December 8, 2011, in category LPAR Configuration comments No comments

How to upgrade an LPAR to a new version of AIX using a mksysb

Checklist

LPAR Information

VIO Information

Backup SSH Files

• copy ssh file to new location

   cp –r /etc/ssh /etc/ssh.backup 

Map disks from VIO

• If new disks are being allocated for additional storage or JFS to JFS2 Conversion, you will need to map the disk through the VIO server
• Once storage is allocated through SVC, login to VIO server
• detect new hard ware

   cfgdev 

• confirm LUN ID with disks allocated by storage team

   lsdev –dev hdiskx –vpd 

• create virtual device on NIM

   mkvdev –vdev hdiskx –vadapter vhostx –dev [vtd] 

• view all devices mapped with specified vhost

   lsmap –vadapter vhostx 

Confirm Disks on LPAR

• review attributes

   cfgmgr 

   lsattr –El hdiskx 

• In order to maximize consistency between redundant VIO servers, I prefer an hcheck interval of 60 and a vscsi1 path priority of 2

 chdev -l [hdiskx] -a hcheck_interval=60

chpath -l [hdiskx] -p vscsi0 -a priority=2
 

Backup rootvg

•  mount –o soft [nim server]:[mksysb image location] [mount point] 

•  /usr/bin/mksysb –i –X –p [full path including new mksysb image name] 1>[log file] 2> [err log file] 

Backup inittab, disable all non-OS processes from booting up to ensure minimum reboot time

•  cp /etc/inittab /etc/inittab.backup 

• add : to comment out in inittab.  This is important because you will initiate multiple reboots and starting up all the processes and applications can be both a waste of time and may not shut down properly (if you force a shut down etc).  It is also important to note that you DO NOT want to simply restore the inittab after installation as the os upgrade may have added entries to the inittab.  The backup is simply a precaution, and you always want to only uncomment out what you added in

   vi /etc/inittab 

Check for currently running processes not owned by root

•  ps –ef 

From NIM, ensure LPAR ready for BOS installation

• Smit nim_bosinst -> rte install -> Version Images/Spot

i.      Accept Licenses -> Yes

ii.      Initiate Reboot -> No

iii.      Accept Licenses -> Yes

• Ensure Client State is Ready for Installation

   lsnim –l [hostname] 

Re-Boot LPAR from HMC

• Shutdown LPAR (I like to do this from the HMC sending an OS level shutdown to ensure all processes closed nicely
• Activate Profile -> Open Terminal -> Advanced -> Boot Mode -> SMS Mode

Upgrade AIX on LPAR

• 2) Remote IPL -> 1) Ethernet -> 1) IPv4 -> 1) BOOTP -> 1) IP Parameters *Populate if necessary* ESC) Back -> 3) Ping Test -> 1) Run *If failed reconfirm IP or contact networking*
• 5) Boot Option -> 1) Select install/boot order -> 6) Network -> 1) BOOTP -> 1) LAN -> 2) Normal Boot -> 1) Yes

On Boot

1) Enter -> 2) Custom Settings -> Ensure Enterprise Edition selected

• At ~3% you will need to confirm settings etc.

Once Booted

• Confirm updated successfully

   oslevel –s 

• Check error reports

   bootlist –om normal 

   errpt | more 

• Check JFS or JFS2 

   lsvg –l [vg] 

Restore SSH Key

•  mv /etc/ssh /etc/ssh.old 

•  mv /etc/ssh.backup /etc/ssh 

If running JFS2 Conversion

• If splitting rootvg over 2 disks, install bos.alt_disk_*7.1.1.0
• Copy old rootvg to new JFS2 disk

   alt_disk_copy –T –d hdiskx 

• Set new rootvg

   alt_rootvg_opt –X 

• confirm boot device

   bootlist –om normal 

If creating additional VG

• smit vg -> Add VG -> Scalable
  • Change PP Size
  • Physical Location
• confirm creation

   lsvg 

Create jfs2log LV on datavg

• make LV

   mklv –t jfs2log datavg 1 

• format LV

   logform /dev/loglv0x 

If migrating file systems off of rootvg to new VG

• Backup /etc/filesystems

  •  cp /etc/filesystems /etc/filesystems.backup 

• unmount

   unmount [file system] 

  **Subdirectories must be mounted first**

• copy LV to new VG

   cplv –v [destination vg] [lv name] 

• update file system file

   chfs –a dev=/dev/[lv] –a log=/dev/[new jfs2 log] [file system] 

• mount newly created file system

   mount [file system] 

• remove old LV

   rmlv [old lv] 

• Repeat for chosen file systems

If migrating rootvg to new disk

• list disks

   lspv 

• allow VG access to new disk

   extendvg rootvg [new disk] 

• mirror disks

   mirrorvg rootvg [new disk] 

• add boot information to new disk

   bosboot –ad [new disk] 

• set system to boot from new disk

   bootlist –m normal [new disk] 

• clear boot record from old disk

•  chpv –c [old disk] 

• unmirror

   unmirrorvg rootvg [old disk] 

• remove vg off old disk

   reduce rootvg [old disk] 

• remove old disk

   rmdev –dl [old disk] 

• confirm old disk removed

   lspv 

• confirm boot device

   bootlist –om normal 

Create paging space (usually 1:1.5 memory)

• check current paging space

   lsps –a 

• check amount of memory

   lsattr –El  mem0 

• smit lv -> Paging
  • Either add space to rootvg paging space or create new LV on new VG
• confirm added page space

   lsps –a 

If you are left with disks no longer needed, delete from LPAR

• list disks

   lspv 

• delete record of disk from LPAR

   rmdev –dl [disk] 

• delete VG from ODM

   exportvg [vg] 

If you are left with disks no longer needed, delete them from VIO servers

• double check LUN ID of old disks

   lsdev –dev hdiskx –vpd 

•  lsmap –vadapter vhostx 

• remove VTD

   rmdev –dev [vtd] 

• remove backing device

   rmdev –dev hdiskx 

• Send email to Storage to confirm disk ready for retrieval