Cluster server issue

[BVshet]

Hi Team,

We have a SFWHA NBU cluster on windows 2008 R2 systems working on NBU 7.6.0.3.

Due to some issues we have lost the primary node and are now dependent on passive node.

Is it possible for me to rebuild the loast (Prev active node using BMR) 

Will that help.

If not we are planning to build a standalone server and import catalog can you please advice if I can import cluster master catalog into a standalone server with hostname of cluster.

Please help.

Bshet.

[Marianne]

None of your plans will work.

You cannot protect or restore a master server using BMR.

You cannot restore NBU catalog to a standalone server. 
Why would you want to? The catalogs are on shared storage and online on failover node.

Please use your MS cluster documentation for instructions on replacing a failed node.

Once the replaced node has joined the cluster, ensure that this new node is zoned to the shared storage.

When you install NBU and specify the virtual hostname as master server name, you will be presented with option to join existing NBU cluster group.
(See NetBackup Clustered Master Server Administrator's Guide  )

If you are unsure of steps to follow, rather log a Support call with Veritas.

 

[Michal_Mikulik1]

Hello,

it is an interesting question - as for cluster-to-standloane catalog recovery, I also think it is not supported.

As for BMR, I think that you could use BMR procedure for a failed node recovery, if you have a valid BMR-based backup of this node.

Because now, you have functional Master Server, thus you can consider the failed node as a common client which needs recovery.

Regards

Michal

[BVshet]

Hi Marianne/Michal,

Thanks for the inputs, my cluster is hosted on a Veriats storage foundation cluster.

The major concern I have is my primary node is down and not the NBU is working on secondary node.

Whats the procedure to join the primary node back to cluster.

Please advice.

Bshet

[Michal_Mikulik1]

Hello,

if you have a valid BMR based backup of the failed node, you can use BMR procedure, and after this, node is simply joined. Actually, I have tested this with MS clusters, not VCS cluster yet, but it worked for them.

If you dont have valid BMR backup, you should follow some VCS/SF steps - but that would be more effective to solve these steps on appropriate VCS/SF forum.

regards

Michal

[sdo]

Me thinks BMR recovery of what was previously a cluster member node is fraught with danger.

If the previous catalog LUN is still zoned / shared / presented / visible to the recovering node, in fact any previously shared LUN, then there is a very possible danger of it/them being formatted / overwritten / corrupted.

If the previous BMR aware backup of the previously clustered member node has, within the backup set, anything that was backed-up from a volume that was a cluster shared resource... well, isn't BMR going to want to recover that too?  And you cannot stop BMR from wanting to attempt to do that.

Plus VxFS on VxVM on VxDG on VxDMP all come in to play.

.

IMO, I'm with Marianne.  I wouldn't even suggest what you are about to do as having the remotest possibility of success.

[RiaanBadenhorst]

Why are you exploring the most complex solutions first? Just install windows on a new node, and it to the cluster, and then reinstall netbackup on it.