Initialize disk message
We have chosen a Custom installation. Then to set up the partitions, Disk Druid,
(a GUI tool similar to fdisk) is launched. You need to create the partitions. A type
(Linux, Swap,...) has to be chosen. After all partitions are defined, the mount
point of the individual disks has to be defined.
For example, you can choose four partitions on an empty drive with 18 GB
capacity:
򐂰
򐂰
򐂰
򐂰
For 20 MB, primary, select Linux, mount point /boot
For 1 to 2 times the physical memory, select SWAP.
For 3000MB, select Linux, mount point /usr
For the rest select Linux, mount point / (your root file system)
Please consult the Red Hat Linux Advanced Server Installation Guide and
Appendix A., “Storage from the OS view” on page 253 of this book for additional
information on partitions and file systems.
Selecting Have the installer automatically partition for you creates the
following partitions, and mount points:
򐂰 Swap is determined by the amount of memory in your system and the amount
of space available on your hard drive. If you have 128MB of RAM, then the
swap partition created can be 128MB - 256MB (twice your RAM), depending
on how much disk space is available. (The maximum swap partition size
possible is 2047MB)
򐂰 47 MB /boot
򐂰 2096 MB /
Chapter 5. Linux on xSeries and BladeCenter
105
In a server environment, it is recommended that you select Manually partition
with Disk Druid (Figure 5-6) and create your own partitions.
Figure 5-6 Manual Partitioning screen
To add a partition, click the New button (Figure 5-7).
106
Implementing Linux with IBM Disk Storage
Figure 5-7 Partitioning with Disk Druid
You will be prompted by the following window (Figure 5-8).
Figure 5-8 Adding a new partition
Chapter 5. Linux on xSeries and BladeCenter
107
To select the mount point, click the drop-down button next to the Mount Point
text box. This will present you with a menu of mount points such as /boot, /usr,
etc. You do not need to specify a mount point for a Swap partition.
Type in the size of the partition in megabytes.
To select the partition type, click the drop-down button next to the Filesystem
Type text box. This will present you with a menu of file system types such as
ext3, swap, and so on. For your Swap partition, use swap. For all other Linux
partitions use ext3.
Restriction: If you plan to attach to ESS and use the SDD driver, do not select
ext3, as it is not supported. Instead, use ext2, which is basically ext3 without
journaling.
Next, the Boot Loader is installed (Figure 5-9).
Figure 5-9 Boot Loader installation screen
Select the boot loader that the server will use. GRUB is the default.
If you choose to install a boot loader (GRUB or LILO), then you must determine
where it will be installed. If GRUB is the only boot manager you use, then choose
Master Boot Record (MBR). If you use the OS/2® boot manager, then GRUB
108
Implementing Linux with IBM Disk Storage
has to be placed on the boot partition (if you have a partition with /boot otherwise
/). We recommend installing GRUB on the Master Boot Record.
After making your selections, click Next.
If you selected the GRUB boot loader you will be asked if you wish to use a boot
loader password.
For highest security we, recommended setting a password, but this is not
necessary for more casual users. If used, enter a password and confirm the
password. Click Next to continue.
At the next step, the network is configured (Figure 5-10).
Figure 5-10 Network Configuration screen
Each network adapter will have a corresponding tab. For each adapter, you
should check the appropriate box to indicate whether the adapter:
򐂰 Will be configured via DHCP
򐂰 Should be activated on boot
If you are not using DHCP (highly likely for a server), you have to enter the
following:
򐂰 IP address
Chapter 5. Linux on xSeries and BladeCenter
109
򐂰 Netmask, Network, and Broadcast are automatically entered, but can be
edited if different.
򐂰 Hostname
򐂰 Gateway
򐂰 Primary DNS
򐂰 Secondary DNS (optional)
򐂰 Ternary DNS (optional)
At this point, a firewall can be configured. The proper configuration of a firewall
would be beyond the scope of this book. In our example, we used the
preselected Medium settings. Select your security level, and click Next.
Note: A properly configured firewall can greatly increase the security of your
system. For more details of the three security levels see the Red Hat Linux
Advanced Server Installation Guide.
Next, is the Language Support Selection. Red Hat Linux Advanced Server
supports multiple languages. You must select a language to use as the default
language. Choosing only one language significantly saves on disk space.
Now, the time zone has to be configured. Select your time zone by clicking the
appropriate location on the world map, or by selecting a location from the list
shown in the dialog. Click Next on the screens to proceed.
The next window takes you to the account configuration.
First, you set the Root Password. Please do this carefully to prevent unauthorized
use! Additionally, you can add other user accounts as well. Click Next.
Next, the authentication configuration is done (Figure 5-11).
110
Implementing Linux with IBM Disk Storage
Figure 5-11 Authentication Configuration screen
Clicking Enable MD5 passwords allows passwords up to 256 characters,
instead of the standard 8 characters or less.
Clicking Enable shadow passwords replaces the standard /etc/passwd file with
/etc/shadow, which can only be read by root user.
Clicking Enable NIS will allow you to add the Linux system to an existing NIS
domain via a specific server or broadcast.
Clicking Enable LDAP will allow authentication via an LDAP directory server.
Clicking Enable Kerberos will allow authentication via a Realm, KDC, or Admin
Server.
Clicking SMB will allow PAM to use an SMB server to authenticate users.
By default, Enable MD5 passwords and Enable shadow passwords are
selected.
After making your selection, click Next.
Chapter 5. Linux on xSeries and BladeCenter
111
The Selecting Package Group dialog box allows you to choose the packages you
wish to install. Please keep in mind that for any changes involving driver
compilation, the kernel sources have to be installed too (Figure 5-12).
Figure 5-12 Selecting Package Group screen
For the custom installation, you can select groups of packages such as
Mail/WWW/News Tools, or individual packages within these groups.
Scroll down the list and select the groups of packages you would like to install by
clicking the check box next to them.
If you wish to select individual packages, click the Select individual packages
check box. You should also check this box if you wish to install the supplied
summit kernel (Figure 5-13).1
1 If you install the summit kernel by selecting the individual packages, you will have to edit
/etc/grub.conf in order to boot the kernel, or see it in the GRUB menu. A preferred method
is to use the kernel-summit RPM post-install. The RPM script will automatically edit
/etc/grub.conf for you. See 5.5.1 “Configuring for the Summit kernel (Red Hat)” on
page 121.
112
Implementing Linux with IBM Disk Storage
Figure 5-13 Individual Package selection screen
Next, you can configure your video. It is useful to know which video card or
chipset you have in your system, however, you can have the installation probe
your system to see if it finds any hardware it recognizes.
You can also select the Skip X Configuration check box if you wish to configure
X after installation or not at all.
Now all required settings are made and you will get a screen telling you that the
installation is about to begin. Click Next to begin copying the operating system
files to your hard disk.
A screen similar to the one in Figure 5-14 keeps you informed over the progress
of the installation.
Chapter 5. Linux on xSeries and BladeCenter
113
Figure 5-14 Copying files screen
Once the file copy is complete, you are asked whether you want to have a boot
disk created. Insert a diskette if you require the boot disk, otherwise click the
Skip boot disk creation check box.
If you choose to use X you will configure your monitor and desktop settings on
the two next screens. Select the monitor model or the generic model that most
closely resembles it and click Next.
Important: Wrong monitor settings can damage your monitor!
Next, you can choose your default desktop and whether you would like to use a
text or graphical login.
Now the installation is complete and it is time to reboot. Please remove the CD,
and if generated, remove the boot disk (Figure 5-15).
If you intend to install on Red Hat on BladeCenter, please proceed to the next
section.
For post-installation tasks an issues, please proceed to 5.5, “Post-installation” on
page 121.
114
Implementing Linux with IBM Disk Storage
Figure 5-15 End of installation
5.4.2 Installation steps for Red Hat on BladeCenter
Ensure that all BIOS and firmware are at the latest levels and any switches are
installed prior to OS installation. See the Installation and User’s Guides for the
HS20 Fibre Channel Expansion Card, BladeCenter 2-Port Fibre Channel Switch
Module, and Management Module for more details.
It is possible to install the operating system over the network. Using PXE boot
allows multiple systems to be installed in parallel.
The following instructions assume you are installing Red Hat Enterprise Linux on
a local IDE disk and not SCSI or external Fibre Channel storage. We will also be
using CDs to install as we are only installing two servers. In the real world you
may be installing to fourteen servers in a single BladeCenter chassis. In this
situation, it would make more sense to perform a network installation. This would
allow you to install to the fourteen servers simultaneously.
Chapter 5. Linux on xSeries and BladeCenter
115
Note: At the time of writing, Red Hat Enterprise Linux was only supported by
IBM on the BladeCenter with IDE disks. Booting from SCSI was at that time
under test. Check the ServerProven® site for updates:
http://www.pc.ibm.com/us/compat/index.html
To install Red Hat Linux Advanced Server 2.1 on a BladeCenter you need a
special set of boot disks to allow the installer to recognize the USB CD-ROM and
USB floppy drives. The boot disks also allow graphical installs.
In addition, you need a driver disk to make use of the LSI SCSI controller, as well
as the included Broadcom gigabit network adapters.
Obtain the Red Hat Enterprise Linux boot diskette and driver diskette from the
following site:
http://redhat.com/support/partners/ibm/ibm_netfinity.html
Create the boot disk: boot-AS-2.1-bladecenter-<version>.img using dd under
Linux or rawrite under Microsoft Windows.
Create the driver disk: AS-2.1-dd.img using dd under Linux or rawrite under
Microsoft Windows. This disk is required for networking and if using SCSI disks.
Select the media tray and KVM on the first system and boot the system from the
diskette. Type linux dd when prompted for a graphical installation or text dd for
a text install.
Insert the driver diskette when prompted and press Enter.
Proceed with installation; see 5.4.1 “Installation steps for Red Hat on xSeries” on
page 102 for more details and screen shots.
Note: Installation on the BladeCenter will give you an additional prompt asking
if the installation method will be using Local CD-ROM or Harddisk, select Local
CD-ROM.
For post-installation tasks and issues, consult 5.5, “Post-installation” on
page 121.
116
Implementing Linux with IBM Disk Storage
5.4.3 Installation steps for SLES8 on xSeries
Power on your server and insert the SuSE SLES 8 CD in the CD drive, after
making sure that your system is set to boot from CD-ROM (this may require
changes to the BIOS settings). The screen shown in Figure 5-16 is displayed.
Figure 5-16 Boot prompt
The boot options field at the bottom of the screen allows you to specify additional
options for loading the kernel. On older systems, for instance, you may want to
disable DMA access to the CD-ROM by specifying ide=nodma as a boot option.
The push buttons at the very bottom of the screen allow you to choose text or
graphical installation modes.
Tip: If your system does not boot after changes to the bootloader, please
choose Rescue System , pick manually the module required to connect to
your storage, and boot. The rescue system allows you to make the required
repairs like writing a new MBR.
Press Enter or wait until the kernel is loaded automatically.
Chapter 5. Linux on xSeries and BladeCenter
117
Once the kernel is loaded, the YaST installation utility starts and displays the
licence agreement, followed (upon your acceptance) by the language selection
screen.
The installation utility is now probing your hardware, loading (if available) the
appropriate modules for the devices found. It is important to note that SuSE will
only install drivers required to boot: from a disk driver standpoint, this avoids the
problem potentially caused by having different drivers attach differently to the
SCSI implementation; loading a second driver may cause a shift in the
numbering of the SCSI devices.
After completing the hardware detection cycle, YaST presents you the screen
shown in Figure 5-17. You can review and change Installation settings and
make sure they fit your actual requirements. In particular, you will most likely
change the boot disk partitioning and the software packages selection.
Figure 5-17 SLES 8 Installation Settings screen
Please refer to section “Disk partitions” on page 255 in Appendix A for more
information on disk partitioning under Linux. There are some system
requirements that must be respected when you partition the disk, but it is also
118
Implementing Linux with IBM Disk Storage
important to make sure that you select a file system suited for the type of
application and storage system you are using.
When going through the software package selection, be careful to include all
required sources (e.g. ncurses), kernel sources, compiler, and libraries. These
packages contain the development environment you will need to compile the
device drivers for the HBAs. After you have made your changes, click the Accept
button (see Figure 5-17) to set and save the new settings.
YaST informs you that all settings are saved and that the installation is about to
begin. Before clicking the Yes button, make sure one more times that you have
set the appropriate settings; once you accept, the installation process starts with
partitioning and data on disk, if any, will be lost.
The process continues with the installation of the selected packages from the
SLES 8 and the UnitedLinux CDs.The system prompts you whenever it requires
another CD in the set.
The process is complete when you are prompted to reboot; please remember to
remove any bootable CD and diskette.
After reboot, you are now ready to configure your system. You begin the
configuration by defining the root password, then you can eventually create
additional users.
YaSt now displays suggested defaults for your desktop settings. Select Change if
you want to adjust or select your own video, mouse, and keyboard setup. Once
you accept, the values are committed.
YaST now probes your system for additional peripherals like printers.
On the next screen (shown in Figure 5-18) you can review and change the
settings for printer and all communication devices including the network, firewall,
and server services. By default, the detected network adapters are set for DHCP.
If you want a static address, which is usually the case for a server, click the
Network Interfaces headline to make the appropriate changes.
Clicking Next saves the network configurations and applies all changes to the
system (see Figure 5-19).
The system now reboots and starts the configured services with the settings
selected.
The installation is complete.
Chapter 5. Linux on xSeries and BladeCenter
119
Figure 5-18 Network and peripherals setup
Figure 5-19 Apply network configuration
120
Implementing Linux with IBM Disk Storage
5.4.4 Installing SuSE Linux on BladeCenter
The Installation of UnitedLinux/SLES on the BladeCenter servers is very similar
to 5.4.3 “Installation steps for SLES8 on xSeries” on page 117. There are a few
particularities to be aware of though.
Mouse during installation
The mouse will not work initially when the installer starts. Use the keyboard to
start the installation until you get to the Installation Settings screen. Press Alt+C
to change the settings. Select Mouse from the menu, scroll down the list and
select USB Mouse. Press Alt+T to test and then click Accept when finished.
Single processor or hyper-threading
If the server has more than one processor or has hyper-threading enabled, select
Booting, and add the parameter acpi=oldboot in the Kernel Boot Parameters
field.
5.5 Post-installation
This section addresses post-installation tasks that are critical, and thus required
depending on your hardware or Linux distribution.
5.5.1 Configuring for the Summit kernel (Red Hat)
This only applies to Red Hat.
IBM’s new high-end Intel based servers take advantage of the new IBM Summit
chipsets. Summit is the name for a set of technologies inspired by proven
mainframe capabilities. Summit enables mainframe inspired capabilities in the
areas of availability, scalability, and performance. Enhanced features include:
򐂰 Memory mirroring, redundant bit steering, chipkill memory
򐂰 L4 cache providing high speed communications between memory and CPU
subsystems
򐂰 High performance PCI-X allows adapters significantly better memory access
򐂰 Increased scalability
We are using x440 servers, which include the Summit chipset. If you are using an
x360 or a server without the Summit chipset, then you can move to the next
section.
Chapter 5. Linux on xSeries and BladeCenter
121
If you have an SMP system and you have booted the SMP or enterprise kernel,
you will find that the loading of Linux will hang. You should boot into the system
BIOS and enable hyper-threading:
Press F1 when prompted at the system BIOS. From the menu select Advanced
Setup -> CPU Options. Enable Hyper-Threading Technology.
Once enabled you will be able to boot with the SMP or Enterprise kernels. The
Summit kernel should be used on servers with this technology except the x360.
Once the Summit kernel is installed hyper-threading can be switched off again if
not required.
The summit kernel, source, and patch are supplied with the distribution CDs if
required.
Obtain the latest supported summit kernel, headers, and source from the
following site:
http://www.rhn.redhat.com
You will need to register your licensed copy of Red Hat Enterprise Linux to obtain
updates for this OS.
Install the updates:
rpm
rpm
rpm
rpm
-ivh
-Uvh
-Uvh
-ivh
kernel-summit-<version>.rpm
kernel-headers-<version>.rpm
kernel-source-<version>.rpm
kernel-<version>.src.rpm
An entry will be added to /etc/grub.conf , which will allow you to boot from the
Summit kernel. Edit the default field to boot from this kernel by default.
When using the summit kernel, it is also necessary to install a patch that is
supplied with the kernel<version>.src.rpm file. Type the following commands to
install the patch:
cd /usr/src/linux-<version>
patch -p1 < /usr/src/redhat/SOURCES/linux-<version>-summit.patch
cp -f configs/kernel-<version>-i686-summit.congif.config
5.5.2 Updating the kernel on BladeCenter (Red Hat)
This only applies to Red Hat. Obtain the latest supported kernel, headers, and
source from the following site:
http://www.rhn.redhat.com
You will need to register your licensed copy of Red Hat Enterprise Linux to obtain
updates for this OS.
122
Implementing Linux with IBM Disk Storage
Install the updates:
rpm
rpm
rpm
rpm
-ivh
-Uvh
-Uvh
-ivh
kernel-<version>.rpm
kernel-headers-<version>.rpm
kernel-source-<version>.rpm
kernel-<version>.src.rpm
An entry will be added to /etc/grub.conf , which will allow you to boot from the
new kernel. Edit the default field to boot from this kernel by default.
5.5.3 Issues on BladeCenter
This section applies to Red Hat and SuSE.
Floppy after installation
You may be unable to mount the floppy after installation due to it being a USB
device.
Red Hat
The floppy resides on the last SCSI device. For example, if you have LUNs
configured on the fibre storage (/dev/sda, /dev/sdb), the floppy will reside on
/dev/sdc. It is possible to manually mount the floppy using this information, or
/etc/fstab can be edited with the following where X=device letter:
/dev/sdX
/mnt/floppy
auto
noauto,user
0,0
Create a directory for the floppy under /mnt
mkdir /mnt/floppy
The system is now configured to mount the floppy at boot time. To mount the
floppy now type the following:
mount /mnt/floppy
SuSE
Type the following to load the USB storage module:
modprobe usb-storage
The floppy resides on the last SCSI device. For example, if you have LUNs
configured on the fibre storage (/dev/sda, /dev/sdb), the floppy will reside on
/dev/sdc. It is possible to manually mount the floppy using this information, or
/etc/fstab can be edited with the following where X=device letter:
/dev/sdX
/media/floppy
auto
noauto,user,sync
0 0
Create a directory for the floppy under /media
Chapter 5. Linux on xSeries and BladeCenter
123
mkdir /media/floppy
The system is now configured to mount the floppy at boot time. To mount the
floppy now type the following:
mount /media/floppy
CD-ROM after installation
You may be unable to mount the CD-ROM after installation due to it being a USB
device.
Red Hat
The CD-ROM will fail to mount if the media tray had been previously switched to
another Blade system. Edit /etc/fstab to change /dev/cdrom to /dev/scd0:
/dev/scd0
/mnt/cdrom
iso9660
noauto,owner,kudzu,ro
0,0
The system is now configured to mount the CD-ROM at boot time. To mount the
CD-ROM, now type the following:
mount /mnt/cdrom
If this system is rebooted and the media tray is moved to another system then
this process may need to be repeated.
SuSE
The CD-ROM device will be known as /dev/sr0, so remove the device /dev/cdrom
and create a link from sr0 to cdrom. Type the following:
rm /dev/cdrom; ln -s /dev/sr0 /dev/cdrom
Edit /etc/fstab to include the following line of text:
/dev/cdrom
/media/cdrom
auto
ro,noauto,user,exec
0 0
Type mount /dev/cdrom /media/sr0 to manually mount a CD now.
Mouse after installation - SuSE only
The mouse may not work after the server restarts. Type the following to regain
mouse controls:
/sbin/modprobe mousedev
Type the following to avoid having to repeat this every time the server reboots:
echo “/sbin/modprobe mousedev” >> /etc/init.d/boot.local
124
Implementing Linux with IBM Disk Storage
5.6 Setup for Fibre Channel attachment
In this section we discuss the setup required for Fibre Channel attachment for
Red Hat Enterprise Linux and Suse Linux Enterprise Server.
First, install the Host Bus Adapters in your server.
Next, replace and update the BIOS on the adapter and apply the proper settings.
򐂰 For a Red Hat installation, you must now replace the default driver that was
automatically installed by an appropriate driver for your storage (ESS or
FAStT).
򐂰 For a Suse installation, no default driver was installed, and you must now
install the appropriate driver for your storage (ESS or FAStT).
Finally, install SDD if you attach to ESS, or configure your paths if you attach to
FAStT.
As described in 5.3 “Configurations used for this redbook” on page 101, we used
IBM FC2 Host Bus Adapters (HBA). Although this HBA is also known as the IBM
FAStT FC2, it does not mean that it is exclusively used for attachment to a FAStT.
Note also that the device driver supports all IBM HBAs labelled with FAStT (FC1
and FC2).
The use of the IBM FC-2 adapter and the configuration described here after
apply whether you will attach to FAStT or ESS. The only difference is that the
ESS requires the single path driver for the HBA.
5.6.1 Install the adapters
If the IBM FC2 HBAs are not already installed, shut down your system and install
the adapters. At this stage it is best to either leave the cards disconnected from
the SAN, or have the SAN switched off. Otherwise, if you boot your server to the
OS, it is now possible that the sequence of the SCSI devices could change, in
which case the boot would fail. This situation will be corrected later in this
chapter.
Red Hat system
If you are running an “out-of-the-box” installation of Red Hat Enterprise Linux,
then KUDZU probes for new hardware during boot. All detected hardware is
compared with entries in the KUDZU database at /etc/sysconfig/hwconfig. If this
file does not exist for any reason, then /etc/modules.conf,
etc/sysconfig/network-scripts, and /etc/X11/XF86Config are scanned instead. If
new hardware is detected, a dialog offers three choices:
Chapter 5. Linux on xSeries and BladeCenter
125
Configure:
Include the adapter(s) into your configuration.
Ignore:
Ignore the new hardware in the future.
Do Nothing:
Do nothing, continue to boot.
Select Configure: A driver will be installed for the hardware.
The standard driver delivered with the distribution is not capable of multipathing.
If you attach to ESS, you can use this single path driver. If you attach to FAStT,
you need to replace it with an available multipath driver.
SuSE system
SuSE ignores the adapters, and you can simply proceed with the installation of
the new device drivers.
5.6.2 Prepare the Fibre Channel Host Bus Adapter cards
Obtain the latest IBM FAStT BIOS update file:
http://www.pc.ibm.com/support/
Create the bootable diskette and use it to boot the server. You should end up with
a DOS C:\ prompt. The update will do all like HBAs in the server at the same
time. (Dissimilar HBAs need to be updated separately.)
򐂰 For xSeries
Type the following to update the HBA BIOS:
flasutil /f /l
Type the following to set the defaults:
flasutil /u
Remove the boot diskette and restart the server.
򐂰 BladeCenter
The /i option is required for the BladeCenter to identify the vendor. Type the
following to update the HBA BIOS:
flasutil /f /l /i
Type the following to set the defaults:
flasutil /u /i
Remove the boot diskette and restart the server.
Settings need to be changed in the HBA configuration for use with Linux. These
settings should be changed after the BIOS update. Watch for the IBM FAStT
BIOS and press Ctrl-Q when prompted to enter the setup utility.
126
Implementing Linux with IBM Disk Storage
Select the adapter to change. Select Host Adapter Settings.
򐂰 Loop reset delay - Change this setting to 8.
򐂰 Adapter Port Name - Note this number as it will be required later for storage
partitioning.
Select the advanced adapter settings:
򐂰 LUNs per target - This should be 0 (zero)
򐂰 Enable target reset - This should be yes
򐂰 Port down retry count - Change this to 12
Repeat this process for any other HBAs in the server, save the settings, and exit
to restart the server.
5.6.3 Installing the Fibre Channel Host Bus Adapter driver
Obtain the latest IBM FAStT HBA driver from IBM. The BladeCenter driver will be
a separate package from the xSeries driver.
As mentioned earlier, two different packages exist for the IBM FAStT HBA family:
򐂰 A fail-over (or multipath) device driver for FAStT attachment
򐂰 A singlepath driver for ESS attachment
The procedure to install the sources is identical for both drivers.
