vgchange

[Paul Barmettler]

Hi,
We are running a two node campus cluster with mirrored volumegroups on XP256/512 in different datacenters. For a desaster test we would like to shut down one of the datacenters.
Now my question: can I activate the vg's (vgchange -a e -q n vgxx) without the missing disks on the remaning node? (cluster lockdisks are available)
Thanx

[Stefan Farrelly · Accepted solution]

If you are sure you have the HP product which mirrors the 2 XP frames between datacentres then yes, you will be able to activate the volume group if the entire other datacentre is down.

The question is will you need another command first, to tell the XP's that you are accessing the VG locally rather than remotely. EMC's equivalent is called SRDF, and if we lose the remote datacentre we have to do an SRDF command first - to pull the remote VG's to our local HP box, then activate the volume group.

[Carsten Krege]

When you are running Campus Cluster where you are using diskarrays mirrored with MirrorDisk/UX on each other, you must use the "vgchange -a e -q n" command, as exactly one half of your disks is missing after the failure of one datacenter.

Stefan gave a useful hint that you would need to perform further actions if you use replication technology like Continous Access (=SRDF on XP arrays).

Carsten