Operating System - HP-UX
1836590 Members
1550 Online
110102 Solutions
New Discussion

Disk replacement in MC -Service Guard cluster-11.09(OS 11.00)

 
abiisharma
Occasional Contributor

Disk replacement in MC -Service Guard cluster-11.09(OS 11.00)

I have a shared vg, of which 2 non mirrored disks are gone bad and that need replacement. Server OS is HPUX 11.00. I am looking for a adequate plan to perform this activity. Will Cluster be halted during this activity or only package will go down? MC Service guard version is 11.09.

 

pvchange -a N doesnot work on it. (because its 11.00).

 

Please suggest.

1 REPLY 1
Matti_Kurkela
Honored Contributor

Re: Disk replacement in MC -Service Guard cluster-11.09(OS 11.00)

If one of the failed disks is used as a cluster lock disk, you may have to halt the entire cluster. Otherwise, only the package needs to go down.

 

Run "cmgetconf" to see the current cluster configuration. The cluster lock is identified with keywords FIRST_CLUSTER_LOCK_VG and FIRST_CLUSTER_LOCK_PV.

 

As the disks are not mirrored, any data on the disks that is newer than your latest backup will be lost.

 

The plan would be basically:

- halt the package (or the cluster if necessary)

- make sure the package filesystems are unmounted and VGs deactivated (the package halt operation may fail as disks are bad)

- replace disks

Note: if the new disks are bigger than the old ones, it might be good to rmsf the old devices before replacement and use insf to re-create device files for new disks. This should make sure the ioconfig has updated disk sizes.

- confirm that new disks are CLAIMED in ioscan and show the correct size with "diskinfo -v"

- vgcfgrestore (or if the new disks are bigger than the old VG can handle, you might want to re-create the entire VG with a larger pe_size and max_pv)

- if you had to halt the cluster, restart it. Do not start the package yet; if AUTO_RUN is enabled for the package, it will attempt to auto-start on cluster startup and will fail since package filesystems are not there.

- activate the package VG manually (vgchange -a e since this is a shared VG)

- mkfs the filesystem(s) on the replaced disks

- mount the filesystems, restore data from backups

- unmount the package filesystems, deactivate the package VG (vgchange -a n)

- start the package

- make a plan to upgrade to supportable hardware/software versions, and to donate these old war horses to a museum :-)

MK