Copy the i2xLNX-<driver version>-fo-dist.tgz file to a folder on your system
and type the following to extract the files:
tar zxvf *.tgz
This also creates an ./i2x00-<driver version> directory.
Change to this directory and run the drvrsetup script to extract the source:
cd i2x00-<version>
sh drvrsetup
Table 5-1 shows the contents of the driver archive folder.
Table 5-1 Contents of the driver archive
drvrsetup
Script file to copy driver source files
included in the driver source tgz file
i2xLNXsrc-<driver version>.tgz
driver source archive.
libinstall
Script file to install/setup HBA API
library.
Chapter 5. Linux on xSeries and BladeCenter
127
libremove
Script file to remove HBA API library.
qlapi-<api lib version>-rel.tgz
Compressed binary distribution file
for API library.
ipdrvrsetup
Script file to copy IP driver source
files included in the IP driver source
tgz file
qla2xipsrc-<IP driver version>.tgz
Compressed binary distribution file for
IP driver sources
For Red Hat, the drivers require that the source headers be prepared before
compilation. Change to the kernel source directory:
cd /usr/src/linux2.4
Change the Extraversion field in the Makefile to correspond with your kernel
version. The default setting with Red Hat Enterprise Linux is custom.
Run make menuconfig to load the Linux Kernel Configurator tool.
For SuSE you may need to run make cloneconfig to pull in the current kernel
configuration prior to running make menuconfig. Check the processor setting is
correct (SMP, cyclone support for x440 (not required for x360, not necessary but
suggested for x440 with one CEC; this features introduces a generalized time to
all CECs), exit and save the configuration. Run make dep to rebuild the kernel
dependencies.
The fibre HBA driver source can now be compiled. Change back to the directory
containing the driver source. The following commands will generate qla2200.o
and qla2300.o modules:
For a uni-processor system type:
make all
For an SMP system type:
make all SMP=1
For SuSE systems add OSVER=linux, for example:
make all SMP=1 OSVER=linux
The BladeCenter Fibre Channel Expansion cards use the qla2300.o module.
After removing the existing drivers, copy the required driver module to the
following location:
/lib/modules/’uname -r’/kernel/drivers/addon/qla2200/
128
Implementing Linux with IBM Disk Storage
5.6.4 Loading the FAStT HBA driver
The following sections explain how to load the FAStT HBA driver for Red Hat and
SuSE respectively.
Red Hat
Edit /etc/modules.conf to ensure that the IBM FAStT modules are loaded after
any local SCSI module. Note in the example that the ServeRAID module, ips, is
loaded before the IBM FAStT modules. Your modules.conf file should look similar
to the following:
alias
alias
alias
alias
eth0 bcm5700
scsi_hostadapter ips
scsi_hostadapter2 qla2300
scsi_hostadapter3 qla2300
The following line should also be added in order to support more than one SCSI
device per adapter (Red Hat Enterprise Linux supports up to 40 LUNs by
default):
options scsi_mod max_scsi_luns=128
Save modules.conf and exit. Run depmod -a to update the modules.dep file.
You can load the fibre HBA module manually by typing the following:
modprobe qla2300
Alternatively, you can have it loaded at boot time by having it load as part of the
ramdisk image. To build a new ramdisk image type the following:
mkinitrd /boot/<newname>.img <kernel version>
In our case, we typed the following:
mkinitrd /boot/initrd-storage-2.4.9-e.12summit.img 2.4.9-e.12summit
Edit /etc/grub.conf so that the boot loader knows to use the new ramdisk
image. Edit the relevant kernel entry to load to new ramdisk image. At this point,
you can also edit the default entry to load your new kernel and ramdisk image
as default. It would be advisable to test that it boots successfully before changing
the default.
Shut down the server and connect the SAN or bring the SAN online. Once online,
bring the server back up. (As an alternative to bringing the server down you could
unload the IBM FAStT module, (modprobe -r qla2300) attach the SAN, and
reload the IBM FAStT module (modprobe qla2300).)
If you will attach to FAStT, you may proceed to 5.8, “Configuring FAStT” on
page 137.
Chapter 5. Linux on xSeries and BladeCenter
129
SuSE
Edit /etc/sysconfig/kernel and add qla2x00 to the string behind
INITRD_MODULES where x represents either 2 or 3:
INITRD_MODULES=”aic7xxx reiserfs qla2300” .
Run depmod -a next.
The script /sbin/mk_initrd uses this string as input. By launching mk_initrd , a
new version of the initial ramdisk is built in the /boot folder. Even though this is
convenient, you might want to built a new ramdisk rather than replacing the
existing one.
Before you restart you should add the kernel parameter max_scsi_luns=128 to
the grub.conf or lilo.conf file. This is necessary because the kernel does not
allow the full range support of LUNs (most desktop versions support no LUN out
of the box). If you are using LILO run /sbin/lilo to commit the changes.
You can load the fibre HBA module manually by typing the following:
modprobe qla2300
Shut down the server and connect the SAN or bring the SAN online. Once online
bring the server back up. (As an alternative to bringing the server down, you
could unload the IBM FAStT module, (modprobe -r qla2300) attach the SAN, and
reload the IBM FAStT module (modprobe qla2300)).
If you will attach to FAStT, you may proceed to 5.8, “Configuring FAStT” on
page 137.
5.6.5 Attaching to ESS
Linux distributors do not offer a multipath driver that can be used when attaching
to ESS as is the case with FAStT. However, one can achieve redundant pathing
and load balancing by using a multipath extension called Storage Device Driver
(SDD), available from IBM. The particularity and advantage of SDD is that it
operates independently of the HBA device driver (SDD is a pseudo device
driver). Therefore, the following applies for any HBA compatible with the ESS.
The SDD introduces a new tree in the /dev file system with the name /vpath.
This is one reason why SDD does not support LVM. In LVM the supported
components of the /dev directory are hard coded.
SDD actually consists of a driver and a server daemon. The server daemon runs
in the background at all times (it starts automatically after the SDD driver
package is installed). It scans all paths at regular intervals for an indication of
130
Implementing Linux with IBM Disk Storage
failure. A path can have the states OPEN, CLOSE DEAD, DEAD, and INVALID.
The SDD server reclaims paths and changes their state according to Table 5-2.
Table 5-2 SDD path probing and reclaiming
Path status before probe
Path status while
probing
New path status
INVALID, CLOSED DEAD
OPEN
OPEN
CLOSED DEAD
CLOSE
CLOSE
DEAD
OPEN
OPEN
OPEN
not operational
DEAD
CLOSE
not working
CLOSE DEAD
Installing the SDD driver
Before you install SDD, please make sure that you meet all the hardware and
software requirements. For up-to-date information, consult:
http://www.storage.ibm.com/disk/ess/supserver.htm
Note: At the time of writing this redbook SDD was still under testing for the
Linux kernel we used.
Before installing SDD, you must configure the Fibre Channel adapters and the
adapter drivers that are attached to the Linux host system, as explained in 5.6.2,
“Prepare the Fibre Channel Host Bus Adapter cards” on page 126, and 5.6.3,
“Installing the Fibre Channel Host Bus Adapter driver” on page 127.
In addition, before you install SDD, you must configure your ESS for multiport
access for each LUN (use the ESS Specialist for that purpose). SDD requires a
minimum of two independent paths that share the same LUN to use the
load-balancing and path failover protection features.
Strictly speaking, note that a host system with a single Fibre Channel adapter
connected through a switch to multiple ESS ports is considered a multipath Fibre
Channel connection, but is not sufficient (SDD will not provide failover
capabilities even if you install it in this case).
To install SDD, type:
rpm -iv IBMsdd-N.N.N.N -1.i686.redhat.rpm
rpm -iv IBMsdd-N.N.N.N -1.i686.suse.rpm
(where N.N.N.N represents the current version release modification level
number; N.N.N.N = 1.3.0.1 for example).
Chapter 5. Linux on xSeries and BladeCenter
131
The installation copies the driver and its utilities in the/opt/IBMssd directory.
Table 5-3 lists all components installed and their location.
Table 5-3 SDD components for a Linux host system
File
Directory
Description
sdd-mod.o-xxxy
/opt/IBMsdd
SDD device driver file
(where XXX stands
for the kernel level of your
host system
and y represents smp or up
vpath.conf
/etc
executables
/opt/IBMsdd/bin
/usr/sbin
sdd.rcscript
/etc/init.d/sdd
SDD configuration and
status tools
Symbolic links to the SDD
utilities
Symbolic link for the SDD
system startup option
Symbolic link for the SDD
manual start or
restart option
/usr/sbin/sdd
Configuring SDD
You can manually or automatically load and configure SDD on your Linux host
system. Manual configuration requires that you use a set of SDD-specific
commands, while automatic configuration requires a system restart.
Table 5-4 CLI for SDD
132
Command
Description
cfgvpath
Configures vpath devices
cfgvpath query
Displays all sd devices
lsvpcfg
Displays the current devices that are
configured and their
corresponding paths
rmvpath
Removes one or all vpath devices
sdd start
Loads the SDD driver and automatically
configures disk devices for
multipath access.
sdd stop
Unloads the SDD driver (requires no vpath
devices currently in use)
Implementing Linux with IBM Disk Storage
Command
Description
sdd restart
Unloads the SDD driver (requires no vpath
devices currently in use), and then loads
the SDD driver and automatically
configures disk devices for multipath
access.
Using multiple commands for SDD configuration
To load and configure the SDD on your system, manually change to the directory
/opt/IBMsdd and enter the command insmod modulename (in our case, and the
module name was sdd-mod.o-2.4.19.SuSE-152.k_smp-2.4.19-233 for SuSE
SLES 8).
If you do not know your exact kernel version, use the command uname -a to list it.
To verify whether the SDD sdd-mod driver is loaded type cat /proc/modules.
To verify that the driver is loaded correctly please enter lsmod. Figure 5-20 shows
an example of the expected output.
linux:/etc # lsmod
Module
videodev
ide-cd
isa-pnp
ipv6
sdd-mod
st
sr_mod
cdrom
sg
joydev
evdev
input
usb-uhci
usbcore
af_packet
bcm5700
lvm-mod
qla2300
aic7xxx
Size
6272
30564
32608
183636
355256
29260
14520
29120
30272
6112
4800
3488
24492
64832
15112
83592
68192
202944
129752
Used by
Tainted: P
0 (autoclean)
0 (autoclean)
0 (unused)
-1 (autoclean)
0
0 (autoclean) (unused)
0 (autoclean) (unused)
0 (autoclean) [ide-cd sr_mod]
0 (autoclean)
0 (unused)
0 (unused)
0 [joydev evdev]
0 (unused)
1 [usb-uhci]
1 (autoclean)
1
0 (autoclean)
0
2
Figure 5-20 Loaded SDD module
Chapter 5. Linux on xSeries and BladeCenter
133
Change to the /opt/IBMsdd/bin directory and enter cfgvpath query as shown in
Figure 5-21.
linux:/opt/IBMsdd/bin # cfgvpath
/dev/sda ( 8, 0) host=1 ch=0
serial=xxxxxxxxxxxx
/dev/sdb ( 8, 16) host=2 ch=0
serial=70024678
/dev/sdc ( 8, 32) host=3 ch=0
serial=70024678
query
id=13 lun=0 vid=IBM-ESXS pid=ST318452LC
ctlr_flag=0 ctlr_nbr=0 df_ctlr=0 X
id=0 lun=0 vid=IBM
pid=2105800
ctlr_flag=0 ctlr_nbr=0 df_ctlr=0
id=0 lun=0 vid=IBM
pid=2105800
ctlr_flag=0 ctlr_nbr=0 df_ctlr=0
Figure 5-21 Output of cfgvpath
The sample output shows the name and serial number of the SCSI disk device,
its connection information, and its product identification. A capital letter X at the
end of a line indicates that SDD currently does not support the device.
Type the cfgvpath as shown in Figure 5-22 to configure the SDD vpath devices.
This merges the SCSI devices sdb and sdc into one virtual device, vpatha
(actually, the effect is to make sdb and sdc dress the same target).
linux:/opt/IBMsdd/bin # cfgvpath
crw-r--r-1 root
root
253,
0 Apr 4 06:03 /dev/IBMsdd
Making device file /dev/vpatha 247 0
Added vpatha ...
linux:/opt/IBMsdd/bin # ls /dev/vpath*
/dev/vpatha
/dev/vpatha11 /dev/vpatha14 /dev/vpatha3 /dev/vpatha6
/dev/vpatha9
/dev/vpatha1
/dev/vpatha12 /dev/vpatha15 /dev/vpatha4 /dev/vpatha7
/dev/vpatha10 /dev/vpatha13 /dev/vpatha2
/dev/vpatha5 /dev/vpatha8
linux:/opt/IBMsdd/bin # cat /etc/vpath.conf
vpatha 70024678
Figure 5-22 Configure SDD vpath devices
The configuration information is saved by default in the /etc/vpath.conf file to
maintain vpath name persistence in subsequent driver loads and configurations.
You can remove an SDD vpath device by using the rmvpath xxx command,
where xxx represents the name of the vpath device that is selected for
removal.Type cfgvpath ? or rmvpath ? for more information about the cfgvpath
or rmvpath commands.
134
Implementing Linux with IBM Disk Storage
To verify the vpath configuration enter lsvpcfg or datapath query device. If you
successfully configured SDD vpath devices, output similar to the following is
displayed by lsvpcfg:
000 vpath0 (247,0)70024678 =/dev/sdb /dev/sdl /dev/sdcg /dev/sdcp.
In the following sections we describe three different modes of operation:
Using a single command for SDD configuration
You can also manually load and configure SDD by issuing the sdd start
command. Successful execution of the sdd start command performs all the
tasks described in “Using multiple commands for SDD configuration” on
page 133.
Use the sdd stop command to unconfigure and unload the SDD driver. Use the
sdd restart command to unconfigure, unload, and then restart the SDD
configuration process.
Configuring SDD at system startup
You can set up SDD to automatically load and configure when your Linux system
boots. SDD provides the startup script sdd.rcscript file in the /opt/IBMsdd/bin
directory and creates a symbolic link to /etc/init.d/sdd.
The following steps are necessary to launch SDD at your Red Hat system
startup:
1. Log on to your Linux host system as the root user.
2. Type chkconfig --level X sdd on to enable run level X at startup (where X
represents the system run level).
3. Type chkconfig --list sdd to verify that the system startup option is enabled
for SDD configuration.
4. Restart your host system so that SDD is loaded and configured.
5. If necessary, you can disable the startup option by typing chkconfig --level
X sdd off.
The following steps are necessary to launch SDD at your SuSE system startup:
1.
2.
3.
4.
Log on to your Linux host system as the root user.
Type insserv sdd
Restart your host system so that SDD is loaded and configured.
If necessary, you can disable the startup option by typing insserv -r sdd
In order for SDD to automatically load and configure, the HBA driver must
already be loaded. This can be assured at start time by adding the appropriate
Chapter 5. Linux on xSeries and BladeCenter
135
driver(s) to the kernel’s initial ramdisk (see 5.8 “Configuring FAStT” on page 137
for details to set up the initial ramdisk).
Partitioning SDD vpath devices
Disk partitions are known as logical devices. SDD for Linux allows configuration
of whole devices only. The SDD naming scheme for disks and disk partitions
follows the standard Linux disk naming convention. The following description
illustrates the naming scheme for SCSI disks and disk partitions:
򐂰 The first two letters indicate the SCSI device.
򐂰 The next letter (or two letters), a-z, specifies the unique device name.
򐂰 A number following the device name denotes the partition number. For
example, /dev/sda is the whole device, while /dev/sda1 is a logical device
representing the first partition of the whole device /dev/sda. Each device and
partition has its own major and minor number.
Similarly then, a specific device file /dev/vpathX is created for each supported
multipath SCSI disk device (where X represents the unique device name; as with
sd devices, X may be one or two letters).
Device files /dev/vpathXY are also created for each partition of the multipath
device (where Y represents the corresponding partition number). When a file
system or user application wants to use the logical device, it should refer to
/dev/vpathXY (for example, /dev/vpatha1 or /dev/vpathbc7) as its multipath
logical device. All I/O management, statistics, and failover processes of the
logical device follow those of the whole device.
The output in Figure 5-22 on page 134 demonstrates how the partitions are
named.
Note:
򐂰 SDD does not support the use of root (/), /var, /usr, /opt, /tmp and swap
partitions.
򐂰 For supported file systems, use the standard UNIX fdisk command to
partition vpath devices (e.g. fdisk /dev/vpatha).
To partition the new device please refer to Appendix A, “Storage from the OS
view” on page 253. Please keep in mind that you have to start fdisk with the
proper vpath as parameter (e.g. fdisk /dev/vpatha).
136
Implementing Linux with IBM Disk Storage
5.7 BladeCenter Fibre Switch Module configuration
The BladeCenter SAN Utility application is used to access and configure the
BladeCenter Fibre Channel switch modules. The SAN Utility can be installed on
a BladeCenter HS20 Blade server or an external network management
workstation configured with a supported version of Linux.
Before you configure your switch module, be sure that the management modules
are properly configured. In order to access your switch module from an external
environment, you may need to enable certain features, such as external ports
and external management over all ports. See the applicable BladeCenter Unit
Installation and User’s Guide publications on the IBM eServer BladeCenter
Documentation CD.
Installing the BladeCenter SAN Utility is detailed in Chapter 8. “BladeCenter SAN
Utility” on page 177.
5.8 Configuring FAStT
It is necessary to set up the FAStT and configure LUNs to be accessed by the
Linux system. The configuration of the FAStT is done using the Storage Manager
tool and the FAStT MSJ (Management Suite for Java). Please refer to Chapter 9.
“FAStT Storage Manager” on page 187, and Chapter 10. “FAStT MSJ
(Management Suite for Java)” on page 211 on how to install and use these.
5.9 Configuring ESS
It is necessary to set up the ESS and configure LUNs to be accessed by the
Linux System. In addition, as mentioned earlier, you must configure your ESS for
multiport access for each LUN before you install SDD. This is done using the
Storage Specialist. Please refer to Chapter 7., “Configuring ESS for Linux” on
page 155.
Chapter 5. Linux on xSeries and BladeCenter
137
138
Implementing Linux with IBM Disk Storage
6
Chapter 6.
Red Hat Cluster Manager
This chapter describes the preparation, installation, and customization tasks
required for the Red Hat High Availability Cluster Manager.
For details, refer to the Red Hat Cluster Manager Installation and Administration
Guide, which can be found at:
http://www.redhat.com/docs/manuals/enterprise/
Cluster Manager is supplied with Red Hat Enterprise Linux AS. It allows you to
set up a high availability cluster containing two nodes and using shared disk
storage. Future versions will support multiple nodes.
Restriction: IBM does not currently support high availability clustering
products under Linux with xSeries servers. These applications are supported
by the vendor.
© Copyright IBM Corp. 2001 2003. All rights reserved.
139
6.1 Preparing to install Cluster Manager
High Availability clusters should ideally have no single point-of-failure. This
reliability provides data integrity and application availability in the event of a
failure. Redundant hardware with shared storage along with cluster and
application failover mechanisms mean a high availability cluster can meet the
needs of the enterprise market.
Figure 6-1 shows the setup done for our experiment.
EXP 700
Fibre Channel
FAStT700
Fibre Channel
Ethernet Heartbeat
x440
x440
Ethernet
Figure 6-1 Configuration of HA cluster
6.1.1 Quorum partitions
It is important to note that the Quorum partitions require raw devices. Cluster
Manager requires a Quorum partition and backup Quorum partition to store
cluster state and configuration information. These should ideally be created on a
140
Implementing Linux with IBM Disk Storage
logical drive on a RAID1 array. Create two partition no smaller than 10 MB but no
larger than 20 MB, there is no need for it to be larger than this. Do not create file
systems on the Quorum partitions, as these need to remain raw.
6.1.2 Software watchdog
A software watchdog is used to kill a hung node if no hardware devices such as
power switches are being used for STONITH (shoot the other node in the head).
The cluster quorum daemon (cluquorumd) will periodically reset the timer
interval. If the daemon fails to reset the timer, the failed cluster member will
reboot itself. It is necessary to create the device special for the watchdog timer.
Carry out the following on both nodes of the cluster:
cd /dev
./MAKEDEV watchdog
6.1.3 NMI watchdog
When using the software watchdog, it is recommended that you also use the NMI
watchdog timer. The NMI watchdog timer may not work on some legacy systems.
Edit /etc/grub.conf on both systems and add nmi_watchdog=1 to the end of the
relevant kernel line (Figure 6-2).
Figure 6-2 grub.conf
Chapter 6. Red Hat Cluster Manager
141
The system needs to be rebooted in order to check that the system supports the
NMI watchdog. Once the server is back online check /proc/interrupts
(Figure 6-3). If the NMI entry is not zero, then the system supports the NMI
watchdog timer. If the entry is zero then try changing the append in grub.conf to
nmi_watchdog=2 and reboot. If it is still zero then the server does not support the
NMI watchdog timer.
Figure 6-3 /proc/interrupts
6.1.4 Rawdevices
The /etc/sysconfig/rawdevices file has to be edited on both systems for use by
the primary and backup Quorum partitions. Add the following lines to the file
where X = device letter (Figure 6-4):
/dev/raw/raw1 /dev/sdX1
/dev/raw/raw2 /dev/sdX2
142
Implementing Linux with IBM Disk Storage
Figure 6-4 /etc/sysconfig/rawdevices
Save the changes and restart the rawdevices service:
service rawdevices restart
6.1.5 Log file for Cluster Messages
The /etc/sysconf.log file can be edited to log events from the cluster to a
different file away from the default log file. The cluster utilities and daemons log
their messages to a syslog tag called local4.
Edit /etc/syslog.conf on both systems and add local4.none to the following
section to avoid events being duplicated in the cluster log and the default log:
# Log anything (except mail) of level info or higher.
# Don’t log private authentication messages!
*.info;mail.none;authpriv.none;cron.none;local4.none
/var/log/messages
Add the following at the end of the file to log the cluster messages into a separate
file:
# Cluster messages on local4
local4.*
/var/log/cluster
The /etc/syslog.conf file should look similar to Figure 6-5.
Chapter 6. Red Hat Cluster Manager
143
Figure 6-5 /etc/syslog.conf
Restart the syslog service:
service syslog restart
6.2 Installing Cluster Manager
This section takes you through the installation and basic setup of Cluster
Manager.
6.2.1 Installation
It is possible that Cluster Manager was installed during the installation of Red Hat
Enterprise Linux AS. Uninstall this version on both systems and obtain the latest
version of clumanager. Type the following to uninstall:
rpm -e clumanager
Type the following to install the latest version:
rpm -ivh clumanager-<version>.rpm
144
Implementing Linux with IBM Disk Storage
6.2.2 Basic setup of Cluster Manager
We can now configure and setup the cluster software.
Type the following on only one of the cluster nodes:
/sbin/cluconfig
The execution of the script takes place triggering a number of prompts for
information such as an IP alias for the cluster, heartbeat details, Quorum
partitions, any watchdogs used. In most instances, the default answers in
[square brackets] can be accepted.
In our setup we use two Ethernet cards per system as heartbeats. A serial
interface can also be used for heartbeat if available on the system. Where
network heartbeats are used, /etc/hosts, should be updated to include the
heartbeat IP Addresses and host names.
The following is an example of the script and the answers we entered.
Example 6-1 Cluster setup
[email protected] rhcm]# /sbin/cluconfig
Red Hat Cluster Manager Configuration Utility (running on node1)
Enter cluster name [Red Hat Cluster Manager]: SG246261-Cluster
Enter IP address for cluster alias [NONE]: 192.168.1.30
-------------------------------Information for Cluster Member 0
-------------------------------Enter name of cluster member [node1]:
Looking for host node1 (may take a few seconds)...
Enter number of heartbeat channels (minimum = 1) [1]: 2
Information about Channel 0
Channel type: net or serial [net]:
Enter hostname of the cluster member on heartbeat channel 0 [node1]: hb1node1
Looking for host hb1node1 (may take a few seconds)...
Information about Channel 1
Channel type: net or serial [net]:
Enter hostname of the cluster member on heartbeat channel 1: hb2node1
Looking for host hb2node1 (may take a few seconds)...
