How to stop VCS from acting?

[eu22106]

Hi all,

I want to change diskconfiguration. (Adding disks to a mirror and than breaking mirrors) but don't want to stop the ServiceGroups. So while VCS is up (and all SQL instances) I want to change the diskconfiguration.

Is that possible and how?

Ivo

[Wally_Heim]

Hi eu22106,

Yes, you can open VEA and add disks to a mirror with the volume being live and the cluster completely unaware of it.  There is nothing that you really need to do from the cluster (VCS) side of things to do this.  The VCS cluster is monitoring that the disk group is imported, the volumes are available and that the volumes are mounted where expected.  Altering mirror settings of a volume does not alter any of the things that the cluster is monitoring for to determine if the VMDg and MountV resources are online.

However, if you are overly concerned, you can freeze the service group temporarily to perform the change and then unfreeze the service group when done.  In Cluster Manager you right click on the service group name and select freeze (temporary or permenant.)  To unfreeze you right click on the group and select unfreeze. 

Freezing a service group makes is so that the cluster does not react for resource faults in the service group.  Temporary freeze will freeze the group but will be automatically reset during a reboot of the node.  A permenant freeze will keep the group frozen even after the cluster nodes have been rebooted.

Thanks,

Wally

[eu22106]

Adding things is not an issue, but removing is where I don't succeed.

Somebody messed up the configuration while expanding the volumes. We usually have two disks (mirrored. However now we have a situation where multiple DG are using the same disks.
e.g
DG1 disk1 (10Gb), disk2(10Gb)
DG2 disk1 (10Gb), disk3 (5Gb)
DG3 disk3 (5Gb), disk4 (5Gb)

What I was planning to do, was to add 2 new disks (to all the DG's) as mirror, than remove disk1 to 4.
Unfortunately I now get errors like ("disk is not empty" when removing disk from DG or "volume is configured as a VCS resource" when splitting the volume)

[Wally_Heim · Accepted solution]

Hi eu22106,

It sounds like you are referring to internal disk names and not actual harddisk numbers.  You cannot have one disk that belongs to multiple disk groups.  However, each disk group can and does having an internal disk named "Disk1". 

For example, "Disk1" for DG1 maybe harddisk1 and "Disk1" for DG2 maybe harddisk2.  However, harddisk1 can not be in both DG1 and DG2 at the same time. 

I would not start to make drastic changes to the disk configuration until you have a good understanding of what is actually configured and what your end goal is.

The output of vxprint or a couple of screen shots of VEA would help us to understand you disk group configuration better.

Thanks,

Wally

[eu22106]

Sorry for this late reaction....... (holiday, ill and what have you.)

We were able to solve this. We added clean disks to the volumes, added a mirror and deleted the "old" disks. Deleting was done on SAN-level. Now the instances kept running.