Cluster files get excluded from quorum backups

dreamz

ADSM.ORG Member
Joined
Dec 30, 2008
Messages
162
Reaction score
1
Points
0
Dear Friends,

We recently configured 2 windows 2008 clustsers and setup TSM cluster schedulers like SQL and Quorum. Both of them appear to work fine. Though the backup status from Q eve is completed, one of the boxes is reporting the issue below in Activiy log during backup of quorum. :

2010-01-10 20:00ANE4987EError processing '\\plspdb00\q$\Cluster\0.hive': the object is in use by another process (SESSION: 59732)2010-01-10 20:00ANE4987EError processing '\\plspdb00\q$\Cluster\0.hive.LOG1': the object is in use by another process (SESSION: 59732)

The other box reports a "completed" status in the report and no error like above is seen in act log.

could anyone pls advice on what could be wrong here? i gave an explicit include statement in the dsm.opt file of quorum and also added the lines for VSS open file support.

As always, thanks for your time and help in advance:)
 
I don't backup the quorum disk since the data on it changes and will be useless for any restores.

Even with open file support, there are files that cannot be backed up once 'held' by the system.
 
You have no problems restoring without the Quorum drive? I never back it up either, I seem to remember somewhere long ago making me think it wasn't necessary, but I can't find that info. Anyway, I never back it up and now we're in a DR test and the Windows team said it was required. We pulled it from the system back home since we have access, but without it they said they couldn't restore.

You have some info to the contrary that I can look to for more guidance on this?
 
Hi GregE,

we created a separate VLAN and building up cluster in the test environment. we are trying to avoid TSM quorum backups for building cluster and see if that works. i have an answer hopefully in a day or two.
 
We backup the quorum but have found it to be not required for a recovery of a 2003 MS Cluster as the data is backed up as part of the system state on one of the nodes. Once you have recovered the node you should be able to start the clustter service with the /fixquorum switch and then use the clusterrecovery.msi tool to replace the disk resources and the quorum disk. The data that resides on the quorum volume will then be recreated from the copy on the node that you have recovered.
 
We backup the quorum but have found it to be not required for a recovery of a 2003 MS Cluster as the data is backed up as part of the system state on one of the nodes. Once you have recovered the node you should be able to start the clustter service with the /fixquorum switch and then use the clusterrecovery.msi tool to replace the disk resources and the quorum disk. The data that resides on the quorum volume will then be recreated from the copy on the node that you have recovered.

Greg,
we found the samething as told by andyj above. Our Windows SME didnt completely rule out the requirement to backup the quorum (with a fear of boomaranging in the event of a failed restore:)) but also told that we need not worry about this and it's good to have quorum backups which may help in case of worst possible situation.
 
Thanks. So how do you set quorum for backup since it can be owned by all systems in the cluster? Do you create a separate TSM node for it and store dsm.opt on the Quorum drive itself so that it "floats" between cluster members?
 
Thanks. So how do you set quorum for backup since it can be owned by all systems in the cluster? Do you create a separate TSM node for it and store dsm.opt on the Quorum drive itself so that it "floats" between cluster members?

As I posted a while back, backing up the quorum is useless.

I have done a lot of restores and the data in the quorum backed up. The system just restarts and recreates a new one even if you have a the quorum data in there.
 
Last edited:
I know, but I have Windows admins doing a proof-of-concept restore of a cluster and are insisting that we backup the quorum so they can see if it's needed. If they come back and say they need it, I'll have to establish it's backup on a schedule with the rest of the TSM nodes related to this cluster. Hopefully will find it's not needed.
 
Last edited:
I know, but I have Windows admins doing a proof-of-concept restore of a cluster and are insisting that we backup the quorum so they can see if it's needed. If they come back and say they need it, I'll have to establish it's backup on a schedule with the rest of the TSM nodes related to this cluster. Hopefully they can see it's not needed.

Then backup the quorum as part of the cluster shared resources.
 
I've created a separate node, with only Q: as the domain.
 
I think that each node in the cluster has it's own copy of what is on the Quorum volume which is held in c:\windows\cluster\. When we do a DR test we use the the proceedure below to start the cluster service.

http://www-01.ibm.com/support/docview.wss?uid=swg1IC45961

From the url you can see that VSS backs up the Quorum DB but restores it to \Cluster_Backup. If you re-instate the CLUSDB from the backup using the 'Local Fix' Section - Make sure that you set the Cluster Service to Manual and reboot otherwise you will not be able to rename the old CLUSDB.then you will fine that the cluster service will start with the /fixquorum switch allowing you to carry out what I suggested in my last post

What I don't know is, as I have said above, is this only the case for the node that currently owns the Quorum Group when the backup is done or is it the same for all nodes in the cluster.

Put it this way... 2 weeks ago we did a DR test of our file server cluster (3 nodes) and we just happen to pick the 1st node to restore but we didn't pick it because it was the owner of the Quorum at the time of the test. It worked fine and we were able to get a sign-off for the recovery
 
GregE,

Yes, i too have setup all quorum groups in the same way, assign a different nodename and creating a cluster scheduler..

andyJ, moon-buddy,

we did perform a restore recently without using quorum backup from TSM with what was found by you, localfix thing.

But still our Windows admin prefered to have quorum backups. So just left them as they are after a good no of debates.
 
Hi,

I have left my quorum backups enabled as wellbut as I said previously we have never used them.

Was your restore test successful?
 
Yes. It was. we restored the SQL cluster successfully. One point to mention is, we made it a single node cluster not active/passive during the restore.
 
I lose. Windows admin restored quorum in DR testing and says we need to start backing it up.
 
Back
Top