Information about Quorum Partitions
Enter Primary Quorum Partition [/dev/raw/raw1]:
Enter Shadow Quorum Partition [/dev/raw/raw2]:
Chapter 6. Red Hat Cluster Manager
145
Information About the Power Switch That Power Cycles Member 'node1'
Choose one of the following power switches:
o NONE
o RPS10
o BAYTECH
o APCSERIAL
o APCMASTER
o WTI_NPS
o SW_WATCHDOG
Power switch [NONE]: SW_WATCHDOG
-------------------------------Information for Cluster Member 1
-------------------------------Enter name of cluster member: node2
Looking for host node2 (may take a few seconds)...
Information about Channel 0
Enter hostname of the cluster member on heartbeat channel 0 [node2]: hb1node2
Looking for host hb1node2 (may take a few seconds)...
Information about Channel 1
Enter hostname of the cluster member on heartbeat channel 1: hb2node2
Looking for host hb2node2 (may take a few seconds)...
Information about Quorum Partitions
Enter Primary Quorum Partition [/dev/raw/raw1]:
Enter Shadow Quorum Partition [/dev/raw/raw2]:
Information About the Power Switch That Power Cycles Member 'node2'
Choose one of the following power switches:
o NONE
o RPS10
o BAYTECH
o APCSERIAL
o APCMASTER
o WTI_NPS
o SW_WATCHDOG
Power switch [sw_watchdog]:
Cluster name: SG246261-Cluster
Cluster alias IP address: 192.168.1.30
-------------------Member 0 Information
-------------------Name: node1
Primary quorum partition: /dev/raw/raw1
Shadow quorum partition: /dev/raw/raw2
Heartbeat channels: 2
146
Implementing Linux with IBM Disk Storage
Channel type: net, Name: hb1node1
Channel type: net, Name: hb2node1
Power switch IP address or hostname: node1
Identifier on power controller for member node1: unused
-------------------Member 1 Information
-------------------Name: node2
Primary quorum partition: /dev/raw/raw1
Shadow quorum partition: /dev/raw/raw2
Heartbeat channels: 2
Channel type: net, Name: hb1node2
Channel type: net, Name: hb2node2
Power switch IP address or hostname: node2
Identifier on power controller for member node2: unused
-------------------------Power Switch 0 Information
-------------------------Power switch IP address or hostname: node1
Type: sw_watchdog
Login or port: unused
Password: unused
-------------------------Power Switch 1 Information
-------------------------Power switch IP address or hostname: node2
Type: sw_watchdog
Login or port: unused
Password: unused
Save the cluster member information? yes/no [yes]:
Writing to configuration file...done
Configuration information has been saved to /etc/cluster.conf.
---------------------------Setting up Quorum Partitions
---------------------------Running cludiskutil -I to initialize the quorum partitions: done
Saving configuration information to quorum partitions: done
Do you wish to enable monitoring, both locally and remotely, via the Cluster
GUI
? yes/no [yes]:
---------------------------------------------------------------Configuration on this member is complete.
To configure the next member, invoke the following command on that system:
Chapter 6. Red Hat Cluster Manager
147
# /sbin/cluconfig --init=/dev/raw/raw1
Refer to the Red Hat Cluster Manager Installation and Administration Guide
for details.
[[email protected] rhcm]#
The basic setup is complete on the first node. Type the following on the second
node:
/sbin/cluconfig --init=/dev/raw/raw1
The script looks for the required information on the Quorum partitions and pulls it
in. You are then be prompted if you wish to save the configuration. Type yes and
press Enter.
The cluster service will start automatically when the system is booted. To start
the service manually now type the following:
service cluster start
Type the following to check that the daemons are running:
service cluster status
Type the following to view the cluster node, heartbeat, and service status, where
-i is the update interval in seconds (Figure 6-6).
clustat -i 2
148
Implementing Linux with IBM Disk Storage
Figure 6-6 Cluster status
Use cluadmin to add, delete, edit, and monitor the cluster and services. See the
Red Hat Cluster Manager Installation and Administration Guide for more details.
6.2.3 Setting up an NFS clustered share
This section goes through the steps of how to set up an NFS clustered share.
This share remains available even if the owning server dies as the other node
takes ownership of the share. See the Red Hat Cluster Manager Installation and
Administration Guide for more details of setting up a service and which services
are available. Red Hat Cluster Manager does not require cluster-aware software.
A start-stop script can be created for almost any application you choose to make
highly available. An example start-stop script is provided by Red Hat.
NFS and portmap should be started on startup of the system OS. Type the
following on both nodes to enable this:
chkconfig --level 345 nfs on
chkconfig --level 345 portmap on
Start the cluadmin command line utility on one node only:
/sbin/cluadmin
Chapter 6. Red Hat Cluster Manager
149
You will drop into the cluadmin shell. Type the following to add a service:
service add
You will receive various prompts (Example 6-2) for answers from the script such
as service name, preferred node, IP Address, disk device, mount point, exports,
etc. You can assign a preferred node for the service to start on. You can also
decide if you would like to relocate the service back to the preferred node in the
event of a failover where the faulty node has come back online.
Services can be shared between the systems rather than having an
active-passive setup where the passive system just sits there waiting for the
other to fail. In an active-active setup you should ensure that there are enough
system resources for a single system to take over in the event of a failure.
The following is an example of the script and the answers we entered.
Example 6-2 Adding an NFS service
[email protected] root]# /sbin/cluadmin
Tue Mar 25 09:30:27 GMT 2003
You can obtain help by entering help and one of the following commands:
cluster
service
help
apropos
version
quit
cluadmin> service add
clear
exit
The user interface will prompt you for information about the service.
Not all information is required for all services.
Enter a question mark (?) at a prompt to obtain help.
Enter a colon (:) and a single-character command at a prompt to do
one of the following:
c - Cancel and return to the top-level cluadmin command
r - Restart to the initial prompt while keeping previous responses
p - Proceed with the next prompt
Currently defined services:
Service name: nfs_share
Preferred member [None]: node1
Relocate when the preferred member joins the cluster (yes/no/?) [no]:
User script (e.g., /usr/foo/script or None) [None]:
Status check interval [0]: 30
150
Implementing Linux with IBM Disk Storage
Do you want to add an IP address to the service (yes/no/?) [no]: yes
IP Address Information
IP address: 192.168.1.40
Netmask (e.g. 255.255.255.0 or None) [None]:
Broadcast (e.g. X.Y.Z.255 or None) [None]:
Do you want to (a)dd, (m)odify, (d)elete or (s)how an IP address, or are you
(f)inished adding IP addresses [f]:
Do you want to add a disk device to the service (yes/no/?) [no]: yes
Disk Device Information
Device special file (e.g., /dev/sdb4): /dev/sdc1
Filesystem type (e.g., ext2, or ext3): ext3
Mount point (e.g., /usr/mnt/service1) [None]: /mnt/nfs
Mount options (e.g., rw,nosuid,sync): rw,nosuid,sync
Forced unmount support (yes/no/?) [yes]:
Would you like to allow NFS access to this filesystem (yes/no/?)
[no]: yes
You will now be prompted for the NFS export configuration:
Export directory name [/mnt/nfs]:
Authorized NFS clients
Export client name [*]:
Export client options [None]: rw
Do you want to (a)dd, (m)odify, (d)elete or (s)how NFS CLIENTS, or are you
(f)inished adding CLIENTS [f]:
Do you want to (a)dd, (m)odify, (d)elete or (s)how NFS EXPORTS, or are you
(f)in
ished adding EXPORTS [f]:
Would you like to share to Windows clients (yes/no/?) [no]:
Do you want to (a)dd, (m)odify, (d)elete or (s)how DEVICES, or are you
(f)inishe
d adding DEVICES [f]:
name: nfs_share
preferred node: node1
relocate: no
user script: None
monitor interval: 30
IP address 0: 192.168.1.40
netmask 0: None
broadcast 0: None
device 0: /dev/sdc1
mount point, device 0: /mnt/nfs
mount fstype, device 0: ext3
mount options, device 0: rw,nosuid,sync
Chapter 6. Red Hat Cluster Manager
151
force unmount, device 0: yes
samba share, device 0: None
NFS export 0: /mnt/nfs
Client 0: *, rw
Add nfs_share service as shown? (yes/no/?) yes
0) node1
1) node2
c) cancel
(preferred)
Choose member to start service on: 0
Added nfs_share.
cluadmin>
Once complete, the service should automatically start. If you run clustat -i 2
you should see a screen similar to Figure 6-7 showing the service on the node
you assigned it to start on.
Figure 6-7 Clustered NFS service added
To now use this service from a client system, you should create a folder for the
NFS share to mount into. Add an entry into /etc/hosts if you added an IP
152
Implementing Linux with IBM Disk Storage
address for the service. Edit /etc/fstab to mount the NFS share on startup. The
entry should be similar to the last line of the following example (Figure 6-8).
Figure 6-8 Example of fstab
To mount the NFS share now without restarting the client system, type the
following:
mount -a
To test the cluster, cause the node which owns the server to hang, crash, or
reboot.
The software watchdog will kill the hung server and the healthy node will take
over the NFS service. Meanwhile, there should be no, or very little, interruption to
the NFS service. You could ping the IP address of the NFS service during failover
to verify this.
Another good demonstration of this failover using NFS shares is to place MP3
files on the share. Use the client system to stream the MP3 files and play the
music from another system. During failover there should be little or no
interruption to the music or the streaming.
Figure 6-9 shows the clustat output during failover.
Chapter 6. Red Hat Cluster Manager
153
Figure 6-9 Clustat during failover
You can cleanly relocate the service using the following command:
cluadmin -- service relocate nfs_share
154
Implementing Linux with IBM Disk Storage
7
Chapter 7.
Configuring ESS for Linux
This chapter describes functions of the IBM TotalStorage Enterprise Storage
Server Specialist (ESS Specialist). The information presented is an extract and
summary from the ESS user guide with focus on how to configure fixed block
storage in the ESS, as required by Linux systems.
This chapter is intended for Linux users not familiar with the function and
operations of the ESS Specialist.
Remember that Linux for zSeries requires S/390 storage (CKD) to IPL (boot)
from. For detailed information on how to configure the ESS for CKD hosts as well
as open systems environments, please refer to the redbook IBM TotalStorage
Enterprise Storage Server: Implementing the ESS in Your Environment,
SG24-5420, and to IBM TotalStorage Enterprise Storage Server Web Interface
User’s Guide, SC26-7448.
The settings and procedures described in this chapter apply to the ESS Model
800.
© Copyright IBM Corp. 2001 2003. All rights reserved.
155
7.1 IBM TotalStorage ESS Specialist
The ESS includes the ESS Specialist, which is a network enabled management
tool that allows the storage administrator to monitor and manage storage from
the IBM TotalStorage Enterprise Storage Server Master Console (ESS Master
Console), or from a remote workstation using a Web browser.
By using a secure Internet connection (LAN with a Web browser), such as
Netscape Navigator, or Microsoft Internet Explorer, your storage administrator
can coordinate the consolidation effort and easily integrate storage capacity into
the ESS.
The ESS Specialist provides you with the ability to do the following:
򐂰 Monitor error logs: If a problem occurs, a description of the problem including
the failed component, the problem severity, and who is to be automatically
notified is described.
򐂰 View the ESS status: Logical schematic of the ESS environment including the
host attached ports, controller and cache storage, device adapters, devices
and host icons may be checked.
򐂰 View and update the configuration: A color schemed view of the storage,
including the amount of space allocated and assigned to one or more hosts,
space allocated and not yet assigned, and space not allocated to logical
volumes may be viewed.
򐂰 Add host systems or delete host systems
򐂰 Configure host ports
򐂰 Add volumes, remove volumes, and reassign volumes between different
servers. Volumes can be reassigned between hosts as follows:
– Removing volumes (or unassigning volumes from hosts). Volumes can be
removed by removing all logically attached host connections to the logical
volume.
– Adding volumes. Volumes can be added from subsystem capacity that has
never been defined or after an array has been reinitialized.
– Reclaiming previously defined logical volumes
򐂰 View communication resource settings, such as TCP/IP configuration and
users
򐂰 View cluster Licensed Internal Code (LIC) levels. You can view the active
level, next level yet to be activated, and the previous level.
򐂰 Select one of the following authorization levels for each user:
– Viewer. A viewer can view the current configuration and status information.
156
Implementing Linux with IBM Disk Storage
– Operator. An operator can perform view and operation functions, such as
changing the remote service and PE password.
– Configurator. A configurator can view the current configuration and status
information and can make changes to the configuration.
– Administrator. An administrator can define new user IDs, delete old IDs,
and assign, change, or revoke passwords and levels of authorization.
򐂰 Web support for ESS Copy Services (PPRC and FlashCopy)
7.2 Using the ESS Specialist
The ESS Specialist interface (and the ESS Copy Services interface) consist of a
set of Java applets, which are programs that are dynamically loaded by the
browser, and which execute within your browser. When you request a change to
the configuration, the Java applets communicate with the microcode running on
the ESS clusters to retrieve the current configuration data, submit the requested
configuration change, and display the outcome of the request.
You must use a browser that contains the proper Java Virtual Machine (JVM)
implementation to support these applets. The browser software provided by
different companies, and even different versions of the same browser, vary
widely with respect to their JVM support. Consequently, not all browsers are
capable of supporting the ESS Specialist or ESS Copy Services.
The ESS Web interfaces support both the Netscape Navigator and the Microsoft
Internet Explorer (MSIE) versions listed in Table 7-1.
Table 7-1 Web browsers supported by ESS Web interfaces
Netscape level (See Note 1)
MSIE level (See Notes 2, 3, 4)
Netscape 4.04 with JDK 1.1 fixpack
MSIE 4.x with Microsoft Java Virtual
Machine (JVM) 4.0 or 5.0
Netscape 4.05 with JDK 1.1 fixpack
MSIE 5.x with Microsoft JVM 4.0 or 5.0
Netscape 4.06 (no fixpack required)
Netscape 4.5x (no fixpack required)
Netscape 4.7x (no fixpack required)
Chapter 7. Configuring ESS for Linux
157
Netscape level (See Note 1)
MSIE level (See Notes 2, 3, 4)
Notes:
1. The ESS Web interfaces do not support Netscape above version 4.7.x
2. If your ESS is running with ESS LIC earlier than level 1.3.0 or SC01206, the
performance of the ESS Web interfaces on MSIE 5.0 with JVM 5.0 is slower than with
Netscape. It is recommended that you use Netscape as the browser or move to LIC level
1.3.0 or higher
3. MSIE 5.0 with JVM 4.0 is supported with all levels of ESS code. However, it is not
recommend that you change JVM 5.x to JVM 4.0 on the ESSNet machine in order to
improve performance. It is not trivial to change the JVM to a lower level.
4. The ESS Master Console running Linux does not support the MSIE browser.
Using the ESS Specialist client from Linux
In order to start and successfully work with the ESS Specialist client from a Linux
workstation, you have to install a supported version of the Netscape Web
browser. We used Netscape Version 4.79, which is available from:
http://wp.netscape.com/download/archive/client_archive47x.html
Note: In preparation of this redbook, we have tried several different
combinations of current Web browsers, plug-ins, and Java virtual machines
(e.g. Mozilla, Konqueror, Netscape, IBM, and Sun JDKs). We were not able to
get any combination of the more recent software to work properly. Our
recommendation is to just use the older Netscape version that includes a
supported JVM, which worked without any problems.
First logon
When you open a Web browser and log on to your ESS Specialist for the first
time, you will get prompted to accept the site certificate from the ESS. After you
accepted the certificate you get a welcome screen, as shown in Figure 7-1, from
where you can start the actual ESS Specialist Java client application.
158
Implementing Linux with IBM Disk Storage
Figure 7-1 ESS welcome screen
When you click the ESS Specialist menu link, you are prompted for a user name
and a password.1 Depending on the performance of your client system, it might
take awhile for the ESS Specialist applet to execute and display the main screen,
as shown in Figure 7-2.
1 To be able to make changes to the ESS configuration, you need a user of the class Configurator on the ESS
master console, which you have to obtain from the administrator or which you create during the ESS setup and
installation.
Chapter 7. Configuring ESS for Linux
159
Figure 7-2 ESS Specialist main screen
The navigation frame of ESS Specialist contains six buttons that provide access
to the major categories of information and tasks that you can perform. The
buttons are:
Status
The Status button accesses the operational status of your
ESS, in graphical and log views.
Problem Notification The Problem Notification button accesses panels that
enable you to set up ways to have the ESS automatically
alert you and those you designate of operational
problems.
Communications
The Communications button accesses panels that enable
you to set up remote technical support.
Storage allocation
The Storage Allocation button accesses panels that
enable you to set up and view the storage segmentation in
your ESS.
Users
The Users button accesses panels that enable you to give
ESS users various levels of access to the ESS.
Licensed Internal Code The Licensed Internal Code button accesses panels that
display the licensed internal code (LIC) in your ESS, the
160
Implementing Linux with IBM Disk Storage
cluster on which the LIC is installed, and the licensed
feature codes that you have purchased.
7.2.1 ESS attachment configurations
ESS Specialist displays the Storage Allocation panel after you click Storage
Allocation from the navigation frame, which is shown in Figure 7-3. The Storage
Allocation panel contains a logical view of the components of the ESS storage
facility. It shows the server’s logical configuration of host systems, host ports,
storage adapters, arrays, and volume assignments. The Storage Allocation panel
provides access to a graphical view and a tabular view of ESS components. The
graphical view depicts the hardware components of the ESS with interconnecting
lines to indicate a particular assignment or logical connection. The tabular view
presents the same information in table format.
Figure 7-3 Storage Allocation panel
The screen contains the following elements:
򐂰 The icons in the top row of the Storage Allocation-Graphical View panel
represent the host systems that are attached to the ESS. Table 7-2 shows the
types of host icons.
Chapter 7. Configuring ESS for Linux
161
Table 7-2 Host icons and their descriptions
Host icon
Description
A rectangular icon divided by two vertical lines
represents a S/390 or zSeries host. It can also represent
a Linux partition on that host. See Note 1.
An icon of a display screen next to a box represents a
AS/400® or iSeries host, except for Linux partitions
within the host (UNIX icons). See Note 2.
An icon of a display screen on top of a thin box
represents a Microsoft Windows, Linux (x86), or Novell
host. See Note 2.
An icon of a display screen and a keyboard represents a
UNIX host, such as Sun Solaris, IBM RS/6000, HP
9000, and Linux partitions in iSeries and pSeries hosts.
See Note 2.
A low, wide rectangular icon represents a Cisco iSCSI
gateway (the SN 5420). See Note 2.
Notes:
1. When the icon has a FiconNet or EsconNet label, it is a pseudo-host representing all
attached S/390 and zSeries hosts that are not yet defined to the ESS, attached through
FICON or ESCON protocol, respectively.
2. When the icon has an Anonymous label, it represents open systems connected
through the Fibre Channel protocol.
When one of the four types of icons represents a host not explicitly defined to
the ESS, but which has access to one or more volumes configured on the
ESS, the icon is called a “pseudo-host.” ESS Specialist adds a pseudo-host
whenever you configure the LUN access mode for a Fibre Channel port to
access-any mode.
162
Implementing Linux with IBM Disk Storage
򐂰 The icons in the second row represent the host adapter cards installed in the
ESS. The lines between them indicate the four host adapter bays in the ESS.
Each of the bays has a capacity of four host adapter cards. Depending on
how many host adapters you purchased, the host bays may not be fully
populated. Each host adapter card can be one of three types shown in
Table 7-3, and cards of all types can be mixed in any order in any of the bays.
Table 7-3 Host adapter types
Adapter icon
Description
ESCON adapter card (2 ports per card).
Fibre Channel adapter cards (1 port per adapter, using
FICON or FCP protocols)
SCSI adapter card (2 ports per adapter).
Note that the ESCON and SCSI adapters contain two ports per adapter. The icons for
these adapters are divided in half so that clicking on the left side of the adapter selects
port A and clicking on the right side of the adapter selects port B. Fibre-channel adapters
have only one port.
򐂰 The graphical area below the row of host adapter icons displays the
relationship of storage areas in the two ESS clusters. It is split into two
columns each representing a cluster, and each row in each column has an
icon at the outside of the row that represents a device adapter card. The lines
between the device adapter icons represent the connections between the disk
groups supported by the device adapter cards.
There are normally four pairs of device adapter cards installed in an ESS, four
cards in each cluster. The ESS uses SSA device adapters, each of which
supports two SSA loops.
The disk groups are the rectangles on the SSA loops. There are normally 16
disk groups installed in an ESS, and two disk groups on each SSA loop. The
minimum ESS configuration is four disk groups, the maximum configuration is
Chapter 7. Configuring ESS for Linux
163
48 disk groups (that is, six disk groups per SSA loop). The disk groups are
usually split evenly among the two clusters.
The individual disk groups are drawn as small rectangles on the SSA loops
and between the SSA adapter pairs of the clusters. When you select a host
and associated port, each associated disk group is filled with a color-coded
bar graph representing the space utilization within that group. The color
coding for the bar graph is described in the legend box on the right side of the
panel, has the following meaning:
Violet
Host Storage: Violet represents the storage space
occupied by volumes that are assigned to the currently
selected host system.
Red
Assigned: Red represents the storage space
occupied by volumes that are assigned to one or more
host systems, but not to the currently selected host
system.
Yellow
Unassigned: Yellow represents the storage space
occupied by volumes that are allocated, but are not
currently assigned to any host systems. This is usually
a temporary condition, since unassigned space is
essentially wasted space.
Green
Not allocated: Green represents free storage space,
that is, storage where no volumes have yet been
allocated.
򐂰 At the bottom of the Storage Allocation-Graphical View panel are two buttons:
– S/390 Storage - This button accesses the S/390 Storage panel, which you
use for configuring storage for data in count-key-data (CKD) format
attached to S/390 and zSeries (mainframe) host systems.
– Open Systems Storage - This button accesses the Open Systems
Storage panel, which you use for configuring storage for data in fixed-block
(FB) format attached to UNIX-based, Intel-based and AS/400 host
systems.
For this book we only use the Open Systems Storage function, since we
concentrate on attachment of Fibre Channel storage for all the Linux hardware
platforms. Even though the Linux for zSeries uses S/390 storage (CKD) as well
to IPL (boot) from, the necessary steps to configure fixed block storage on the
ESS are the same as for the other platforms. The explanation of how to configure
S/390 storage is beyond the scope of this book. Please refer to the IBM
TotalStorage Enterprise Storage Server Web Interface User’s Guide, SC26-7448,
for information on how to configure S/390 storage.
164
Implementing Linux with IBM Disk Storage
7.2.2 Storage allocation procedure
Use the Open System Storage panel to define and modify the definitions for
Fibre Channel and SCSI-attached Linux host systems, and to assign physical
storage units for them.
1. If you are not already on the Storage Allocation panel, click the
corresponding button in the navigation frame of ESS Specialist. The Storage
Allocation-Graphical View panel opens.
2. At the bottom of the panel, click Open System Storage. The Open System
Storage panel opens as shown in Figure 7-4.
Figure 7-4 Open System Storage panel
3. Inspect the list of defined open-systems hosts and storage allocations in order
to determine the remaining storage configuration tasks that you need to
perform.
4. You generally perform configuration actions in the following order by clicking
the associated buttons at the bottom of the Open System Storage panel:
a. Click Modify Host Systems to open the Modify Host Systems panel and
to define new hosts to ESS Specialist.
b. Click Configure Host Adapter Ports to open the Configure Host Adapter
Ports panel and identify to ESS Specialist the characteristics of the new
Chapter 7. Configuring ESS for Linux
165
host adapter ports. For SCSI ports, you also use this panel to associate a
port with a host system.
c. Click Configure Disk Groups to open the Fixed Block Storage panel and
select a storage type and track format for a selected disk group.
d. Click Add Volumes to open the Add Volumes (1 of 2) panel and define
numbers and sizes of volumes for selected arrays.
e. Click Modify Volume Assignments to open the Modify Volume
Assignments panel and modify volume assignments, such as creating
shared access to a volume, or removing an assignment.
7.2.3 Defining Linux hosts
On the Open System Storage panel click Modify Host Systems. The Modify
Host Systems panel opens as shown in Figure 7-5.
