Forum Discussion

sdo's avatar
sdo
Moderator
7 years ago

Catalog recovery, appliance to own build...

Is it supported to perform a catalog recovery from an Appliance based catalog backup to a new own build master?  Assumng I use the same mount-point folder names, then I can't see what would stop this from being a supported customer side action.

My thinking was that a catalog backup is a catalog backup, and that it if stick to same OS family/type as the Appliance, then I should be able to.

From: Master/Media 5230, four shelves, v2.7.3, on RHEL 6.6

To:     Master (VM on Hyper-V 2012 R2), v7.7.3, on RHEL 6.5

.

Any detractors, gotchas, warnings, lessons?

5 Replies

  • Linux to Linux shouldn't be a problem.

    So Appliance to Linux will work.  Just keep the Master name the same.

  • You can even do it with symbolic links that simulates Appliance path. The only gotchas that I see is the Advance Disk and MSDP pool. Since you cannot attach those two to the new master, best to empty out both disk pools and remove from config before final catalog backup.
    • sdo's avatar
      sdo
      Moderator

      Thanks eduncan - yes name will remain same.

      Marianne - thank you as ever, great point as usual... my thinking is that want to keep the original 5230 as pristine as possible as my roll-back point (I will get networks team to permanently down/down the LAN switch NICs) (but leave IPMI) at the appropriate time... so... now I can't help but ask whether you know that I should remove the defunct (after all backups have expired) MSDP disk pool and storage server entries because if I don't then my restored master on a VM will be unusable?  I mean, I was going to remove the defunct storage unit and disk pool and storage server after the catalog recovery.

      • Marianne's avatar
        Marianne
        Level 6
        Hopefully you can test the procedure in a lab environment first.
        I seem to remember horror stories of how people tried to clean up with disk being removed without following all the recommended steps. In the end they had to log support call for SQL scripts to remove orphaned entries...