I am trying to determine if my current strategy is the best way to handle clusters or if someone else has a better way. Here's what I have today: PhysicalNodeA PhysicalNodeB VirtualA (J: Volume assigned) VirtualB (X: Volume assigned) QuorumNode
I have on policy that backups up PhysicalNodeA and PhysicalNodeB. The file list is C: and System State. I have another policy that backs up the QuorumNode and the file list is Q: (quorum disk). I have a 3rd policy that backs up VirtualA and the file list is J: and a 4th policy for VirtualB with file list as X:.
If I create one policy for the two physical nodes and backup ALL_LOCAL_DRIVES, what am I missing? The problem I have now is that Engineers go out and create either a new virtual node or assign another volume to an existing virtual node and I don't hear about it until someone wants a file restored. Also, there's no easy way to verify coverage. Any one doing things differently?
We basically do the brute force ALL_LOCAL_DRIVES for each physical node Seems to get everythinng. The only issue occurs when the nodes failover, in which we sometime backup MORE than we need, which we live with. Better than trying to micro-manage it all.
I totally agree that this seems like the easiest solution. In the event of a complete cluster meltdown, am I going to be missing any pieces to the puzzle when it comes time to restore the cluster from scratch?
I believe you would be covered (although we have never had a complete cluster meltdown ourselves)
I have looked this over and it seems like we would have everything: http://www.microsoft.com/technet/prodtechnol/windowsserver2003/technologies/clustering/sercbrbp.mspx
There's also a coverage tool built into NBU that might help you check coverage as well. It's pretty slow, but might expose some things you are missing and didn't know about!
You should find it under installpath\NetBackup\bin\goodies. It's called "check_coverage.cmd". In there you'll find the bpcoverage command that they use to check coverage.
Here, http://ftp.support.veritas.com/pub/support/products/NetBackup_Enterprise_Server/268188.pdf Page 74-75, there is a couple of policies to set-up, with local disk, shared disk, virtual node...
Thanks Julien, this is how we currently have our backups configured. Although it's reassuring to know I'm doing it the Veritas recommended way, I was hoping for a configuration that required less management. But I'm thinking I better leave well enough alone and go with the recommended configuration.
Keeping it the veritas recommended way is a good idea. This configuration avoid code 71 when backing-up ALL_LOCAL_DRIVE with multiple data stream and get the 71 with the qorum on the inactive node.