Figure 7-5 Modify Host Systems panel
Enter the characteristics of the new host in the Host Attributes sub-panel:
1. In the Nickname field, type a name that uniquely identifies the associated host
or host attachment within the ESS. The host nickname must not exceed 29
characters. ESS Specialist ignores the leading and trailing blanks that you
entered.
166
Implementing Linux with IBM Disk Storage
2. Select a host type from the list in the Host Type field. There are three types of
Linux hosts that you can choose from:
– Linux (x86): PC servers with the Linux operating system with v2.4 kernel.
Although you can configure 256 LUNs on the ESS, a Linux host can only
support 128 LUNs.
– Linux (iSeries/pSeries)
– Linux (zSeries)
3. From the Host Attachment list you would usually select the type of interface
used to connect the host to the ESS, but since the ESS supports Linux hosts
only through the FCP protocol, Fibre-channel attached is the only option
available in the Host Attachment list.
4. In the Hostname/IP Address field, if the host is connected to the IP network,
you have the option of entering the host name or the dotted decimal IP
address. You should enter information in this field if you are using the IBM
TotalStorage Expert software package.
5. The Worldwide Port Name field is enabled if you select Fibre-channel
attached from the Host Attachment list. If the host to be identified is
Fibre-Channel attached, select the appropriate WWPN from the list in the
Worldwide Port Name field, or type the WWPN in manually. The WWPN list
contains the WWPNs of all host Fibre Channel adapters that are currently
connected (logged in) but not yet defined to the ESS. The ESS discovers the
WWPN from your host Fibre Channel adapter when you connect your host
system to the ESS. If the connection is made through a single ESS Fibre
Channel port, that port identifier is listed in parentheses following the WWPN.
If the connection is made through multiple ESS Fibre Channel ports, the ports
are not indicated in the list.
6. Scroll down in the Host Attributes sub-panel to display Fibre Channel Ports
list in this field. The highlighted entries indicate the ports that this particular
host system can use to access its assigned volumes in the ESS. If the first
entry, All installed ports, is highlighted (which is the default) then this host
system can use all Fibre Channel ports in the ESS to access its assigned
volumes. Alternatively, to limit the ports through which this host can access
the ESS, select one or more individual ports in the list.
After you have filled in the complete information click Add. The Host Systems
List, which is on the right side of the Modify Host Systems panel, displays the
data that you entered about the new host, along with the attributes of all currently
defined host systems.
Next click Perform Configuration Update at the bottom of the panel to apply the
configuration changes to the ESS. A progress bar indicates the progress of the
configuration action as the ESS processes it. Alternatively, click Cancel
Chapter 7. Configuring ESS for Linux
167
Configuration Update to cancel the information that you entered. Clicking either
button returns you to the Open System Storage panel, which displays the
characteristics as you entered them of the new host or hosts that you defined.
7.2.4 Configure the host adapters
At the bottom of the Open System Storage panel, click Configure Host Adapter
Ports. The Configure Host Adapter Ports panel opens. Click an icon for a Fibre
Channel adapter (the icon has one port; see Table 7-3). Alternatively, select the
adapter from the Host Adapter Ports list that is below the icon row. Figure 7-6
shows an example of the fields that are enabled on the Configure Host Adapter
Ports panel when you click a Fibre Channel adapter.
Figure 7-6 Configure Host Adapter Ports panel
1. The Fibre Channel Access Mode field in the Storage Server Attributes box
shows the current Fibre Channel access mode for the ESS, either:
– Access-any
– Access-restricted
Note: If you want to change the Fibre Channel access mode, ask your
Service Support Representative to perform this change during a service
action or during installation of a new host adapter.
168
Implementing Linux with IBM Disk Storage
2. Select one of the following attributes from the Fibre-Channel Topology list in
the FC Port Attributes box:
– Point to Point (Switched Fabric). For unconfigured ports, point-to-point
topology is the only choice if you access the Configure Host Adapter Ports
panel from the S/390 Storage panel.
– Arbitrated Loop (Direct Connect)
Note: If you want to connect a Linux for zSeries system, only a switched
environment is supported. Thus, you have to select Point to Point
(Switched Fabric).
If the port is already configured and you want to change its settings, you must
first remove the configuration for the port by selecting Undefined from the list
and then clicking Perform Configuration Update.
You can click Reset Selected Port at any time to cancel any pending
configuration changes made to the currently selected port. To configure another
Fibre Channel port, select the port from the Host Adapter Port list or from the
port graphic, then repeat the configuration steps above. When all entries are
complete for the Fibre Channel port, click Perform Configuration Update to
apply, or Cancel Configuration Update to cancel the configuration step for all
modified ports.
7.2.5 Configure RAID arrays
Before you can allocate storage space as logical volumes for use by specific host
systems, you must define disk groups on the Fixed Block Storage panel. Defining
a disk group means that you make two formatting selections for the selected disk
group: storage type and track format. If you are allocating storage for an
open-systems or Linux host (in other words, for a system other than one using
the count-key-data (CKD) data format), use the Fixed Block Storage panel.
At the bottom of the Open System Storage panel, click Configure Disk Groups.
The Fixed Block Storage panel opens as shown in Figure 7-7.
Chapter 7. Configuring ESS for Linux
169
Figure 7-7 Fixed Block Storage panel
1. From the Available Storage table, select a disk group. If you select a disk
group that has a value of Undefined in the Storage Type column, you can be
sure that the disk group is not already being used. If you have a system that
contains RAID-10 disk groups, you can gain capacity by converting them to
RAID 5, or if you have a system containing RAID-5 disk groups, you can
increase performance by converting them to RAID 10.
2. Select one of the following storage types from the Storage Type list:
– RAID-5 Array
– RAID-10 Array
If you select RAID-10 Array or RAID-5 Array, the value in the Track Format
field is automatically set to Fixed Block.
Repeat the sequence of steps on the Fixed Block Storage panel as needed to
define or undefined RAID disk groups. Defined appears in the Modification
column in the table to indicate which entries will be affected when you commit
your changes. When you complete your desired changes, click Perform
Configuration Update to apply the changes, or click Cancel Configuration
Update to cancel the changes. In either case, the Open System Storage panel
replaces the Fixed Block Storage panel in the browser window. Alternatively, you
can click one of the buttons on the navigation pane to open another panel without
committing the changes.
170
Implementing Linux with IBM Disk Storage
7.2.6 Allocating storage for Linux
Click Add Volumes at the bottom of the Open System Storage to panel to create
fixed-block volumes on the ESS for use by your Linux hosts. This section divides
the task of assigning open-systems volumes into two subsections, one for each
panel. First the Add Volume (1 of 2) panel opens as shown in Figure 7-8.
Figure 7-8 Add Volumes panel (1 of 2)
1. In the top row of icons, click the host icon that represents the host system to
which you want to assign the new volumes. The host adapter ports that are
connected to this host system become highlighted. If the Linux host has any
volumes already assigned, the disk groups that contain those volumes also
become highlighted. You can select only one host system at a time for an
add-volumes sequence. To refresh the view before selecting a different host
system icon, click Clear View at the bottom of the panel.
The host system that you select is highlighted in yellow and appears with a
black background to indicate selection. The name of the host system appears
in the Information box at the bottom right of the panel. A thin black line is
drawn from this host system to one or more host ports in the second row that
is attached to this system.
Chapter 7. Configuring ESS for Linux
171
Note: Before adding a LUN (a synonym for volume when speaking of
open-systems volumes), or in any way modifying the LUN assignments for
a host system port, ensure that the affected LUNs are offline or not in use
by any attached host system. After adding a LUN, it might be necessary to
restart the attached host systems in order to establish connectivity to the
added LUN.
2. Select a highlighted host adapter port. The background of the port you
select changes to black. Volumes are assigned to the selected host and are
visible on all the ESS Fibre Channel ports that were defined for that host. You
need to select the highlighted ports associated with the attached host to limit
the host’s access to those specific ports. A Fibre Channel attached
open-systems host shows a connection to all of its configured Fibre Channel
adapters.
After you select a port, all fixed-block disk groups are selected by default and are
highlighted. This provides a quick overview as to where the new volumes can be
created. Optionally, select one or more fixed-block groups. The Information box
on the right side of the panel summarizes the list of selected disk groups. If you
select any groups, they will be the only ones used for the add-volume process.
Alternatively, if you do not select any groups, the ESS will use all available disk
groups for the add-volume process.
To proceed click Next and the Add Volumes panel (2 of 2) opens as shown in
Figure 7-9:
1. In the Available Free Space table (at the top of the Add Volumes (2 of 2)
panel), select the table row that corresponds to the type of storage that you
want to allocate. You must select a type for which the available capacity is
greater than 0.
The information in the table is based on the capacities of the storage areas
selected on the first Add Volumes panel. Each row in the table shows the
available capacity for a specific storage type. Whether RAID 5 or RAID 10 is
displayed in the table depends first on the storage types that are available for
your ESS, and on what storage type you selected when you configured your
fixed-block disk groups (see 7.2.5, “Configure RAID arrays” on page 169).
The Available Capacity column indicates the total available space in the
selected storage areas, and the Maximum Volume Size column displays the
largest contiguous free space in the selected storage areas, by storage type.
If you selected particular disk groups on the Add Volumes (1 of 2) panel, the
defined storage types available on those selected disk groups are reflected in
the Available Free Space table.
172
Implementing Linux with IBM Disk Storage
Figure 7-9 Add Volumes panel (2 of 2)
2. In the Volume Attributes section:
a. Select a Volume Size from the so named list. The Available Free Space
table at the top of the panel displays the maximum volume size that can be
allocated in the selected storage areas. Size is specified in gigabytes.
b. In the Number of Volumes field, type the number of volumes that you want
to add.
c. If you have not already used the Available Free Space table to select the
type of disk group in which to place the new volumes; you can scroll down
in the Volume Attributes section and select the type from the Storage Type
list.
3. When you are creating multiple volumes and you have selected multiple disk
groups, you can choose how the volumes should be placed. Near the bottom
of the panel, select one of the following Volume Placement options:
– Place volumes sequentially, starting in first selected storage area (the
default): This choice creates volumes starting in the first storage area, and
continues allocating them there until the capacity in that storage area is
depleted. At this point, allocation continues in the second storage area,
and so on.
Chapter 7. Configuring ESS for Linux
173
– Spread volumes across all selected storage areas: This choice
allocates the first volume in the first storage area, the second volume in
the second area, the third volume in the third area, and so on. Wrapping
occurs if necessary, so that after allocating a volume in the last area, the
next volume is allocated in the first area. If a particular volume is too large
for a particular area, that area is skipped.
4. To update the New Volumes table click Add. The Available Capacity and
Maximum Volume Size columns in the Available Free Space table at the top of
the panel are also updated. The New Volumes table lists all defined volumes
that are pending creation. The last row in the table indicates the total amount
of space, in gigabytes, that has been defined. To remove volumes from the list
to be created, select one or more volumes from the New Volumes table, then
click Remove. The last row in the table and the Available Capacity and
Maximum Volume Size columns at the top of the panel are updated
appropriately.
Note: It is important that you carefully plan, create, and assign your LUNs.
Once you perform the configuration update, the only way to physically
delete the LUN is to reformat the whole array.
When you are finished adding or deleting volumes, click Perform Configuration
Update to apply the changes to the fixed-block volume configuration or Cancel
Configuration Update to cancel the changes. A progress bar indicates the
progress of the configuration action as the ESS processes it. After the
configuration changes are completed or cancelled, you are returned to the Open
System Storage panel.
Attention: When volumes are created and assigned to a port, the ESS
initiates a formatting process that can exceed one hour for one or more
volumes. A volume is not available to the host system while the ESS is
formatting it. Do not attempt to use the newly created volumes until the ESS
reports that it has completed the configuration. You can click Refresh Status
on the Modify Volume Assignments panel to check the progress of the
formatting process on a volume. If the formatting process is complete, no
progress indication appears. Note also that the refresh function requires some
time to run.
Modify the volume assignments
After using the Add Volumes panels to create a volume and assign it to a Linux
host, you can use the Modify Volume Assignments panel to modify the
assignment of the volume, including removing it from the host or assigning it to
an additional host. Click Modify Volume Assignments in the Open System
Storage panel for the new panel shown in Figure 7-10.
174
Implementing Linux with IBM Disk Storage
Figure 7-10 Modify volume assignments
Click one or more rows in the table to select the volumes you want to assign.
When you select a row, the two check boxes in the Action box below the table are
enabled:
򐂰 In the Action box, select Assign selected volume(s) to target hosts.
The optional Use same ID/LUN in source and target action becomes
enabled. Select this if you want to keep the same volume ID and LUN for
source and target hosts. The display changes to show in the Target Host box,
hosts that are compatible for sharing the selected volume.
– The Target Host box is populated with all the hosts that are connected by
SCSI or Fibre Channel adapters to the ESS. Use the scroll bar to see the
complete list. Select the host to which you want to assign the volume.
Click Perform Configuration Update to apply the modifications; or click Cancel
Configuration Update to cancel the modifications to the volume assignments.
Perform the following steps to remove a volume assignment from a host:
򐂰 Open the Modify Volume Assignments panel, as described previously.
򐂰 In the Volume Assignments table, select the volume-host association to be
removed. If you want to remove more than one association at one time, you
can use the table to do so by selecting more than one row.
Chapter 7. Configuring ESS for Linux
175
– Select Unassign selected volume(s) from target host.
– Select the host in the Target Host box. Depending on the characteristics of
the associations between the volume and its associated hosts, the Host
Nicknames field for a particular row might contain more than one host. In
that case, the Target Host box would display all of the host names. Also,
the Target Host box would display more than one host name if you
selected more than one row in the Volume Assignments table. In the case
where you have more than one host in the Target Host list, select each
host that you want to remove from association with the selected volume.
Click Perform Configuration Update to apply the modifications, or click Cancel
Configuration Update to cancel the modifications to the volume assignments.
After the changes are made, ESS Specialist leaves you on the panel and
refreshes the view, so it is easy to make changes incrementally.
Note: Removing a volume from a target host does not affect the volume
definition, nor does it affect any of the data on the volume. The volume is not
deleted; only the assignment to the target host is deleted.
176
Implementing Linux with IBM Disk Storage
8
Chapter 8.
BladeCenter SAN Utility
This chapter provides the details for installing the BladeCenter SAN Utility.
© Copyright IBM Corp. 2001 2003. All rights reserved.
177
8.1 Installing the BladeCenter SAN Utility
The BladeCenter SAN Utility application is used to access and configure the
BladeCenter Fibre Channel switch modules. The SAN Utility can be installed on
a BladeCenter HS20 blade server or an external network management
workstation configured with a supported version of Linux.
Before you configure your switch module, be sure that the management modules
are properly configured. In order to access your switch module from an external
environment, you may need to enable certain features, such as external ports
and external management over all ports. See the applicable BladeCenter Unit
Installation and User’s Guide publications on the IBM eServer BladeCenter
Documentation CD or at:
http://www.ibm.com/pc/support
Obtain the latest version of the BladeCenter SAN Utility. Install the SAN Utility
using the following command:
sh Linux_<version>.bin
You should see a screen similar to Figure 8-1, read the details and click Next.
Figure 8-1 SAN Utility Installation Introduction screen
Choose the location for the installation. Stay with the default location or enter
another location and click Next (Figure 8-2).
178
Implementing Linux with IBM Disk Storage
Figure 8-2 SAN Utility Installation Folder screen
You then get the following options regarding the creation of links (Figure 8-3):
򐂰 In your home folder
򐂰 Other: /opt is default
򐂰 Do not create links
Figure 8-3 SAN Utility Create Links screen
Chapter 8. BladeCenter SAN Utility
179
Make your choice and click Next.
The next screen requires a location for log files. Stay with the default or enter a
different location and click Next (Figure 8-4).
Figure 8-4 SAN Utility Log Location screen
Next, select the browser you would use for any online help and click Next
(Figure 8-5).
180
Implementing Linux with IBM Disk Storage
Figure 8-5 SAN Utility Browser Select screen
The next screen gives a pre-installation summary. This is the last chance to go
back and change any settings before the software starts to copy files to disk.
Make any changes if required and click Install (Figure 8-6).
Figure 8-6 SAN Utility Pre-installation Summary screen
Chapter 8. BladeCenter SAN Utility
181
Installation begins (Figure 8-7).
Figure 8-7 SAN Utility Installation screen
Once installation is complete click Done (Figure 8-8).
Figure 8-8 SAN Utility Install Complete screen
182
Implementing Linux with IBM Disk Storage
To run the SAN Utility type the following where X is the location where you
installed the software:
/X/runBladeCenterSANUtility
You should see a screen similar to Figure 8-9.
Figure 8-9 SAN Utility screen
Click Add and enter the required details in the fields of the Add New Fabric
screen (Figure 8-10).
The default user ID is USERID, the default password is PASSW0RD (the sixth
character is a zero, not the letter O). The user ID and password are case
sensitive.
Chapter 8. BladeCenter SAN Utility
183
Figure 8-10 Add New Fabric screen
Highlighting the name of the fabric on the left hand column shows the Topology
view (Figure 8-11).
Figure 8-11 SAN Utility Topology View screen
Highlighting the Fibre Switch Module on the left hand side shows the Faceplate
view (Figure 8-12).
184
Implementing Linux with IBM Disk Storage
Figure 8-12 SAN Utility Faceplate view
Attention: If using two Fibre Switch Modules it is necessary to delete the
factory default zoning in order for the switches to communicate with the
storage.
The SAN Utility software can be used to change such things as port settings,
zoning, alarms.
Once the Fibre Switch Modules are configured, you are ready to start preparing
the storage.
Chapter 8. BladeCenter SAN Utility
185
186
Implementing Linux with IBM Disk Storage
9
Chapter 9.
FAStT Storage Manager
This chapter contains a description of the FAStT Storage Manager. FAStT
software is used to define storage in the FAStT and to manage its availability to
fibre-attached hosts.
This is not an exhaustive coverage as there are plenty of redbooks on the
subject. While our emphasis here is on using the Storage Manager in the Linux
environment, we also try to keep the content of this chapter as host-neutral as
possible, with the host specific details being reserved to their respective
chapters. We cover these topics:
򐂰
򐂰
򐂰
򐂰
򐂰
Storage Manager concepts
FAStT Storage Manager client: Getting started
Updating the FAStT firmware
Setting up arrays, LUNs, and Storage Partitioning
Configuring the LUN attachment
© Copyright IBM Corp. 2001 2003. All rights reserved.
187
9.1 FAStT Storage Manager
The Fibre Array Storage Technology1 (FAStT) requires a set of software and
firmware to integrate with a host. The software includes a management
component that provides a user interface to administer the storage system and
additional programs such as an agent and utilities that may be required. Also
included in the software may be a host-level driver that enables I/O path transfer
within the host in the event of path failure in a multipath attachment. Additional
software may also be needed to implement an appropriate environment on the
management station, the host using storage provided by the FAStT, or both.
In the general case, The FAStT Storage Manager Software includes the Storage
Management Client, the Storage Management Agent, the Storage Management
Utilities, and the Redundant Disk Array Controller (RDAC). Not all installations
require all components of the FAStT Storage Manager suite.
9.1.1 Storage Manager concepts
Management of the FAStT storage subsystem requires an instance of the FAStT
Storage Manager client on a management station. Because the FAStT Storage
Manager is a Java-based application, any suitable management station2 may be
used provided it can communicate with the FAStT storage subsystem in one of
two ways:
in-band
The Storage Manager client communicates with an agent process over either a
physical or loopback network connection. The agent process runs on one or
more of the systems connected via Fibre Channel to the FAStT storage
subsystem. The agent process communicates with the storage controllers over
the Fibre Channel by means of a special interface in the controller known as an
Access LUN. This method is also known as indirect control because of the
intermediary agent process. No connection is required to the network ports on
the FAStT controllers.
out-of-band
The Storage Manager client communicates over physical network connections to
the FAStT controllers. The Access LUN is not used. This method is also known
as direct control since no additional agent process is required, and no control
communication takes place over the Fibre Channel. The default settings are
192.168.128.101 and 192.168.128.102. Linux communicates with the FAStT
out-of-band, so if these IPs do not fit in with your network, then these will need to
be changed.
1
2
188
See also the redbook, Fibre Array Storage Technology - A FAStT Introduction, SG24-6246
Supported systems include Linux, AIX, Windows, Solaris, Novell NetWare and HP-UX
Implementing Linux with IBM Disk Storage
Setting the IP and mask on a FAStT Controller
Connect a serial null modem cable between a serial port on the server (otherwise
use a client system) and Port A on the FAStT. This allows the configuration of
Controller A.
Use a terminal program such as minicom under Linux or HyperTerminal under
Microsoft Windows.
Use the following settings:
򐂰
򐂰
򐂰
򐂰
򐂰
Serial device (devv/ttyS1)
Speed (9600)
Data/parity/stop bits (8N1)
Hardware Flow Control (Yes)
Software Flow Control (No)
If using minicom, press Ctrl-A F to send break. If using HyperTerminal, press
Ctrl-Break every 5 seconds until the ASCII characters become human readable.
Press Escape to enter the FAStT shell.
The shell prompts you for a password (Figure 9-1). The default is infiniti
Figure 9-1 Password prompt for shell
Type netCfgShow to view the current settings (Figure 9-2).
Chapter 9. FAStT Storage Manager
189
Figure 9-2 netCfgShow screen
Type netCfgSet to change the settings. Change only My IP Address and Subnet
Mask. Press Enter to bypass the other settings.
Important: Do not change any other settings unless advised to do so by IBM
Level 2 Technical Support. Any other alterations will place the FAStT in an
unsupported configuration.
Once complete type sysReboot to restart the controller blade.
Wait for the reboot to complete and repeat this whole process again for Controller
B by connecting the null modem cable to Port B, and reconnecting the terminal
software.
Is there a Storage Manager agent for Linux?
The FAStT Storage Manager software does not include an agent process for all
operating systems. For example, there is no agent that runs under Linux. A Linux
system can run the Storage Manager client and manage a FAStT system using
in-band control if there is a host that has a Fibre Channel connection to the
FAStT, and runs the Storage Manager agent. In this case, the Storage Manager
client on the Linux system would communicate with the agent on the non-Linux
system. However, if the only hosts that are attached to the FAStT storage
subsystem are Linux hosts, or if there are no attached hosts for which agent
190
Implementing Linux with IBM Disk Storage
software is available that can be considered as management agent candidates,
then out-of-band management must be used.
For this discussion we assume that the environment is purely Linux, and that
out-of-band control will be used to manage the FAStT.
Is there an RDAC driver for Linux?
At the time of this writing, the Redundant Disk Array Controller (RDAC) is not
available for Linux. This means that Linux hosts using storage on the FAStT have
volume transfers within the FAStT managed by the FAStT itself. Multipathing is
done using the appropriate driver for the Host Bus Adapter (HBA) and additional
software called FAStT MSJ, that is used to manage the paths with the Host Bus
Adapters. FAStT MSJ is described in Chapter 10., “FAStT MSJ (Management
Suite for Java)” on page 211.
9.1.2 FAStT Storage Manager client: Getting started
The FAStT Storage Manager client is the graphical user interface (GUI) that
controls one or more FAStT storage subsystems. Because this is a Java
application, a Java runtime environment (provided on the FAStT Storage
Manager CD) must be installed as well.
Linux Install of the FAStT Storage Manager client
Linux distributions of interest to us in this redbook come equipped with the rpm
package manager. This tool maintains a dependency list such that if one
package requires another, the user will be notified. When installing the Storage
Manager client under Linux, you should install the Java runtime environment
before the client software. In our implementation, the FAStT Storage Manager
client resides on a Linux management station that can communicate with the
FAStT controllers via the network. Such a management station may also be a
host that is attached to the FAStT storage via Fibre Channel, but this is not
required. To be clear, in the Linux environment, there is no Storage Manager
agent available.
So, even if the Storage Manager client is installed on a host that is attached via
Fibre Channel to the storage, the management still takes place out-of-band and
requires network connectivity to the FAStT controllers. Both the client and the
runtime environment are provided on the FAStT CD, but the latest versions (as
well as the latest firmware for the FAStT controllers and expansion modules;
please see Appendix 9.1.3, “Updating the FAStT firmware” on page 195 for
detailed instructions on updating firmware) can be obtained from the following
URL:
http://www.storage.ibm.com
Chapter 9. FAStT Storage Manager
191
From there, navigate to Disk Storage Systems,3 then FAStT Storage Server, and
then select your FAStT model; use the pull-down menu to select downloads,
then select FAStT Storage Manager Downloads as appropriate for your
version.
Read and accept the terms and conditions, and you now see a selection of both
software and documentation. Please obtain and read the documentation entitled
“IBM TotalStorage FAStT Storage Manager Version X.Y Installation and Support
Guide for Linux” (where X.Y corresponds to your version number). Click your
browser’s “back” button and then select LINUX operating system; this action
takes you to the download page for the code and the README file. Please look
at the README file before installing, as it contains the latest information for the
code release.
The package is rather large (over 30 MB) and requires another 70 MB to install,
including the Java Runtime Environment. Make sure you allow enough room for
unpacking the code, storing the rpm files, and the installed code. Also, please
note that the Storage Manager client GUI requires the X window system be
running on the management workstation.
You need to have root privilege to install the client. Site administrators may make
their own decisions regarding access control to the Storage Manager client. The
FAStT itself can be protected by password from changes to the storage
configuration independently of access to the Storage Manager client.
First, uninstall any previous version of Storage Manager before installing the new
version. To check for prior installations, query the packages that are installed and
select only those containing the letters “SM” with the following command:4
rpm --query --all | grep SM
Use the output from this command to remove any prior version of the Storage
Manager software by package name. For example:
rpm --uninstall SMruntime-XX.YY.ZZ-1
When you have obtained the latest version of Storage Manager, extract the
contents of the zipped tar file:
tar zxvf *.tgz
The required rpm files for installation are extracted into ./Linux/SMXclientcode
where X=Storage Manager major version.
The Storage Manager components should be installed in the following order:
3
Navigation details may change from time to time, these instructions were correct as of April 2003
We have used the long option names for clarity. Experienced users will know of shorthand versions
of the options. Please see the rpm documentation (man rpm) for more information
4
192
Implementing Linux with IBM Disk Storage
rpm --install --verify SMruntime-<version>.rpm
rpm --install --verify SMclient-LINUX-<version>.rpm
rpm --install --verify SMutil-<version>.rpm
The Storage Manager software is installed in /opt/IBM_FAStT, with the client and
command line interface, SMcli, installed in the /opt/IBM_FAStT/client directory.
Launching Storage Manager and connecting to the FAStT
To launch Storage Manager, open a terminal window and type SMclient 5
Alternatively, you can launch Storage Manager from the graphical interface
menus:
Click the Gnome or KDE icon -> Programs -> Utilities -> IBM FAStT Storage
Manager.
The screen shown in Figure 9-3 appears, asking if you wish to automatically
discover the storage. This method will not find direct-attached storage
controllers. Close this screen to see the Enterprise Management Window
(Figure 9-4).
Figure 9-3 Automatic Discovery screen
5
You may need to add the directory to your path. Otherwise, type the full path to SMclient
Chapter 9. FAStT Storage Manager
193
Figure 9-4 Storage Manager Enterprise Management window
Select Edit -> Add Device from the menu. On the next screen, enter a Hostname
or IP Address for each controller blade in the FAStT (Figure 9-5).
Enter a Host name or IP address you set on the FAStT and click Add. Repeat the
process for the second controller blade.
Figure 9-5 Add Device screen
Highlight Storage Subsystem <unnamed> and select Tools -> Manage
Device.
194
Implementing Linux with IBM Disk Storage
If there was no configuration previously on the storage, you should see a
Subsystem Management screen similar to Figure 9-6.
Figure 9-6 Subsystem Management screen
9.1.3 Updating the FAStT firmware
Obtain the latest firmware and NVSRAM for the FAStT controller as described in
“Linux Install of the FAStT Storage Manager client” on page 191. Unzip the file to
a temporary folder:
unzip *.zip
This extracts the firmware and NVSRAM files for all of the current models of
FAStT controllers to the following folders:
./CONTROLLER CODE/FIRMWARE/<controller model>
./CONTROLLER CODE/NVSRAM/<controller model>
Note: You must first update the firmware and then the NVSRAM.
To update the controller firmware select Storage Subsystem -> Download ->
Firmware. The screen shown in Figure 9-7 displays. Enter the location of the
update file and click OK.
Chapter 9. FAStT Storage Manager
195
Figure 9-7 Firmware Download Location screen
You now see a screen similar to Figure 9-8. This graphic persists until the
firmware is transferred to the controllers and updated, which may take a few
minutes.
Figure 9-8 Firmware update screen
Next, update the NVSRAM. Select Storage Subsystem -> Download ->
NVSRAM. Enter the location of the NVSRAM update file and click Add
(Figure 9-9).
196
Implementing Linux with IBM Disk Storage
Figure 9-9 NVSRAM Download Location screen
A a screen similar to the one shown in Figure 9-10 displays.
Figure 9-10 NVSRAM Update screen
Confirm both controllers are at the same level by selecting View -> System
Profile.
Now that the controllers have been updated, we can use the Storage Manager to
prepare the storage for attachment to the host systems.
Chapter 9. FAStT Storage Manager
197
9.1.4 Setting up arrays, LUNs, and Storage Partitioning
This section takes you through the steps to set up storage arrays, LUNs, and
Storage Partitioning. Refer to the redbook, Fibre Array Storage Technology - A
FAStT Introduction, SG24-6246, for more on creating arrays and LUNs, selecting
RAID levels, Storage Partitioning topics, for considerations on the Access LUN
when defining the storage partitions (the sets of related logical drive mappings,
host groups, and host ports).
The system is now ready to start creating arrays, LUNs, and Storage Partitioning.
In this example we set up storage arrays for our High Availability cluster (see
Chapter 6., “Red Hat Cluster Manager” on page 139). We also use small logical
drive sizes to save on initialization time. Your requirements will likely be different
in detail, but similar in many respects.
Highlight the Unconfigured Capacity and select Logical Drive -> Create from
the menu.
Select Linux from the Default Host Type screen and click OK (Figure 9-11).
Figure 9-11 Default Host Type screen
198
Implementing Linux with IBM Disk Storage
The Create Logical Drive wizard starts. The wizard takes you through the stages
of creating arrays and logical drives. The first step is to select from the following
options:
򐂰 Free capacity on existing arrays
򐂰 Unconfigured capacity (create new array)
As we have no existing arrays at the stage we selected Unconfigured capacity,
and click Next (Figure 9-12).
Figure 9-12 Logical Drive Wizard screen
Next, select a RAID level, and the number of drives to be included in the array.
You also get the following choice:
򐂰 Automatic - Select from list of provided capacities/drives
򐂰 Manual - Select your own drives to obtain capacity
In most cases we recommend that you select Automatic. This is because the
FAStT allocation heuristic does a reasonable job of distributing I/O traffic across
available resources. However, manual selection may be preferable if there are
but few physical disks to be allocated, or if other special circumstances warrant.
With Automatic mode of this example,click Next (Figure 9-13).
Chapter 9. FAStT Storage Manager
199
Figure 9-13 Create Array screen
Now, you must specify the logical drive parameters. Enter the logical drive
capacity and a name. You also have the following options:
򐂰 Use recommended settings
򐂰 Customize settings (I/O characteristics, controller ownership, logical drive to
LUN mappings)
Select Customize and click Next (Figure 9-14).
200
Implementing Linux with IBM Disk Storage
Figure 9-14 Logical Drive Parameters screen
You can then specify advanced logical drive parameters. You can set the drive
I/O characteristics, preferred controller ownership, and drive-to-LUN mapping.
Make any changes to suit your environment. Logical drive-to-LUN mapping
should be left at Map later with Storage Partitioning. Click Finish
(Figure 9-15).
Note: Carefully consider any changes as they could dramatically affect the
performance of your storage.
Chapter 9. FAStT Storage Manager
201
Figure 9-15 Advanced Logical Drive Parameters screen
The next screen (Figure 9-16) prompts you: Would you like to create another
logical drive and gives you the following options:
򐂰 Same array
򐂰 Different array
In our case, we selected Different array to create a RAID5 array and then
selected Same array to create four logical drives within it (Figure 9-17).
Figure 9-16 Create a New Logical Drive screen
202
Implementing Linux with IBM Disk Storage
Figure 9-17 Arrays and logical drives
The systems involved in Storage Partitioning should be up and running in order
to get the Host Port Identifiers. You should not use the Default Group for security
reasons, but you should not delete this group.
Select the Mappings View tab in the Subsystem Management window. This
displays the Mappings Startup Help message. Read the text and close the
window (Figure 9-18).
Chapter 9. FAStT Storage Manager
203
Figure 9-18 Mappings Startup Help screen
Create a group for your Storage Partitioning. Select Mappings -> Define -> Host
Group. Enter a name a click Add (Figure 9-19).
Figure 9-19 Host Group added
204
Implementing Linux with IBM Disk Storage
Highlight the new Host Group and select Mappings -> Define -> Host. Enter a
host name and click Add. Repeat this for each system, which is to be a part of
this group. We are setting up a High Availability cluster, so we have two systems
in the group (Figure 9-20).
Figure 9-20 Hosts added
Highlight the first Host in the new group. Select Mappings -> Define -> Host
Port. Select the Host Port Identifier (collected earlier in the Fibre HBA BIOS
setup) which matches the first card in this system. Select Linux as the Host Type
(Figure 9-21).
Tip: If the Host Port Identifier drop-down is empty or missing the correct
entries, close Storage Manager, and restart the relevant system.
Chapter 9. FAStT Storage Manager
205
Figure 9-21 Define Host Port screen
Repeat this process for each Fibre HBA in the system. Repeat the process again
for each system in the group (Figure 9-22).
Figure 9-22 Host Ports added
206
Implementing Linux with IBM Disk Storage
Highlight the new Host Group, select Mappings -> Define -> Storage
Partitioning. The Storage Partitioning Wizard will appear. Read the text and click
Next (Figure 9-23).
Figure 9-23 Storage Partitioning Wizard screen
Select a Host Group or single Host for this partition. Click Next (Figure 9-24).
Figure 9-24 Select Host Group or Host screen
Chapter 9. FAStT Storage Manager
207
Select the logical drives to include in the partition and assign LUN IDs. These
IDs must start at zero and continue sequentially, you cannot miss a number out.
If you later remove a LUN you must re-assign the LUN IDs to continue without
skipping a number (Figure 9-25).
Figure 9-25 Select Logical Drives/LUNs screen
Highlight the logical drives to be included and click Add (Figure 9-26). Once you
have added all of the required logical drives, click Finish.
208
Implementing Linux with IBM Disk Storage
Figure 9-26 Logical drives added
The Access LUN (31) will probably be listed under Defined Mapping in your new
Host Group. This LUN is used by some other operating systems. Delete this LUN
as it may cause problems with multipathing under Linux. Highlight LUN31, press
Delete, click Yes to confirm.
As final step you have to reboot your system. If you want an automatic mount of
your volumes, you have to enter the appropriate parameters in /etc/fstab.
Important: Remember to make a copy of the FAStT profile.
For more about Storage Area Networks, please consult any of IBM’s Redbooks
on the topic.6
6
Introduction to Storage Area Networks, SG24-5470-01 and Designing and Optimizing an IBM
Storage Area Network, SG24-6419-00 are just two examples.
Chapter 9. FAStT Storage Manager
209
210
Implementing Linux with IBM Disk Storage
10
Chapter 10.
FAStT MSJ (Management
Suite for Java)
This chapter contains a description of the FAStT Management Suite for Java
Diagnostic and Configuration Utility. FAStT MSJ software is used to manage
paths available from fibre-attached hosts to storage.
© Copyright IBM Corp. 2001 2003. All rights reserved.
211
10.1 FAStT MSJ
If your installation uses the IBM FAStT Host Bus Adapter, then you will use the
IBM FAStT HBA driver. If you have multiple paths to your storage, this driver can
provide a failover capability, but the paths must be configured. This is the function
of the IBM FAStT Management Suite Java (MSJ) Diagnostic and Configuration
Utility. This tool allows you to view the current HBA configuration, attached
nodes, LUNs, and paths. It will allow you to configure paths and perform some
diagnostics.
Note: If you are not using IBM FAStT HBAs and the IBM FAStT driver, you can
skip this part.
When using two adapters, it is necessary to hide one path to the LUN away from
Linux. This is necessary because, like most OSs, Linux does not support
multipathing by itself. At boot time, the adapters negotiate the preferred path that
will be the only visible one. However, with the management tool, it is possible to
make an adequate configuration instead of getting the results of FC protocol
negotiations and driver design. Additionally, it is possible to choose the LUNs
visible for the OS. Also included is static load balancing that distributes access to
the individual LUNs via several paths.
FAStT MSJ has a GUI component. It also has an agent that runs on the host
containing the HBAs. The agent, qlremote, is quite small but the GUI is rather
large, so allow sufficient space (say 50 MB) for unpacking and installing the
product.
Installing FAStT MSJ
Obtain the latest version of the IBM FAStT MSJ from the same Web location as
described in FAStT Storage Manager client: Getting started.
The FAStT MSJ version is closely tied with the HBA version. Ensure that you
have the correct version. Extract the gzipped tar file:
tar zxvf *.tgz
This will extract the required files to ./FAStT_MSJ/Linux
As root, install FAStT_MSJ using the following command:
sh ./FAStTMSJ_install.bin
You should see a screen similar to Figure 10-1.
212
Implementing Linux with IBM Disk Storage
Figure 10-1 Installation Splash screen
You should then see the Introduction screen (Figure 10-2). Read the text, click
Next, and continue through the License screen and the Readme screen.
.
Figure 10-2 FAStT_MSJ Introduction screen
Chapter 10. FAStT MSJ (Management Suite for Java)
213
The next screen (Figure 10-3) allows you to choose which features of the product
you wish to install. You have four choices:
򐂰
򐂰
򐂰
򐂰
GUI and Linux Agent - This installs both on one system
GUI - Console for remote administration
Linux Agent - For systems to be managed remotely
Custom - To customize the components to be installed
The default selection installs the GUI and Linux Agent. Click Next.
Figure 10-3 Product Features screen
Next, you are prompted for a location in which to install the software. The default
folder is /opt/IBM_FAStT_MSJ. Click Next to accept the default, or choose a
different location according to your preference. Software installation will take a
few moments.
Note: During installation, a script is generated to uninstall the MSJ if desired.
Once installation is complete, click Done.
10.1.1 Configuring the LUN attachment
This section introduces you to the basic steps for administering your Fibre
Channel environment. Prior to describing how the settings are made, lets take a
moment to understand how the fibre-attached devices (or any devices, in
general) become effective. Device interfaces to the kernel are done using
214
Implementing Linux with IBM Disk Storage
software components called device drivers. Linux has two different ways of
integrating device drivers:
򐂰 Directly compiled to the kernel
򐂰 As loadable modules
In our work we used the second approach, and we used the FAStT drivers as
loadable modules. If it is necessary to have those drivers available at boot time to
launch the required scripts in time or to have access to the device during boot,
they may be compiled into the kernel, or the boot loader (GRUB) can load an
initial ramdisk, which contains all necessary drivers (see Figure 10-4).
Figure 10-4 Building a ramdisk image
You can see that mkinitrd scans modules.conf for the required modules,
including all options for the particular modules. As you can see in this case, the
drivers for the network, the ServeRAID, the Fibre Channel, and other modules
are included. In addition, a long options list (generated by FAStT MSJ, we will
see how to do this in 10.1.2, “Use FAStT MSJ to configure Fibre Channel paths”
on page 216) is submitted, too. These options contain all the necessary path and
LUN information to set up the FAStT multipath driver during boot. This leads to
the following results:
Chapter 10. FAStT MSJ (Management Suite for Java)
215
򐂰 Every time a change to your configuration occurs, you have to generate a
new initial ramdisk and reboot (or unload and reload the Fibre driver).
򐂰 It is quite useful to keep an additional bootable ramdisk with no options for the
FAStT HBA in reserve. After booting this ramdisk, the driver detects attached
devices and sets preferred paths, but these are not overruled by the options.
Then, with MSJ, you can generate a new, error free option string.
Note: Any change in the configuration of your SAN causes you to
reconfigure your paths and create a new initial ramdisk.
10.1.2 Use FAStT MSJ to configure Fibre Channel paths
In this section we set up the preferred and alternate paths for the LUNs.
The Linux agent is qlremote. This has to be running on the host before you open
the GUI client software. Open a terminal window and run qlremote (Figure 10-5).
Figure 10-5 Example of qlremote running
216
Implementing Linux with IBM Disk Storage
If the MSJ software is installed on the same host, leave this to run and open
another terminal window. Otherwise, open a terminal window on the
management workstation. In either case, type the following to run FAStT MSJ
client:
/opt/IBM_FAStT_MSJ/FAStT
When the FAStT MSJ is launched a screen similar to Figure 10-6 should appear.
Figure 10-6 FAStT_MSJ screen
Click Connect to connect to the agent you wish to administer. Enter the IP
address of the system where the qlremote agent is running and click Connect.
You should see a screen similar to Figure 10-7 showing you storage and LUNs.
The management system’s /etc/hosts file or the site DNS services should be up
to date as this program will try to resolve a hostname to the IP address. You will
get errors if this is not correct.
Chapter 10. FAStT MSJ (Management Suite for Java)
217
Figure 10-7 FAStT_MSJ connected
After connecting, the FAStT MSJ displays the adapters installed in the specific
host. In addition, you can see the Fibre Channel nodes attached to the fibre
HBAs.
To configure your environment, highlight either the host machine or an adapter
and click Configure. You will receive an error message (Figure 10-8). This is
because there is no configuration setup yet. Click OK and continue.
Figure 10-8 Invalid Configuration screen
You should then see a screen similar to Figure 10-9. If your system is connected
through an external switch, the labels will be slightly different (for example,
displaying “Fabric” instead of “Local”).
218
Implementing Linux with IBM Disk Storage
Figure 10-9 Port Configuration screen
For the easiest way to configure the ports, click Device -> AutoConfigure. You
will receive a message asking: Also configure LUNs After clicking Yes, the
preferred path is shown in white, while the hidden paths are shown in black.
Hidden means that these paths are not visible to the operating system. Because
the multipath driver is handling all I/O, they can be used as redundant paths
without interfering with the operating system view.
Before balancing the LUNs, the configuration will look something like
Figure 10-10. Select Configure, highlight the Node Name, select Device ->
Configure LUNs. You will see the preferred paths are highlighted in blue, while
the alternate paths are yellow. Also, you can see the preferred paths are marked
with a green circle, because this matches the settings that the drivers did at boot
time (and hence, the paths are “seen” by Linux).
Figure 10-10 Paths before balancing LUNs
Chapter 10. FAStT MSJ (Management Suite for Java)
219
Load balancing by path allocation
Select LUNs -> Load Balance -> All LUNs. When prompted to accept the
configuration and click Yes.
Click Save, enter the password, and click OK.
A screen will pop-up (Figure 10-11) informing you that the configuration has been
saved and you must reboot for the changes to take effect. Of course, this reboot
requirement refers to the system containing the HBAs.
Figure 10-11 Configuration Saved screen
The current status of the paths will look something like Figure 10-12, the red
circles indicate the new preferred path. This is marked red as you cannot change
the paths during runtime.
Figure 10-12 Balanced paths before rebooting
Once you have rebooted, you can launch the FAStT MSJ again and view how the
paths have been configured. (Figure 10-13).
220
Implementing Linux with IBM Disk Storage
Figure 10-13 Balanced paths after rebooting
Exit from FAStT MSJ, click File -> Exit. Stop qlremote on the host system by
pressing Ctrl-C in the correct terminal window or type the following:
killall -TERM qlremote
If you now view /etc/modules.conf you will notice that FAStT MSJ has added a
large options string. This string is used by the Fibre HBA drivers (Figure 10-14).
Figure 10-14 Example of modules.conf
As changes have been made to modules.conf we need to rebuild the ramdisk.
First run depmod -a to update the modules.dep file.
Chapter 10. FAStT MSJ (Management Suite for Java)
221
Build the ramdisk as before:
mkinitrd -f /boot/<newname>.img <kernel-version>
The -f option allows you to build a ramdisk to a name that already exists, i.e
overwrite it. In our case we used the following command with Red Hat Advanced
Server (your naming convention may vary from this example):
mkinitrd -f /boot/initrd-storage-2.4.9-e.12summit.img 2.4.9-e.12summit
If you used a new name for the ramdisk, update /etc/grub and reboot the system.
10.1.3 Using FAStT as storage: A summary
As described before, you must set up your storage before building the initial
ramdisk. Otherwise, the module will be loaded without a path configuration and
no manual load balancing will be possible.
First you install the required management software. Install (at least) qlremote
from the FAStT MSJ package on the host. After you have configured your
storage, launch the qlremote agent and access it with the MSJ (either locally or
from a management station). After you have set up your paths, MSJ instructs
qlremote to write an option string to the modules.conf, which is used to configure
the driver module during system boot.
Note: Run qlremote only during the usage of FAStT MSJ. Once you are done
and the output string is written to the modules.conf, stop qlremote
immediately!
You can verify your setup by unloading the module with rmmod qla2x00 and
reload with modprobe qla2x00. During this load the module will use the option
string carrying the path definitions and activate them. Then start qlremote again
and check the setting with FAStT MSJ.
SuSE notes
Before you build the new ramdisk you have to make the following change to the
file /etc/sysconfig/kernel: add qla2200 or qla2300 to the string behind
INITRD_MODULES, depending on your HBA designation:
INITRD_MODULES=”aic7xxx reiserfs qla2300”
The script /sbin/mk_initrd uses this string as input. By running mk_initrd a
new version of the initial ramdisk is built in the /boot directory. Even though this
is convenient you might build a new ramdisk instead of replacing the existing one.
Have a look at the mk_initrd script, it offers you quite some useful (and
sometimes colorful) options.
222
Implementing Linux with IBM Disk Storage
Red Hat notes
Before you restart you should add the kernel parameter max_scsi_luns=128 to
the modules.conf file. This is necessary because the kernel does not probe the
full range of LUNs by default. However, this is only critical if you are using a large
amount of SAN volumes.
Note: Every change in the configuration of your SAN causes the need to
reconfigure your paths and rebuild the initial ramdisk!
For more about Storage Area Networks, please consult any of IBM’s several
Redbooks on the topic.1
As final step, you have to reboot your system. If you want an automatic mount of
your volumes, you have to enter the appropriate parameters in /etc/fstab.
Important: Keep in mind that the failure of a volume prior or during reboot
may cause a change in the SCSI device numbering sequence.
1
Introduction to Storage Area Networks, SG24-5470-01 and Designing and Optimizing an IBM
Storage Area Network, SG24-6419-00 are just two examples.
Chapter 10. FAStT MSJ (Management Suite for Java)
223
224
Implementing Linux with IBM Disk Storage
11
Chapter 11.
Implementing VMware
VMware ESX Server is virtualization software that enables the deployment of
multiple, secure, independent virtual machines on a single physical server. The
purpose of this software is to provide an efficient and high performance platform
for consolidation and accelerated deployment of services.
This chapter starts with a brief overview of VMware and the VMware ESX
architecture, then focuses on the practical aspects of implementing VMware ESX
1.5.2 server on IBM eServer xSeries.
© Copyright IBM Corp. 2001 2003. All rights reserved.
225
11.1 Introduction
VMware, Inc. was founded in 1998 to bring mainframe-class virtual machines to
industry-standard computers. It offers the following products:
򐂰 VMware Workstation: This product allows you to run multiple operating
systems and their applications simultaneously on a single Intel based PC.
򐂰 VMware GSX server: GSX runs in a server OS, like Linux or Windows Server
Editions; it transforms the physical computer into a pool of virtual machines.
Operating systems and applications are isolated in multiple virtual machines
that reside on a single piece of hardware. System resources are allocated to
any virtual machine based on need, delivering maximum capacity utilization
and control over computing infrastructure. It can run up to 64 virtual machines
(provided that the server hardware can sustain the load).
򐂰 VMware ESX Server: Is virtualization software that enables the deployment of
multiple, secure, independent virtual machines on a single physical server, as
illustrated in Figure 11-1. It runs directly on the hardware (in contrast to
VMware Workstation and GSX Server products, which utilize host operating
system to access hardware) to provide a secure, uniform platform for easy
deployment, management, and remote control of other operating systems.
Figure 11-1 VMWare ESX server
The VMware ESX server can be used as platform for server consolidation. Many
Intel based server are often used for services that do not fully utilize the capacity
226
Implementing Linux with IBM Disk Storage
of the machine. Most systems run with an average system load of less than 25%
of that capacity. Servers like the x440 with up to 16 processors offer enough
power and scalability to move such services on one system.
Because VMware emulates the hardware, different operating systems, and the
applications they run are presented with a consistent set of virtual hardware by
each VM, irrespective of the physical hardware available in the system. VMs
have no dependency on the physical hardware, and as such do not require
device drivers specific to the physical hardware. This improves the mobility of a
VM. An installation done in a virtual machine can be moved to another server.
11.1.1 VMware ESX architecture
The design of the VMware ESX Server core architecture implements the
abstractions that allow hardware resources to be allocated to multiple workloads
in fully isolated environments.
The key elements of the system design are:
򐂰 The VMware virtualization layer, which provides the idealized hardware
environment and virtualization of underlying physical resources.
򐂰 The resource manager, which enables the partitioning and guaranteed
delivery of CPU, memory, network bandwidth, and disk bandwidth to each
virtual machine.
򐂰 The hardware interface components, including device drivers, which enable
hardware-specific service delivery while hiding hardware differences from
other parts of the system.
VMware ESX Server incorporates a resource manager and a Linux based
service console that provides bootstrapping, management, and other services.
While the service console is basically used for administration purposes, the
resource manager orchestrates access to physical resources and maps them to
specific VMs. For instance, physical disks are virtualized. Virtual disks are
created as regular files in the VMFS file system, and assigned to the VMs. These
virtual disks are presented to the VMs as SCSI drives connected to a SCSI
adapter.
This is the only disk storage controller used by the guest operating system,
despite the wide variety of SCSI, RAID, and Fibre Channel adapters that might
actually be used in the system.
Chapter 11. Implementing VMware
227
11.1.2 IBM and VMware
Some milestones in excellent IBM and VMware cooperation that help customers
to succeed with server consolidation projects include:
򐂰 December 2000: IBM becomes member of VMware Preferred Hardware
Partner Program.
򐂰 December 2001: IBM certifies VMware ESX Server for IBM Server Proven
Program.
򐂰 February 2002: IBM and VMware enter joint development agreement.
򐂰 August 2002: IBM announces xSeries + ESX Server bundles (with optional
IBM support).
The IBM Server Proven Program helps to ensure customers that a whole
proposed solution based on IBM technologies, including parts from other
partners, will work together smoothly without any compatibility issues.
IBM offers outstanding service and support for VMware ESX products. Support
packages are available through IBM Global Services that cover the entire
solution including the hardware, VMware ESX Server, supported guest operating
systems, and even the applications in most cases.
For more information about IBM and VMware relationship and VMware support
on IBM eServer xSeries servers follow the listed URLs:
http://www.pc.ibm.com/ww/eserver/xseries/vmware.html
http://www.pc.ibm.com/ww/compat/nos/vmware.html
11.1.3 VMware system requirements
Table 11-1 on page 229 summarizes the VMware ESX 1.5.2 system
requirements in regards to the server hardware and device support.
Consult http://www.pc.ibm.com/us/compat/nos/matrix.shtml for certification
status.
Please visit http://www.vmware.com and http://www.vmware.com/support
(restricted access) for supported guest operating systems.
Additionally, refer to the VMware ESX Server User’s Manual, MIGR-44890 for
details about the installation and setup procedures.
You can obtain an installation guide (document ID MIGR-44890) from:
http://www.ibm.com/support
228
Implementing Linux with IBM Disk Storage
Table 11-1 Server requirements
Server hardware
Server device support
򐂰 Processor :
򐂰 Local area networking
– Intel Pentium II,III, 4
– Broadcom 5700 Series Gigabit
Ethernet
– Intel e1000 Gigabit Ethernet
– Common 10//100 Cards
򐂰 Minimum system RAM
– 512 MB
򐂰 Minimum disk space
򐂰 Direct attached storage
– 2GB ++1GB per virtual
machine
–
–
–
–
򐂰 Minimum networking
– 2 Network Interface Ports
IBM ServeRAID
HP SmartArray
Dell Perc
Common SCSI Adaptors
򐂰 Storage Area Networking
– Emulex 8000,9000 series
– QLogic 2200,2300 series
11.2 Configuration
We used the following configuration for the purpose of this redbook:
Server:
IBM xSeries 330 with the following features:
򐂰
򐂰
򐂰
򐂰
Two INTEL Xeon 1.1 GHz
2 GB of memory
Two IBM FC2-133 Fibre Channel host adapter
Two 18 GB hard disk drives
Storage:
򐂰 IBM SAN switch 2109-F16
򐂰 FAStT700
򐂰 One EXP700 with four HDDs
Chapter 11. Implementing VMware
229
11.3 Implementing VMware ESX Server 1.5.2
This section describes the installation and configuration of the VMware ESX
server.
11.3.1 Installation steps
The installation of VMware is rather simple.
To begin the installation insert the VMware ESX CD and start up the server. The
dialog shown in Figure 11-2 displays. Click Install.
Attention: If you are planning to install with redundant paths you must
disconnect all Fibre Channel HBAs from the fabric. Please be aware that loops
are not supported by the QLogic device driver used with VMware ESX server.
The VMware kernel supports up to 7 LUNs by default.
However, you can set up your storage completely. Please refer to Chapter 9.,
“FAStT Storage Manager” on page 187 for further information.
Important: If you are installing on a x440 enter esx apic at the boot: prompt
when it appears.
Figure 11-2 Begin installation
The next dialog shown on Figure 11-3 asks for a driver disk. Actually, no
additional driver is required: Answer No.
230
Implementing Linux with IBM Disk Storage
Figure 11-3 Driver disk
The system now starts probing the hardware, and installing appropriate modules
for the detected hardware.
Upon completion of the hardware probing space, the Disk Partitioning Setup
screen (shown in figure Figure 11-4) displays.
Figure 11-4 Partitioning of console OS
It is essential to keep in mind that the partitioning done at this point pertains to
the console OS (storage partitioning for the virtual machines later will be done
later). Even though Autopartitioning is convenient, we suggest that you
partition the disk manually. Please keep the following recommendations in mind:
򐂰 The root partition should be at least 1.8 GB in size.
򐂰 For /boot about 25 MB are required.
Chapter 11. Implementing VMware
231
򐂰 The /home directory is used for two purposes: storing the configurations of the
virtual machines and storage for suspended virtual machines. This means
that you require 10 MB + (size of VM memory) for each virtual machine (if
your virtual machine has a memory size larger than 2 GB, the VMs must be
attached to VMFS volumes; see Figure 11-27).
– Apply the following rules for deciding the size of the swap partition:
•
•
•
•
•
1-4 VMs -> 128 MB
5-8 VMs -> 192 MB
9-16 VMs -> 272 MB
17-32 VMs -> 384 MB
> 32 VMs -> 512 MB
When the partitioning is done, the system prompts you as to where to locate the
boot loader (LILO), usually in the Master Boot Record (MBR).
The installation proceeds with the setup of a network interface (Figure 11-5). This
network is used by the console OS only. You need a static IP address for this
interface because all configuration and setup communications are carried
through this device.
Figure 11-5 Network setup of console OS
Next, you set up the time zone (screen not shown) and then specify a root
password (Figure 11-6). This password is required for any fundamental
modification to the VMware ESX server.
232
Implementing Linux with IBM Disk Storage
Figure 11-6 Define root password
Click OK.
The the next dialog allows you to define additional users of the system ().
These additional users can be later defined as administrators for the virtual
machines (creation, start, stop, and other maintenance tasks.). We strongly
recommend not to use the root user ID for the setup and administration of virtual
machines.
The installation continues until all the console OS and the VMware ESX server
components have been copied to the disk. When completed, the server reboots
and displays the LILO boot menu. If you do nothing the default setting (VMware
kernel) is loaded.
11.3.2 Configuration of the ESX server
Once the VMware ESX server is loaded, you can access the system either
locally via a console (press Alt+F2 to enter a console) or remotely. To do a
remote access for the first time, start a Web browser and enter the following
destination:
http://<VMware ESX IP address>
The VMware server responds to your request by transmitting a site certificate, as
shown in Figure 11-7.
Chapter 11. Implementing VMware
233
Figure 11-7 Security certificate of VMware ESX server
Accept the certificate, then enter the root user ID and password on the login
screen shown in Figure 11-8.
Figure 11-8 Login on VMware management console
If the login is successful, the next page that appears within your browser displays
the ESX server configuration menu (Figure 11-10) and ESX server management
menu (Figure 11-10).
234
Implementing Linux with IBM Disk Storage
Figure 11-9 VMware ESX server configuration
Chapter 11. Implementing VMware
235
Figure 11-10 VMware ESX server management
Go to the server configuration menu first to finalize the server setup and be able
to access the Fibre Channel storage.
We recommend that you start with license information (here you enter your
activation key) and the security settings.
Next, select Update Boot Configuration/Allocate Devices. A new page in your
browser displays the boot configuration (Figure 11-11) and the device allocation
(Figure 11-12) dialogs.
236
Implementing Linux with IBM Disk Storage
Figure 11-11 Boot configuration
In the boot configuration you can specify the name to display on the LILO screen,
the memory reserved for the console OS, and which VMware kernel file to load at
startup. Additionally, you can mark this configuration as default. For the console
OS memory, use the following rules:
򐂰
򐂰
򐂰
򐂰
򐂰
1-4 VMs -> 128 MB
5-8 VMs -> 192 MB
9-16 VMs -> 272 MB
17-32 VMs -> 384 MB
> 32 VMs -> 512 MB
The device allocation dialog in Figure 11-12 allows you to distribute your devices
between console OS and virtual machines. Although SCSI devices only can be
shared between VMs and console OS, we recommend that you keep them
separate.
Chapter 11. Implementing VMware
237
Figure 11-12 Initial device allocation
If you are planning to install two Fibre Channel HBAs for redundancy, you have to
take some details into account:
򐂰 By default the qla2x00 module is installed; this module does not support
failover.
򐂰 To support failover, both HBAs must be identical.
򐂰 Only Qla2200 and Qla2300 adapters are supported for failover.
򐂰 Both cards must see exactly the same targets.
򐂰 There is no load balancing.
򐂰 Failover is provided for both ESS and FAStT.
To replace the driver modules, you have to either connect to the server via ssh or
work directly from the console (assuming you did not change the security
settings yet). Once you have access to the console OS shell, you need to edit or
add (in case it does not exist yet) the file vmware-devices.map.local.
Figure 11-13 shows the changes required for VMware to recognize the new
settings. The file vmware-devices.map contains the hardware definition for the
VMware kernel.
238
Implementing Linux with IBM Disk Storage
[[email protected] root]# cd /etc/vmware
[[email protected] vmware]# cat vmware-devices.map.local
device,0x1077,0x2200,vmhba,QLA2200,qla2200.o
device,0x1077,0x2300,vmhba,QLA2300,qla2300.o
device,0x1077,0x2312,vmhba,QLA2300,qla2300.o
[[email protected] vmware]#
Figure 11-13 Extension required to use fail-over driver
The file vmware-devices.map.local is used to select different device drivers. At
boot time the file is evaluated. All entries override settings made in
vmware-devices.map.
Once the changes are made, the device allocation console displays the new
driver (Figure 11-15).
Now you can select the device with the correct device drivers for the virtual
machines.
Tip: To load the VMkernel driver manually, you have to enter at a shell
’/usr/sbin/vmkload_mod’ /usr/lib/vmware/vmkmod/qla2xxx/qla2300.o
vmhba
Before you can reboot to commit the changes, you must set up your external
storage. Refer to Chapter 9., “FAStT Storage Manager” on page 187 if you are
using FAStT, or Chapter 7., “Configuring ESS for Linux” on page 155 if you are
using ESS.
Chapter 11. Implementing VMware
239
Figure 11-14 Final assignment of devices
Now, attach the HBAs to your configured storage.
If you are using a FAStT you have to set one flag in the VMware kernel
(Figure 11-15 on page 241, and Figure 11-16 on page 241). This is necessary to
prevent VMware from misinterpreting status codes from FAStT as error codes
(LUN exists, but is not accessible).
240
Implementing Linux with IBM Disk Storage
Figure 11-15 Settings for FCal storage
Figure 11-16 Update VMkernel flag
Once you have saved the configuration, reboot the system to activate your
settings. To reboot, you must select Reboot/Halt System from the Server
Management dialog shown in Figure 11-10 on page 236.
Chapter 11. Implementing VMware
241
11.3.3 Setup disk space for VMs
After reboot, login to the management console again to start setting up disk
space for the virtual machines.
Important: This is a brief and summarized description. For details please
refer to the VMware ESX 1.52 server manual.
For our testing environment, we define three LUNs on the FAStT 700. Next,
following instructions given in the previous section, we install and load an
appropriate HBA driver and connect the server to the SAN. Figure 11-17 shows,
for our installation, that the three LUNs are correctly detected as physical drives.
To partition these drives you have two options:
Create New Partition Partitions the disk automatically; additionally a partition for
a VMware dump (crash dump space for VMware) is
created (Figure 11-18 on page 243).
Expert Mode Fdisk
Allows you to specify exactly, and based on your own
input, how to partition the volumes (Figure 11-19 on
page 244)
In addition to a label, you can specify the VMFS accessibility by three levels:
Private
Only the specified virtual machines can access the
partition.
Public
The partition is available to multiple physical servers and
their virtual machine, but only to one server at a single
time.
Shared
Shared access to the partition is possible; this setting is
required for shared disk solutions like fail-over clustering
between virtual machines.
The menu in Figure 11-9 on page 235 shows that you can create swap space for
the virtual machines (Figure 11-20 on page 244). This feature allows you to
assign more memory to your virtual machines than is physically available on the
server. Please keep in mind that intensive swapping negatively impacts the
system performance.
242
Implementing Linux with IBM Disk Storage
Figure 11-17 Volumes are visible
Figure 11-18 Create new partition
Chapter 11. Implementing VMware
243
Figure 11-19 Expert mode
Figure 11-20 Swap for VMware (more logical than physical memory)
244
Implementing Linux with IBM Disk Storage
11.3.4 Define a VM using the wizard
To create a virtual machine, click Virtual Machine Wizard in Figure 11-10 on
page 236.The page that displays in your browser let you specify the major
characteristics of a VM. The page contains all the elements shown in
Figure 11-21 on page 245 through Figure 11-26 on page 247.
򐂰 In the Basic Settings ()section, you define the guest OS to install, select a
display name for this VM, specify the path for the configuration file, and the
memory size available to the VM.
Figure 11-21 Define VM basic settings
򐂰 In the SCSI Disk section, select one of the partitions you had set up before.
There are four disk modes available:
Persistent
All write accesses by the guest are written
permanently to the disk immediately.
Nonpersistent
All changes are discarded when a virtual machine is
shut down and powered off.
Undoable
All changes are written to a log file and can be
discarded at anytime.
Append
All changes are stored in a log file; the changes can be
discarded or made permanent.
Figure 11-22 Define VM SCSI disks
Chapter 11. Implementing VMware
245
򐂰 In the Networking section you can have a choice between two different types
of NIC:
vmnic
Binds the virtual NIC to a physical NIC; each physical
NIC has its own VMnic device number
vmnet
Binds the virtual NIC to a NIC of the virtual network
provided by the VMware
Additionally, you can select the type of the virtual NIC (vmlance or vmxnet).
Figure 11-23 Define VM networking
򐂰 The CD-ROM section allows you to manage the access of a virtual machine
to the CD drive. Alternatively, you can specify an ISO file instead of a physical
CD-ROM.
Figure 11-24 Define VM CD-ROM
򐂰 The floppy drive section is similar to the CD-ROM section.
Figure 11-25 Define VM floppy drive
246
Implementing Linux with IBM Disk Storage
򐂰 In the Misc section, you can specify the suspend location (necessary, if your
virtual machines uses more than 2 GB of memory).
Figure 11-26 Define VM miscellaneous options and create VM
When you are satisfied with the settings, click Create VM.
The page in your browser now displays a summary screen for the VM just
created. Selecting Return to Overview on this page, takes you to the dialog
shown in Figure 11-27 on page 248.
11.3.5 Install VMware remote console
Before you launch and monitor your virtual machine you have to install the
VMware remote console (see middle left in Figure 11-27 on page 248) on your
management workstation.
You can choose between a Windows or a Linux console. Please refer to the
dialog for installation details.
If you install the remote console on Windows, you can launch it by right-clicking
on the folder symbol located in the upper left corner of the screen (see
Figure 11-27 on page 248) and then select Launch Remote Console from the
pull-down menu.
Chapter 11. Implementing VMware
247
Figure 11-27 Installation of remote console and launch dialog
On Linux workstations you have entered the command vmware-console The
remote console is launched and a dialog asks you for IP address and login
information (user and password) to access the ESX server (Figure 11-28).
Figure 11-28 Linux Remote Console: connecting to VMware server
248
Implementing Linux with IBM Disk Storage
Clicking OK connects to the specified server. The next window shows you the list
of available VM configurations (Figure 11-29). There is only one defined in our
illustration.
Figure 11-29 Linux Remote Console: select configuration
After selecting a configuration the console displays as shown in Figure 11-30. To
start the virtual machine, click Power On.
Figure 11-30 Linux Remote Console
Chapter 11. Implementing VMware
249
The virtual machine boots like a regular system. The pictures in Figure 11-31,
Figure 11-32 on page 251, and Figure 11-33 on page 251) illustrate what
appears within the remote console window when you boot a VM defined as a
Linux system that is about to be installed.
In this scenario, we had previously copied the ISO-images of the Linux
installation CDs to our ESX server and connected the (virtual) CD drive of our
Linux VM to an ISO-image (see Figure 11-24 on page 246). This means that
whenever the guest OS demands a different CD (Figure 11-31) you have to
disconnect the existing ISO file (Figure 11-32 on page 251), define a new one,
and connect it to the virtual machine (Figure 11-33 on page 251).
Figure 11-31 Linux Remote Console: Start guest OS installation
250
Implementing Linux with IBM Disk Storage
Figure 11-32 Connect to different ISO-file part 1
Figure 11-33 Connect to different ISO-file part 2
Chapter 11. Implementing VMware
251
11.3.6 Monitoring a VM
Once you set up virtual machines and start using them (i.e. in a production
environment) VMware allows you to monitor the virtual machines. Figure 11-34
below shows the VM Overview. For tuning purposes you can modify some
parameters of the virtual machine by clicking on Edit VM Resources.
Please refer to the manual for details on optimizing your resources.
Figure 11-34 Active VM
252
Implementing Linux with IBM Disk Storage
A
Appendix A.
Storage from the OS view
In this appendix we discuss the concept of storage from the perspective of the
operating system, and how this storage is presented to an application. Our intent
is to introduce to the person familiar with Windows some of the differences
evident in Linux. We will also briefly survey the file systems available with Linux.
We will show how to use a feature of Linux called the /proc file system to view
attached storage, and briefly mention the file system table /etc/fstab.
© Copyright IBM Corp. 2001 2003. All rights reserved.
253
Disk storage in the Windows environment
As is often the case, one cannot grasp the state of the art (or at least its
idiosyncrasies) without examining the roots of the technology. So, it is with disk
storage in Windows. We will not return to the absolute beginning of computer
technology. Instead, we will begin slightly later to look at drive letters. The
Windows discussion is oversimplified by design, our apologies to readers who
know the development history and technical architecture better than we describe.
Our emphasis, though, is on Linux, so we discuss Windows only in contrast.
Disk letters
In DOS,1 a disk is assigned a letter. Traditionally, “A” is the boot floppy, “B” is the
second floppy. It is a bit of a stretch to recall that this once was the extent of
external storage available. To add non-removable disk storage, a hardware
method was needed. The BIOS provides hard disk access via a hardware
interrupt (INT13). Through the BIOS, a disk looks like one big file. DOS breaks up
this storage into files and directories by means of a File Allocation Table (FAT).
The first unused letter (historically speaking) was “C” thus the notion of the “C”
drive being the first hard disk on an Intel-based system survives to this day. A
subtle but important implication is that the BIOS has a single interrupt to service
this storage type, so only one hard drive can be presented to the operating
system at any given time. This is of no consequence to DOS, as it cannot
multitask.
But the notion of an “attached volume” persisted even as complexity evolved.
With the advent of multiple hard drives, each “volume” was independent of the
others, there was no hierarchy that spanned multiple volumes. The operating
system image had to be resident in memory to permit the mounted volume to be
changed. Again, not an issue with DOS. The FAT was sufficient to locate a file on
the attached volume, and the operating system had no intrinsic recourse to the
disk once it was loaded. This scheme was adequate until the operating system
evolved to use virtual memory, when it had occasion to use disk storage for its
own purposes.
Until Windows 3.11, the underlying disk accesses were all handled through the
BIOS by DOS. Windows NT employed a server-oriented design derived from
OS/2 that was not constrained in this manner. Even so, Microsoft’s desire for
commonality between desktop and server environments ensured the disk
naming convention (if not the access method) would be carried into Windows NT
and beyond.
1
254
Using letters for disk attachment is not original to DOS, but that is not relevant to this introduction.
Implementing Linux with IBM Disk Storage
Disk partitions
Fundamental constraints inherited from earlier designs limited the addressable
size of a mounted volume. Disk capacities outstripped physical addressability at
a precipitous rate. It became necessary to subdivide individual hard drives into
smaller, completely addressable pieces. The concept of disk partitioning is
common to both Windows and Linux, and we will revisit it in subsequent sections.
Suffice it to say that Windows partitioning permits up to four physical divisions of
a disk. The last division may be subdivided, but these subdivisions are not visible
at boot. Disks are not presented to Windows based on their adapter type or
access protocol as is done in Linux.
Disk storage in the Linux environment
Although Linux is (rather emphatically) not UNIX, it is UNIX-like and shares the
UNIX disk access paradigm. Linux, in its simplest invocation, needs only to be
given its bootstrap and a “root device.” The operating system expects to find
everything it needs to run on this root device, which can be a floppy disk, a
ramdisk, a CD-ROM, hard drive, or network. In addition to the initial load of the
kernel, are two necessary elements: a device driver that is particular to the media
to be viewed as a disk, and a nexus in the file system by which the media can be
accessed. It is in some sense a paradox that the operating system should need
an entry in the file system to access the device on which it has a file system. This
is because the access point of the device driver looks like a file. In fact, a design
element of UNIX-like systems is that everything looks like a file such as disks,
pipes, directories, and printers, which mainly look the same.
Characteristically, the disk devices are specified in the file system as “special” file
entries in the /dev directory. With each of these “special” files is associated a set
of indices (major and minor numbers, see Appendix A-1, “/proc/partitions” on
page 259, or the output of the command ls -l /dev) that correlate to the
physical attachment (for example, the controller, target, and logical unit) and the
driver semantics associated with the desired access (for example, character or
block mode access). The first device to be accessed is declared by the hardware
and is termed “xxy” where xx indicates the device adapter type, and y
enumerates the first device on that adapter. A SCSI disk, for example, might be
seen as sda, while the second IDE hard disk might be seen as hdb. Once Linux
is booted, we may expect to see entries such as /dev/sdb0 for the first partition of
the second SCSI disk, or /dev/hda2 for the third (0,1,2...) partition on the first IDE
hard disk, and so forth.
Appendix A. Storage from the OS view
255
Disk partitions
Linux does not use the BIOS, but other systems that do might have to share a
disk with Linux. Thus, Linux is aware of data structures and BIOS limitations that
might influence such co-residency. For a purely Linux disk, these issues do not
affect partition choices except for the adherence to four physical partitions. Since
the topic of sharing a Linux disk with other operating systems (especially the boot
disk) is well documented2 elsewhere, we will not pursue it here. When
unconstrained by co-residency requirements, Linux disks can be partitioned
according to your taste, but there are some things to remember.
First, Linux disks, like Windows disks, can have only four physical partitions.
Additional logical partitions can be created in one of these physical partitions,
known as an extended partition, similar to the method used for Windows.
Second, Linux disks are assigned names in a manner that can cause great
difficulty if changes are made later. Lower SCSI ID numbers are assigned
lower-order letters. If you remove one drive from the chain, the names of the
higher ID number drives will change. A related note that we point out in several
places in this book is to be careful to avoid gaps in the LUN numbering, as Linux
will “stop counting” while probing when it reaches the first unavailable LUN.
Partitioning disks is done to achieve a physical separation of data according to
type or function. A common tool to do this is fdisk but there are others such as
cfdisk or even third-party tools like Partition Magic. No matter what the method,
the goal is the same - to create a structure that can accommodate a file
hierarchy. A hierarchical arrangement of directories (or folders) containing files
can be as simple as a single tree that spans an entire disk. This is often done
with trial installations but is not recommended for production use3. There are
some useful sizing guidelines that we will discuss.
File hierarchy
Work by the Filesystem Hierarchy Standard Group has been published4 to assist
in interoperability of UNIX-like systems (including Linux). This group
recommends segregating data into major subdivisions that derive from the
shareability and variability of data.
For example, data that is shared between multiple systems should be segregated
from data that is private to a single machine. Similarly, data that is highly variable
should be segregated as much as possible from data that is static. Partitions
2
See The Linux Documentation Project, Large Disk HOWTO and the Partition HOWTO at
http://www.tldp.org
3 To be clear, we are discussing the system disk. Additional disks can be subdivided as required to
accommodate the applications.
4
http://www.pathname.com/fhs/2.2/
256
Implementing Linux with IBM Disk Storage
enforce the implementation of this hierarchy by imposing physical constraints on
the segments (files cannot span partitions) and allowing control over the access
protocols to the segments (partitions are a convenient unit of access control as
well as share policies, just as within Windows).
For a Linux example, /usr could be sharable and static, while /boot would be
static, but unshareable. Similarly, /var/mail may be sharable and variable, while
/var/lock would be variable but unshareable.
Regardless of the division policies, all branches of a hierarchical structure must
eventually share a common point of origin, called the “root” (represented by a
forward slash, “/”). The root partition contains enough information to boot, repair,
and recover a system. Because system installation and recovery may require
booting from smaller media, the root partition may be physically quite small.
Essentially, the root partition contains directory entries, some of which may (or
should) be separate physical partitions. The hierarchy is then assembled using
the concept of “mounting” to be discussed in Appendix , “The file system table
/etc/fstab” on page 262.
You can explore the rationale in more detail depending on your interest5 but we
suggest segregating partitions and size ranges for the following mount points:
򐂰
򐂰
򐂰
򐂰
򐂰
򐂰
/ - Small to medium (tens to 100s of MB)
/usr - Large (several GB)
/opt - Medium to large (100s of MB up to several GB)
/var - Medium (100s of MB)
/tmp - Medium (100s of MB) or larger, depending on application requirements
/home - Very small to large, depending on user private space requirements
A separate partition not shown here (because it contains no file hierarchy) is
declared for system swap, usually 1 or 2 times the size of system memory. Also,
a small partition may be mounted on /boot for ease of location on systems with
BIOS constraints causing addressing limitations.
There are some interesting constituents of the file hierarchy. We have briefly
discussed the /dev directory,6 where device special nodes are conventionally
aggregated. In “Identifying your SCSI storage” on page 259 we will introduce the
/proc pseudo-hierarchy. These are not disk partitions; they merely exploit the
“everything-looks-like-a-file” paradigm discussed above. But every partition
where you want to write a file hierarchy needs to be allocated a branch of the tree
that begins with “root” (/). Their names can be arbitrary (although conventions
exist), and they are managed via a table of entries in the plain-text file /etc/fstab.
Before discussing the content of that file in “The file system table /etc/fstab” on
5
6
ibid.
Linux Allocated Devices, http://www.kernel.org/pub/linux/docs/device-list/devices.txt
Appendix A. Storage from the OS view
257
page 262, let’s talk about possible file systems that may be selected for disk
storage with Linux.
File systems
We have discussed a hierarchy of files, which is sometimes referred to as a file
system. There is a more fundamental structure that underlies this hierarchy that
is properly known as a file system, and it is important to choose one that meets
your needs. Just as Windows has the FAT and NTFS file systems, Linux also has
more than one type. In fact, the selection is quite rich. In addition to being able to
read file systems from a number of other operating systems including Windows,
UNIX and OS/2, Linux can reside in and make read/write use of several varieties
including ext2, ext3, reiserfs, xfs, JFS, and others. What are these and why are
there so many choices? The Linux System Administrator’s Handbook introduces
file systems this way:
“A file system is the methods and data structures that an operating system
uses to keep track of files on a disk or partition; that is, the way the files are
organized on the disk. The word is also used to refer to a partition or disk
that is used to store the files or the type of the file system. Thus, one might
say “I have two file systems'' meaning one has two partitions on which one
stores files, or that one is using the ``extended file system'', meaning the
type of the file system.
The difference between a disk or partition and the file system it contains is
important. A few programs (including, reasonably enough, programs that
create file systems) operate directly on the raw sectors of a disk or
partition; if there is an existing file system there it will be destroyed or
seriously corrupted. Most programs operate on a file system, and
therefore will not work on a partition that does not contain one (or that
contains one of the wrong type).
Before a partition or disk can be used as a file system, it needs to be
initialized, and the bookkeeping data structures need to be written to the
disk. This process is called making a file system.”7
SuSE Linux uses the rreiserFS as its default file system, other distributions may
default to ext2 or ext3. The reasons may be historical, but that is seldom the best
criterion for selection. The ext2 file system has a 10-year production history, so
its stability is attractive. It suffers from fragmentation when called upon to store a
large number of small (a few kilobyte) files. The ext3 file system is ext2 plus
journalling, however, the journalling implementation is dependent on kernel
journalling, which can be considered fragile, yet ext3 can also journal both data
and metadata, which is a feature not available with other journalling file systems.
ReiserFS is a journalling file system that excels in both space and time efficiency
7
258
http://tldp.org/LDP/sag/ - chapter 6.8
Implementing Linux with IBM Disk Storage
with variably-sized files because of its internal structure. IBM’s JFS is very
efficient in throughput for server-like environments with large files.
Considerations imposed by higher-level disk aggregation programs and volume
managers such as LVM, md, and Veritas VxVM are beyond the scope of this
discussion, although their use may affect the choice of an underlying file system.
Rather than suggest a default “one-size-fits-all” choice, we encourage you to
research file system types to determine the best one for your application. Dan
Robbins, CEO of gentoo, Inc. has posted a thirteen-part tutorial on file systems
on the IBM developerWorks™ site8. Look for more useful information at IBM’s
developerWorks tutorial site9.
Identifying your SCSI storage
The Linux /proc-file system offers you a large variety of possibilities to checkout
your connections. In this appendix we want to introduce to them shortly.
To check out the SCSI HBAs operating in your system, view the file
/proc/partitions. This output contains information about which SCSI storage
device are detected, and to which device it is associated. Figure A-1 shows you
the look-a-like in our example.
linux:/etc # cat /proc/partitions
major minor #blocks name
rio rmerge rsect ruse wio wmerge wsect wuse
running use aveq
8
0
17774160 sda 6064 16164 177548 25180 1293 327 13088 86720 0
24760 111900
8
1
3144361 sda1 6048 16134 177450 25090 1293 327 13088 86720 0
24670 111810
8
2
1 sda2 1 0 2 10 0 0 0 0 0 10 10
8
5
1048099 sda5 2 3 16 10 0 0 0 0 0 10 10
8
6
13561663 sda6 1 3 8 0 0 0 0 0 0 0 0
8
16
71014400 sdb 15 30 90 1870 0 0 0 0 0 1870 1870
8
17
1 sdb1 1 0 2 0 0 0 0 0 0 0 0
8
21
71007237 sdb5 1 3 8 0 0 0 0 0 0 0 0
8
32
35507200 sdc 19 42 122 4980 0 0 0 0 0 4980 4980
8
33
1 sdc1 1 0 2 0 0 0 0 0 0 0 0
8
37
35507168 sdc5 1 3 8 10 0 0 0 0 0 10 10
Figure A-1 /proc/partitions
8
9
Start here: http://www-106.ibm.com/developerworks/linux/library/l-fs.html
http://www-106.ibm.com/developerworks/views/linux/tutorials.jsp
Appendix A. Storage from the OS view
259
The system has three disks attached (sda, sdb, and sdc). Each disk is
partitioned.
The folder /proc/scsi contains informations about the SCSI adapter as well as
about the attached components. For each SCSI HBA device driver is a folder in
/proc/scsi. Within this folder you will find a file for each HBA using this driver. By
viewing these files you can learn which adapter corresponds with which SCSI
host ID. The file scsi informs you about the physical devices attached to your
SCSI buses including ID and LUN. Figure 11-35 on page 260 shows an example:
on SCSI HBA 2 is a device attached. The device is of the type 1742, which is the
model number of the FAStT700.
linux:/etc # cat /proc/scsi/scsi
Attached devices:
Host: scsi1 Channel: 00 Id: 09 Lun: 00
Vendor: IBM
Model: GNHv1 S2
Type:
Processor
Host: scsi1 Channel: 00 Id: 12 Lun: 00
Vendor: IBM-ESXS Model: ST318305LC
Type:
Direct-Access
Host: scsi2 Channel: 00 Id: 00 Lun: 00
Vendor: IBM
Model: 1742
Type:
Direct-Access
Host: scsi2 Channel: 00 Id: 00 Lun: 01
Vendor: IBM
Model: 1742
Type:
Direct-Access
Rev: 0
ANSI SCSI revision: 02
!# Rev: B244
ANSI SCSI revision: 03
Rev: 0520
ANSI SCSI revision: 03
Rev: 0520
ANSI SCSI revision: 03
Figure 11-35 Content of /proc/scsi/scsi
This device provides two LUNs: 0 and 1. Actually, if the storage is invisible to your
system this is the perfect spot to look for. Linux does not probe for LUNs beyond
gaps. For example, during Storage Partitioning on a FAStT you can assign
individual LUNs to the logical drives associated to a Host Group. However, if
there is either a gap in between (e.g. 0,1,3,4) or the assignment incorrect
(2,3,4,5) Linux does not detect all LUNs.
At the same time both files can help you finding errors in your SAN setup (for
example, wrong port mapping causing seeing devices double).
For more in depth investigation you can collect informations about a generic
devices by using the sg_scan command.
Figure 11-36 shows the sample output of the command.
260
Implementing Linux with IBM Disk Storage
linux:/etc # sg_scan
/dev/sg0: scsi1 channel=0
/dev/sg1: scsi1 channel=0
/dev/sg2: scsi2 channel=0
/dev/sg3: scsi2 channel=0
id=9 lun=0 type=3
id=12 lun=0 type=0
id=0 lun=0 type=0
id=0 lun=1 type=0
Figure 11-36 sg_scan
It delivers the generic devices attached including their ID and LUN. For further
details, launch the command sg_inq with the corresponding device delivered by
sg_scan.
linux:/etc # sg_inq /dev/sg2
standard INQUIRY:
PQual=0, Device type=0, RMB=0, ANSI version=3, [full version=0x03]
AERC=0, TrmTsk=0, NormACA=1, HiSUP=1, Resp data format=2, SCCS=0
BQue=0, EncServ=1, MultiP=0, MChngr=0, ACKREQQ=0, Addr16=0
RelAdr=0, WBus16=1, Sync=1, Linked=0, TranDis=0, CmdQue=1
length=36 (0x24)
Vendor identification: IBM
Product identification: 1742
Product revision level: 0520
Product serial number: 1T23563008
Figure 11-37 sg_inq
Figure 11-37 shows the output of such an inquiry.
Using your storage on Linux
Once you have installed the OS and driver, the next step is to set up multipathing
if it is available. Having a driver we were able to verify that the system “sees” the
storage in a way that is planned or at least expected. For the OS, it does not
matter how the storage is organized (single disks, logical drives in an array, etc.).
All devices are addressed as a physical disk. This means that in order to mount
and use the drive, appropriate partitioning is required. This is usually done with
the command fdisk /dev/sd?
where ? stands for the device as stated in /proc/partitions (e.q. fdisk /dev/sdc).
Once the drive is partitioned the new partition table has to be loaded. This is
usually done by the boot of the system. However, during the implementation
phase it is sometimes feasible to unload the device driver with the command
rmmod <modulename>. The command lsmod lists you all loaded modules. Now
Appendix A. Storage from the OS view
261
reload the module with modprobe <modulename> again. Well, sure enough you
might simply reboot the system.
The final step is writing a file system to the partition (if it is not used as raw
device). To make a ext2 file system, for example, you simply have to enter the
command mkfs -t ext2 /dev/sd?* where ? is the device number and * the
partition number. In order to mount it, you need a mount point which you can
create by the command mkdir /path/my-mount-point. After you have mounted
the drive by mount -t fs /dev/sd?* /path/my-mount-point your storage is now
accessible for the system. To make the mount persistent, make an entry in
/etc/fstab.
To promote a new LUN to the /proc file system from the command line, use the
following command:
echo “scsi add-single-device $host $channel $id $lun” > /proc/scsi/scsi
This is the proper way if you cannot afford to either unload the driver module or
reboot the system.
The file system table /etc/fstab
Mounting a file system is nothing more than attaching a branch to the file
hierarchy, where the branch is contained on a partition. As is common in Linux,
many things can be “mounted” that are not truly partitions (for example, the /proc
pseudo-file system). Our interest here is partitions, and the file that controls the
attachment is /etc/fstab. The structure of this file is described in the online
documentation (man 5 fstab) and we can synopsize it here with an example.
# filesystem
mount-point
fs-type
options
dump
fsck-order
/dev/sda0
/
reiserfs
rw
1
1
/devsda1
swap
swap
pri=1
0
0
proc
/proc
proc
defaults
0
0
Table A-1 Example /etc/fstab entries
The entry under the file system heading specifies the device (or pseudo device,
as in the example of proc) to be attached (“mounted”). The mount point specifies
the location in the tree to which the branch will be attached. The fs-type specifies
the file system type that Linux should expect. Options vary according to file
system type, we will not enumerate the options here. The final two fields specify
the backup-priority and the order in which file systems are checked at boot.
These fields are gradually losing currency as file systems and archiving
262
Implementing Linux with IBM Disk Storage
strategies evolve. Reading the online documentation is your best source of
further information on these fields, as well as the options field.
Appendix A. Storage from the OS view
263
264
Implementing Linux with IBM Disk Storage
Glossary
A
B
Access logical drive. The Access Logical
Drive is a special drive that uses none of the
physical disk drives and should be assigned to
the last (highest) available LUN number.
Typically, the LUN number will be LUN 31. The
Access Logical Drive allows the Storage
Manager Agent to communicate to the Fibre
Channel RAID controllers through the fibre
connection for storage management services.
These services include monitoring,
configuring, and maintaining the RAID storage
device.
bay. Physical space on an ESS rack. A bay
contains SCSI, ESCON, or Fibre Channel /
FICON interface cards.
allocated storage. On the ESS, this is the
space that you have allocated to volumes, but
not yet assigned.
array.The group of physical disk drive
modules (DDMs) that are divided into logical
drives and allocated to hosts.
assigned storage. On the ESS, this is the
space that you have allocated to volumes, and
assigned to a port.
asynchronous operation. A type of operation
in which the remote copy XRC function copies
updates to the secondary volume of an XRC
pair at some time after the primary volume is
updated. Contrast with synchronous
operation.
availability. The degree to which a system or
resource is capable of performing its normal
function.
© Copyright IBM Corp. 2001 2003. All rights reserved.
backup. The process of creating a copy of
data to ensure against accidental loss.
C
cache. A random access electronic storage in
selected storage controls used to retain
frequently used data for faster access by the
host.
CCW. Channel command word.
chunksize. The number of data blocks,
assigned by a system administrator, written to
the primary RAIDset or stripeset member
before the remaining data blocks are written to
the next RAIDset or stripeset member.
channel. (1) A path along which signals can
be sent; for example, data channel and output
channel. (2) A functional unit, controlled by the
processor, that handles the transfer of data
between processor storage and local
peripheral equipment.
channel connection address (CCA). The
input/output (I/O) address that uniquely
identifies an I/O device to the channel during
an I/O operation.
channel interface. The circuitry in a storage
control that attaches storage paths to a host
channel.
265
channel path. The ESA/390 term for the
interconnection between a channel and its
associated controllers.
channel subsystem. The ESA/390 term for
the part of host computer that manages I/O
communication between the program and any
attached controllers.
CKD. Count key data (CKD) is the disk
architecture used by zSeries (and S/390)
servers. Because data records can be variable
length, they all have a count field that
indicates the record size. The key field is used
to enable a hardware search on a key,
however, this is not generally used for most
data anymore. ECKD is a more recent version
of CKD that uses an enhanced S/390 channel
command set. The commands used by CKD
are called Channel Command Words (CCWs);
these are equivalent to the SCSI commands
CKD Server. This term is used to describe a
zSeries 900 server (or a S/390 server) that is
ESCON or FICON connected to the storage
enclosure. In these environments the data in
the storage enclosure for these servers is
organized in CKD format. These servers run
the z/OS, OS/390, MVS, z/VM, VM, VSE and
TPF family of operating systems
cluster.
See storage cluster
cluster processor complex (CPC). The
unit within a cluster that provides the
management function for the storage server.
It consists of cluster processors, cluster
memory, and related logic.
concurrent copy. A copy services function
that produces a backup copy and allows
concurrent access to data during the copy.
concurrent maintenance. The ability to
service a unit while it is operational.
consistent copy. A copy of data entity (for
example a logical volume) that contains the
266
Implementing Linux with IBM Disk Storage
contents of the entire data entity from a
single instant in time.
control unit address (CUA). The high order
bits of the storage control address, used to
identify the storage control to the host
system.
Note: The control unit address bits are set
to zeros for ESCON attachments.
CUA. Control unit address.
D
DA. The SSA loops of the ESS are
physically and logically connected to the
Device Adapters (also see device adapter)
DASD. Acronym for Direct Access Storage
Device. This term is common in the z/OS
environment to designate a disk or z/OS
volume
DASD subsystem. A DASD storage control
and its attached direct access storage
devices.
data availability. The degree to which data
is available when needed. For better data
availability when you attach multiple hosts
that share the same data storage, configure
the data paths so that data transfer rates are
balanced among the hosts.
data sharing. The ability of homogenous or
divergent host systems to concurrently
utilize information that they store on one or
more storage devices. The storage facility
allows configured storage to be accessible
to any attached host systems, or to all. To
use this capability, you need to design the
host program to support data that it is
sharing.
DDM. See disk drive module.
dedicated storage. Storage within a
storage facility that is configured such that a
single host system has exclusive access to
the storage.
interconnection logic to make the DDMs
accessible to attached host systems.
device. The zSeries 900 and S/390 term for
a disk drive.
dump. A capture of valuable storage
information at the time of an error.
device adapter. A physical sub unit of a
storage controller that provides the ability to
attach to one or more interfaces used to
communicate with the associated storage
devices.
Device Support Facilities program
(ICKDSF). A program used to initialize
DASD at installation and perform media
maintenance in zSeries 900 and S/390
environments.
DFDSS. Data Facility Data Set Services
(see DFSMSdss)
DFSMSdss. A functional component of
DFSMS/MVS used to copy, dump, move,
and restore data sets and volumes.
disaster recovery. Recovery after a
disaster, such as a fire, that destroys or
otherwise disables a system. Disaster
recovery techniques typically involve
restoring data to a second (recovery)
system, then using the recovery system in
place of the destroyed or disabled
application system. See also recovery,
backup, and recovery system.
disk drive module (DDM). The primary
nonvolatile storage medium that you use for
any host data that is stored within a
subsystem. Number and type of DDMs
within a storage facility may vary. DDMs are
the whole replaceable units (FRUs) that hold
the HDDs
disk-group. In the ESS, a group of 7 or 8
DDMs that are not yet formatted as ranks.
disk. Disks (or maybe logical disks) are the
logical representations of a SCSI or FCP disk
as seen from the server system. In reality, a
disk may span multiple physical disks, and the
size of the disk is set when the disk is defined
to the storage enclosure.
Disk drive. See disk drive module (DDM).
E
ESCON. Enterprise Systems Connection
Architecture. An zSeries 900 and S/390
computer peripheral interface. The I/O
interface utilizes S/390 logical protocols over
a serial interface that configures attached
units to a communication fabric.
ESCON Channel. The ESCON channel is a
hardware feature on the zSeries and S/390
servers that controls data flow over the
ESCON Link. An ESCON channel is usually
installed on an ESCON channel card which
may contain up to four ESCON channels
ESCON Host Adapter. The host adapter
(HA) is the physical component of the storage
server used to attach one or more host I/O
interfaces. The ESS can be configured with
ESCON host adapters (HA). The ESCON host
adapter is connected to the ESCON channel
and accepts the host CCWs (channel
command words) that are sent by the host
system.The ESCON HA in the ESS has two
ports for ESCON channel connection.
disk 8-pack. In the ESS, a group of 7 or 8
DDMs that are not yet formatted as ranks.
drawer. A unit that contains multiple DDMs,
and provides power, cooling, and related
Glossary
267
ESCON Port. The ESCON port is the
physical interface into the ESCON channel. An
ESCON port has an ESCON connector
interface. You have an ESCON port wherever
you plug in an ESCON Link
ESCON Link. An ESCON link is the fiber
connection between the zSeries 900 (or
S/390) server and the storage enclosure. An
ESCON link can also exist between a zSeries
900 (or S/390) processor and an ESCON
Director (fibre switch), and between an
ESCON Director and the storage enclosure
(or other ESCON capable devices)
extended remote copy (XRC). A hardwareand software-based remote copy service
option that provides an asynchronous
volume copy across storage subsystems for
disaster recovery, device migration, and
workload migration.
F
Fabric Fibre Channel employs a fabric to
connect devices. A fabric can be as simple
as a single cable connecting two devices.
The term is most often used to describe a
more complex network utilizing hubs,
switches and gateways.
Fabric-Loop Ports, FL_ports. These ports
are just like the F_ports, except that they
connect to an FC-AL topology. FL_ports can
only attach to NL_ports. The ESS Fibre
Channel adapters do not support the
FL_port functionality, which is found only in
fabrics or hubs.
Fabric Ports, F_ports. These ports are
found in Fibre Channel Switched Fabrics.
They are not the source or destination of
IUs, but instead function only as a
"middleman" to relay the IUs from the
sender to the receiver. F_ports can only
attach to N_ports. The ESS Fibre Channel
adapters do not support the F_port
functionality, which is found only in fabrics.
268
Implementing Linux with IBM Disk Storage
FC See Fibre Channel
FC-AL. Fibre Channel Arbitrated Loop.
Description of Fibre Channel connection
topology when SAN fabric consists of hubs.
This implementation of the Fibre Channel
standard uses a ring topology for the
communication fabric.
FC Adapter. A FC Adapter is a card installed
in a host system. It connects to the Fibre
Channel (fibre) through a connector. The FC
adapter allows data to be transferred over fibre
links at very high speeds (currently 100 MB/s)
and over greater distances than SCSI.
According to its characteristics and its
configuration, they allow the server to
participate in different connectivity topologies
FC Host Adapter. The ESS has a FC Host
Adapter (HA). The FC Host Adapter is
connected to the Fibre Channel and accepts
the FC commands that are sent by the host
system
FCP. Fibre Channel Protocol. When we are
mapping SCSI to the Fibre Channel transport
(FC-4 Upper Layer) then we say FCP
FCS. See Fibre Channel standard
FC-SW. Fibre Channel Switched Fabric.
Description of Fibre Channel connection
topology when SAN fabric consists of
switches.
Fibre Channel. Some people refer to Fibre
Channel as the Fibre version of SCSI. Fibre
Channel is capable of carrying IPI traffic, IP
traffic, FICON traffic, FCP (SCSI) traffic, and
possibly traffic using other protocols, all at the
same level in the standard FC transport. Two
types of cable can be used: copper and fiber.
Copper for shorter distances and fiber for the
longer distances
Fibre Channel Host. A server that uses
Fibre Channel I/O adapter cards to connect to
the storage enclosure
F_Port Fabric Port - a port used to attach a
NodePort (N_Port) to a switch fabric.
Fibre Channel standard. An ANSI standard
for a computer peripheral interface. The I/O
interface defines a protocol for
communication over a serial interface that
configures attached units to a
communication fabric. The protocol has two
layers. The IP layer defines basic
interconnection protocols. The upper layer
supports one or more logical protocols (for
example FCP for SCSI command protocols,
ESCON for ESA/390 command protocols).
G
FICON. An IO interface based on the Fibre
Channel architecture. In this new interface,
the ESCON protocols have been mapped to
the FC-4 layer, i.e. the Upper Level Protocol
layer, of the Fibre Channel Architecture. It is
used in the S/390 and z/series
environments.
FICON Channel. A channel that has a
FICON channel-to-controller I/O interface
that uses optical cables as a transmission
medium.
Fixed Block Architecture (FBA). FC and
SCSI disks use a fixed block architecture, that
is, the disk is arranged in fixed size blocks or
sectors. With an FB architecture the location
of any block can be calculated to retrieve that
block. The concept of tracks and cylinders also
exists, because on a physical disk we have
multiple blocks per track, and a cylinder is the
group of tracks that exists under the disk
heads at one point-in-time without doing a
seek
FlashCopy. A point-in-time copy services
function that can quickly copy data from a
source location to a target location.
F_Node Fabric Node - a fabric attached
node.
GB. See gigabyte.
gigabyte. 1 073 741 824 bytes.
group. A group consist of eight DDMs.
H
hard disk drive (HDD). A nonvolatile
storage medium within a storage server
used to keep the data records. In the ESS
these are 3.5” disks, coated with thin layers
of special substances, where the information
is magnetically recorded and read. HDDs
are packed in replaceable units called
DDMs.
HDD. See hard disk drive.
heterogeneous. Term used to describe an
environment consisting of multiple types of
hosts concurrently (i.e. Windows NT, Linux
and IBM AIX)
homogenous. Term used to describe an
environment consisting entirely of one type of
host (i.e. Windows 2000 only)
host. The server where the application
programs run.
host adapter (HA). A physical sub unit of a
storage controller that provides the ability to
attach to one or more host I/O interfaces.
I
ICKDSF. See Device Support Facilities
program.
Glossary
269
Intel Servers. When we say Intel servers (or
Intel based servers) we are referring to all the
different server makes that run on Intel
processors. This includes servers running
Windows NT and Windows 2000, as well as
Novell Netware and Linux. This term applies to
the IBM Netfinity servers, the most recent
xSeries family of the IBM ^brand, and
the various non-IBM server makes available
on the market that run on Intel processors
Internet Protocol (IP). A protocol used to
route data from its source to its destination
in an Internet environment.
I/O device. An addressable input/output
unit, such as a direct access storage device,
magnetic tape device, or printer.
I/O interface. An interface that you define in
order to allow a host to perform read and
write operations with its associated
peripheral devices.
ITSO. International Technical Support
Organization
J
JBOD. Just a Bunch Of Disks. A disk group
configured without the disk redundancy of the
RAID-5 arrangement. When configured as
JBOD, each disk in the disk group is a rank in
itself.
JCL. See job control language.
Job control language (JCL). A
problem-oriented language used to identify
the job or describe its requirements to an
operating system.
K
270
Implementing Linux with IBM Disk Storage
KB. See kilot .iie( KB)322.0())29.7(.)-9.3( )]TJ /F2 1 Tf 9.84
logical volume. The storage medium
associated with a logical disk. A logical volume
typically resides on one or more DDMs. For
CKD the logical volume size is defined by the
device emulation mode (3390 or 3380 track
format). For open systems hosts, the size is
0.5 GB to the maximum capacity of a rank
LUN. See logical unit number
M
MB. See megabyte.
megabyte (MB). 1 048 576 bytes.
mirrorset. A RAID storageset of two or more
physical disks that maintains a complete and
independent copy of the entire virtual disk’s
data. Also referred to as RAID-1
connections. N_ports can only attach to
other N_ports or to F_ports. The ESS Fibre
Channel adapters support the N_port
functionality when connected directly to a
host or to a fabric.
non-disruptive. The attribute of an action or
activity that does not result in the loss of any
existing capability or resource, from the
customer’s perspective.
nonvolatile storage (NVS). Random
access electronic storage with a backup
battery power source, used to retain data
during a power failure. Nonvolatile storage,
accessible from all cached IBM storage
clusters, stores data during DASD fast write,
dual copy, and remote copy operations.
NVS. See Nonvolatile storage.
O
N
native Linux. An environment where Linux is
installed and operating on the native hardware
platform. This is the most common of Linux
implementations however, the distinction
needs to be drawn as the zSeries servers
allow for Linux to be run as a guest under the
z/VM (or VM) native operating system
Node-Loop Ports, NL_ports.These ports
are just like the N_ports, except that they
connect to a Fibre Channel Arbitrated Loop
(FC-AL) topology. NL_ports can only attach
to other NL_ports or to FL_ports. The ESS
Fibre Channel adapters support the NL_port
functionality when connected directly to a
loop.
Node Ports, N_ports. These ports are
found in Fibre Channel Nodes, which are
defined to be the source or destination of
Information Units (IUs). I/O devices and host
systems interconnected in point-to-point or
switched topologies use N_ports for their
open systems. (or open servers) refers to
the Windows NT, WIndows 2000, Linux, and
various flavors of UNIX operating
environments. Both for IBM and non IBM
servers
operating system. Software that controls the
execution of programs. An operating system may
provide services such as resource allocation,
scheduling, input/output control, and data
management.
P
peer-to-peer remote copy (PPRC). A
hardware based remote copy option that
provides a synchronous volume copy across
storage subsystems for disaster recovery,
device migration, and workload migration.
port. (1) An access point for data entry or
exit. (2) A receptacle on a device to which a
cable for another device is attached.
Glossary
271
PPRC. See peer-to-peer remote copy.
primary device. One device of a dual copy
or remote copy volume pair. All channel
commands to the copy logical volume are
directed to the primary device. The data on
the primary device is duplicated on the
secondary device. See also secondary
device.
PTF. Program temporary fix. A fix to a bug in
a program or routine.
R
rack. A unit that houses the components of
a storage subsystem, such as controllers,
disk drives, and power.
RAIDset. A storageset that stripes data and
parity across three or more members in a disk
array. Also referred to as RAID-5
rank. A disk group upon which a RAID-5
array is configured. For JBOD, each DDM
becomes a rank.
random access. A mode of accessing data
on a medium in a manner that requires the
storage device to access nonconsecutive
storage locations on the medium
RDAC. Redundant Disk Array Controller.
Controller failover facility provided for some
operating systems with the FAStT product line
read hit. When data requested by the read
operation is in the cache.
read miss. When data requested by the
read operation is not in the cache.
recovery. The process of rebuilding data
after it has been damaged or destroyed. In
the case of remote copy, this involves
applying data from secondary volume
copies.
recovery system. A system that is used in
place of a primary application system that is
272
Implementing Linux with IBM Disk Storage
no longer available for use. Data from the
application system must be available for use
on the recovery system. This is usually
accomplished through backup and recovery
techniques, or through various DASD
copying techniques, such as remote copy.
ReiserFS. Reiser FS is a journaling
filesystem which means logs all changes to a
disk and can play them back after a system
failure. Hence long filesystem checks like with
ext2 are obsolete .
remote copy. A storage-based disaster
recovery and workload migration function
that can copy data in real time to a remote
location. Two options of remote copy are
available. See peer-to-peer remote copy and
extended remote copy.
restore. Synonym for recover.
re-synchronization. A track image copy
from the primary volume to the secondary
volume of only the tracks which have
changed since the volume was last in duplex
mode.
S
SCSI. SCSI (Small Computer Systems
Interface) is the protocol that the SCSI adapter
cards use. Although SCSI protocols can be
used on Fibre Channel (then called FCP) most
people mean the parallel interface when they
say SCSI. This is the ANSI standard for a
logical interface to computer peripherals and
for a computer peripheral interface. The
interface utilizes a SCSI logical protocol over
an I/O interface that configures attached
targets and initiators in a multi-drop topology.
SCSI adapter. A SCSI adapter is an I/O card
installed in a host system. It connects to the
SCSI bus through a SCSI connector. There
are different versions of SCSI, some of which
can be supported by the same adapter. The
protocols that are used on the SCSI adapter
(the command set) can be either SCSI-2 or
SCSI-3
SCSI ID. An unique identifier (ID) assigned to
a SCSI device, that is used in protocols on the
SCSI interface to identify or select the device.
The number of data bits on the SCSI bus
determines the number of available SCSI IDs.
A wide interface has 16 bits, with 16 possible
IDs. A SCSI device is either an initiator or a
target
SCSI host. An open systems server that
uses SCSI adapters to connect to the storage
enclosure
SCSI host adapter. The host adapter (HA) is
the physical component of the storage server
used to attach one or more host I/O
interfaces.The ESS can be configured with
SCSI host adapters (HA). The SCSI host
adapter is connected to the SCSI bus and
accepts the SCSI commands that are sent by
the host system.The SCSI HA has two ports
for SCSI bus connection
SCSI port. A SCSI Port is the physical
interface into which you connect a SCSI cable.
The physical interface varies, depending on
what level of SCSI is supported
Seascape architecture. A storage system
architecture developed by IBM for open
system servers and S/390 host systems. It
provides modular storage solutions that
integrate software, storage management,
and technology for disk, tape, and optical
storage.
secondary device. One of the devices in a
dual copy or remote copy logical volume pair
that contains a duplicate of the data on the
primary device.
server. A type of host that provides certain
services to other hosts that are referred to
as clients.
spare. A disk drive that is used to receive
data from a device that has experienced a
failure that requires disruptive service. A
spare can be pre-designated to allow
automatic dynamic sparing. Any data on a
disk drive that you use as a spare is
destroyed by the dynamic sparing copy
process.
Spareset. A collection of disk drives made ready
by the controller to replace failed members of
a storageset.
SSA. Serial Storage Architecture. An IBM
standard for a computer peripheral interface.
The interface uses a SCSI logical protocol
over a serial interface that configures
attached targets and initiators in a ring
topology.
SSR. System Support Representative. The
IBM person who does the hardware
installation and maintenance.
storage cluster. A partition of a storage
server that is capable of performing all
functions of a storage server. When a cluster
fails in a multiple-cluster storage server, any
remaining clusters in the configuration can
take over the processes of the cluster that fails
(fail-over)
storage device. A physical unit which
provides a mechanism to store data on a given
medium such that it can be subsequently
retrieved.
storage server. A unit that manages
attached storage devices and provides access
to that storage and storage-related functions
for one or more attached host systems
Glossary
273
Striped mirrorset. A storageset that stripes
data across an array of two or more
mirrorsets. Also referred to as RAID-0+1
Stripeset. A storageset that stripes data
across an array of two or more disk drives.
Also referred to as RAID-0
striping. A technique that distributes data in
bit, byte, multi-byte, record, or block
increments across multiple disk drives.
synchronization. An initial volume copy.
This is a track image copy of each primary
track on the volume to the secondary
volume.
synchronous operation. A type of
operation in which the remote copy PPRC
function copies updates to the secondary
volume of a PPRC pair at the same time that
the primary volume is updated. Contrast
with asynchronous operation.
volume. When you store multiple volumes
on a single storage medium transparently to
the program, you may refer to the volumes
as logical volumes.
VTOC. Volume table of contents. In a
DASD, the place where the space and
allocation information of the volume is
maintained.
W
World Wide Name (WWN). unique number
assigned to Fibre Channel devices (including
hosts and I/O adapter ports) - analogous to a
MAC address on a network card.
write hit. A write operation where the data
requested is in the cache.
write miss. A write operation where the
data requested is not in the cache.
T
X
TCP/IP. Transmission Control
Protocol/Internet Protocol.
XRC. Extended remote copy.
timeout. The time in seconds that the
storage control remains in a “long busy”
condition before physical sessions are
ended.
U
Ultra-SCSI. An enhanced small computer
system interface.
Y
YaST. Acronym for Yet another Setup Tool.
YaST provides integrated user and group
administration and simplifies system
administration.
V
volume. An ESA/390 term for the
information recorded on a single unit of
recording medium. Indirectly, it can refer to
the unit of recording medium itself. On a
non-removable medium storage device, the
terms may also refer, indirectly, to the
storage device that you associate with the
274
Implementing Linux with IBM Disk Storage
Z
z/Architecture. The IBM 64-bit real
architecture implemented in the new IBM
server zSeries 900 enterprise e-business
servers.
z/OS. The IBM operating systems for the
z/Architecture family of processors.
zSeries 900. Refers to the zSeries 900
family of servers from the IBM server
brand. Also referred to as zSeries. These
servers are the successors to the IBM 9672
G5 and G6 family of processors.
Glossary
275
276
Implementing Linux with IBM Disk Storage
Related publications
The publications listed in this section are considered particularly suitable for a
more detailed discussion of the topics covered in this redbook.
IBM Redbooks
For information on ordering these publications, see “How to get IBM Redbooks”
on page 278. Note that some of the documents referenced here may be available
in softcopy only:
򐂰 IBM TotalStorage Enterprise Storage Server Model 800, SG24-6424
򐂰 IBM TotalStorage Enterprise Storage Server: Implementing the ESS in Your
Environment, SG24-5420
򐂰 IBM TotalStorage Enterprise Storage Server Web Interface User’s Guide,
SC26-7448
򐂰 Fibre Array Storage Technology - A FAStT Introduction, SG24-6246
򐂰 IBM TotalStorage FAStT700 and Copy Services, SG24-6808
򐂰 Linux for S/390, SG24-4987
򐂰 Linux for IBM zSeries and S/390: Distributions, SG24-6264
򐂰 Linux on IBM zSeries and S/390: Large Scale Linux Deployment, SG24-6824
򐂰 Tuning IBM eServer xSeries Servers for Performance, SG24-5287
© Copyright IBM Corp. 2001 2003. All rights reserved.
277
򐂰 IBM Storage Area Network, SG24-6419-00 , SG24-5420
򐂰 The Cutting Edge: The IBM eServer BladeCenter, REDP-3581
򐂰 Getting Started with zSeries Fibre Channel Protocol, REDP-0205
Other publications
These publications are also relevant as further information sources:
򐂰 IBM TotalStorage Enterprise Storage Server Web Interface User’s Guide,
SC26-7448
򐂰 IBM TotalStorage Enterprise Storeage Server: Subsystem Device Driver
User’s Guide, SC26--7478
Online resources
These Web sites and URLs are also relevant as further information sources:
򐂰 IBM Linux Web site
http://www.ibm.com/linux/
򐂰 The Linux Technology Center
http://www.ibm.com/linux/ltc
򐂰 The IBM TotalStorage Web site
http://www.storage.ibm.com/
򐂰 The IBM TotalStorage SAN fabric Web site
http://www.storage.ibm.com/ibmsan/products/sanfabric.html
򐂰 The IBM eServer Web site
http://www.ibm.com/eserver
How to get IBM Redbooks
You can search for, view, or download Redbooks, Redpapers, Hints and Tips,
draft publications and Additional materials, as well as order hardcopy Redbooks
or CD-ROMs, at this Web site:
ibm.com/redbooks
278
Implementing Linux with IBM Disk Storage
Index
A
access LUN 188
AIX 75
array 21, 199, 202
Automatic Discovery screen 193
B
BladeCenter 97–98, 115, 126
Fibre Channel Expansion 128
Fibre Switch Module 137, 184
issues 123
requirements 100
SAN Utility 137, 177
switch module 178
boot
disk 114, 116, 118
parameters 121
boot loader
GRUB 108–109, 215
LILO 108, 130, 232
BSD 3–4
215, 227, 239, 255, 260
(see also SDD)
DHCP 83, 119
disk
boot 118
driver 116
partition 231, 255
Disk Druid 105–106
driver (see device driver)
drvrsetup 127
E
cache 14
CD-ROM 116–117, 124, 246, 255
cfgvpath 132, 134
channel path identifier 28
cloneconfig 128
cluadmin 149
clustat 152
Cluster Manager 139
install 144
log file 143
service 148
control unit address 28
ECKD 26, 28, 32, 164, 169
Emulex 76, 86
device driver 86, 93
multipath driver 76
ESCON 27
ESS 7, 12, 18, 127, 130, 137
cache management 14
copy services 15, 157
disk groups 166
host adapter 163, 168
LIC 160
LIC level 156
Linux host 166
Master Console 156
performance 14
Specialist 65, 137, 156
client 158
Web Browser 157–158
storage allocation 161, 165
storage consolidation 14
target host 175–176
volume 156, 164, 166, 171–174, 176
ext2 258
ext3 258
D
F
DASD 26, 32, 52, 55, 58–59
DDM 13
depmod 129–130
device address 29
device driver 28, 71, 99–100, 118, 126, 130, 188,
failover 238
FAStT 7, 18, 127, 129–130
controller 189, 191
firmware 191, 195
host 205, 207
C
© Copyright IBM Corp. 2001 2003. All rights reserved.
279
host port 205
Management Suite for Java 137, 191
MSJ 100, 137, 211–212
NVSRAM 195
Performance 19
Remote Volume Mirroring 22
Storage Manager 187
agent 191
client 191
Storage Partitioning 22, 198, 201, 203, 207
FAT 254, 258
FC 18
FCP 16–17, 25, 29–30, 33, 35, 38, 64, 67, 163, 167
addressing 29
mapping 31
fdisk 68, 105, 136, 256, 261
Fibre Channel 125
FICON 16–17, 27, 32, 35, 162–163
File Allocation Table (see FAT)
file system 105, 108, 119, 253, 255
disk letter 254
mounting 262
root 136
VMFS 227
filesystem 256, 258
FlashCopy 16, 22
floppy 116, 123, 254–255
VMware
floppy 246
fylesystem
root 105
Integrated Facility for Linux 34
ips 129
ISO image 246, 250
J
JFS 258
journalling 258
K
KUDZU 125
L
LILO 108, 130, 232–233
Linux
distributions 5
load balancing 220, 222
log file 143, 180
logical drive 198–200, 208
Logical Volume Manager. See LVM
LPAR 90
pSeries 74–76, 79, 90, 92
zSeries 26–27, 34, 38, 46, 51
lsmod 133
lsvpcfg 132, 135
LUN 29, 31, 65–67, 129, 131, 162, 172, 174, 198,
200, 208, 212, 242, 260
access 188, 198
configure 214, 219
masking 22
number 256
LVM 71, 130, 259
G
GNU 3–5
GRUB 108–109, 215
H
HBA 20, 100, 119, 125–126, 130, 212, 218, 240,
260
heartbeat 145, 148
Hipersocket 36
Host Bus Adapter. See HBA
hyper-threading 121–122
I
insmod 133
insserv 135
280
Implementing Linux with IBM Disk Storage
M
Management Suite for Java (see MSJ)
Master Boot Record. See MBR
MBR 108, 117, 232
menuconfig 128
mk_initrd 130, 222
mkinitrd 215, 222
modprobe 124, 129–130
mount point 105, 108, 150, 257, 262
mouse 103, 121, 124
MSJ 100, 137, 211–212
install 212
multipath 188, 212, 215
multipath driver 97, 126–127, 130
N
NMI 141–142
NVSRAM 195
O
Open Source 1, 4–5
open source 4
filesystem 105
partition 231
user 89, 110, 119, 135, 192, 212, 233
root device 255
root partition 257
RVM 21–22
S
P
package 88
selection 112, 118–119
package selection 59
partition 58–59, 87, 105, 118, 136, 256, 261
root 257
swap 257
path
configure 216, 220
physical link address 28
port
configure 219
PPRC 15
pSeries 73
requirements 75–76
PXE 115
Q
qlremote 212, 216–217, 221–222
Quorum partition 140, 145
R
RAID 15, 18, 21, 141, 169–170, 199, 202, 227
ramdisk 40–41, 46, 70, 84, 129–130, 216, 221, 255
rawdevice 142
RDAC 188, 191
Red Hat 6–7
BladeCenter 102, 115, 122
boot loader 108
Cluster Manager 139
High Availability Cluster Manager 139
Summit kernel 121
xSeries 102
Redbooks Web site 278
Contact us xviii
reiserFS 258
rmvpath 132, 134
root
file system 136
SAN 16, 125
SAN Utility 137, 177
browser 180
Faceplate view 185
install 178
log file 180
SCSI 29, 68, 115, 129, 136, 163, 227, 237, 255,
259–260
device address 29
SDD 125, 130–132, 136
sdd start 132, 135
sdd stop 132, 135
Seascape 12
SSA 14
device adapters 163
loop 164
loops 163
ssh 38, 45, 238
Storage Manager 21–22, 76, 100, 187
agent 188, 190
client 188
in-band 188, 190
out-of-band 188, 191
Subsystem Device Driver (see SDD)
Summit kernel 121–122
SuSE 6–7
BladeCenter 121
pSeries 73
xSeries 117
zSeries 25, 36
swap 257
switch module 178
switch module (See BladeCenter-Fibre Switch Module)
System Management Services 79, 81
T
telnet 38
TotalStorage 11
Index
281
U
unit address 28
UnitedLinux 6–7, 75
YaST 51, 85, 118
YaST2 87
Z
V
VMware 225
boot configuration 237
CD-ROM 246
configure ESX server 234
console OS 231–233, 237–238
dump 242
ESX 226
architecture 227
GSX 226
installation 230
network 246
remote console 247
requirements 228
resource manage 227
virtualization layer 227
VM boot 250
VM monitoring 252
VM wizard 245
VMFS 227, 232, 242
VMkernel 239
Workstation 226
vmware-console 248
VNC client 38, 45, 51, 76, 78, 83, 85
VNC server 84
volume 254
vpath device 135–136
W
watchdog 141, 145, 153
NMI 141–142
Worldwide Port Name (see WWPN)
WWPN 167
X
X server 38, 45, 62
xfs 258
XRC 15
xSeries 97, 126
requirements 99–100
Y
yaboot 82
282
Implementing Linux with IBM Disk Storage
zSeries 14–15, 25–26
multipath 72
requirements 33–34, 37
z/VM guest 26, 40
Implementing Linux with IBM Disk Storage
(0.5” spine)
0.475”<->0.875”
250 <-> 459 pages
Back cover
®
Implementing Linux with
IBM Disk Storage
Your guide to Linux
implementations on
IBM eServers
Use Linux efficiently
with FAStT and ESS
Explore options
relating to Linux in
SANs
This IBM Redbook explains considerations, requirements,
pitfalls, and possibilities when implementing Linux with IBM
disk storage products. This redbook presents the reader with
a practical overview of the tasks involved in installing Linux on
a variety of IBM Eserver platforms, and it also introduces
the people who are already familiar with Linux to the powerful
IBM TotalStorage servers: ESS and FAStT. The book also
provides the steps required to prepare the storage system for
Linux host attachment.
Among other things, we review the recently added support for
FCP attachment when running SuSE Linux (SLES 8) in an
LPAR on a zSeries machine.
On pSeries we cover the implementation of SLES 8, either
natively or in an LPAR on a p690, and explain how to configure
the Emulex Host Bus Adapter, for attachment to FAStT and
ESS.
For Intel based machines, we look at the implementation of
Red Hat Enterprise AS 2.1 and SLES 8 on xSeries servers and
BladeCenter; in addition, we explain how to set up the Red Hat
High Availability cluster using shared external storage.
The book also contains a chapter on the VMware ESX Server,
focusing on the practical aspects of its implementation and
configuration with external storage.
INTERNATIONAL
TECHNICAL
SUPPORT
ORGANIZATION
BUILDING TECHNICAL
INFORMATION BASED ON
PRACTICAL EXPERIENCE
IBM Redbooks are developed by
the IBM International Technical
Support Organization. Experts
from IBM, Customers and
Partners from around the world
create timely technical
information based on realistic
scenarios. Specific
recommendations are provided
to help you implement IT
solutions more effectively in
your environment.
For more information:
ibm.com/redbooks
SG24-6261-01
ISBN 0738453471
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement