Restore Netbackup tapes from standalone to clustered instance with same name

[vazkr]

Hello! Hope this one is easy? Is it possible to restore a tape from netbackup server 7.x (standalone, non clustered) to netbackup 8.0 (new, clustered instance but same name). I understand catalogs can't be replayed  from standalone to clustered instances and need professional services to manipulate the same but what if I want to restore a tape or two. I always have the offline non clustered server that I can power up but ideally looking to restore random tapes once a new NBU cluster is deployed? I'm aware one has to import the tapes (if no catalog) but need confirmation if a new, clustered (and higher version) server can create the necessary catalog entry from a tape created in a standalone netbackup server and if yes, need to also understand what the conditions/caveats are if possible? 

On a related note, I suppose we need to uninstall and reinstall the remote clients/agents (or will push upgrade work?) as the new master/media server (non-enterprise netbackup edition) is now clustered. 

Many thanks in advance!

[RamNagalla · Accepted solution]

import will work fine, you need to follow the steps to perfrom phase 1 & 2 imports

https://www.veritas.com/support/en_US/article.TECH43584

Netbackup clients & agents need not to update as long as they are in supported version, however you can upgrade if you want to.

to make the successfull backups, client bp.conf needs to be updated with new master server & media server names, does not need any reinstallation.

[Marianne]

'Normal' backups (database and/or filesystem) is exactly the same on clustered or non-clustered master server.

The server writing the backups is then a 'normal' media server. There is nothing on a backup tape that indicates anything about the master. All that is written on the tape is header info and backup data (tar files). Not even media server name is written on a tape.

When you import in the new environment, you choose the media server. 
The media server sends catalog info to the bpdbm on the master (clustered or non-clustered).

The master server writes media server name that performed the import in the catalog header.

Restore is now the same as any other restore.

[vazkr]

Thanks, Ram/Marianne. Couple of points.

Since master/media server is same in NBU server (non enterprise) that supports clustering in Windows (not UX) so as long the server name remains the same, I can remotely upgrade the agents/clients to NBU 8, if I understand correctly?

Also, I'm using a different LTO library to import. I noticed a few users reported errors in importing although that maybe an environmental variable, so although a prefix is added, using the same or different library doesn't really make a difference.

[Marianne]

Seems I misunderstood what you are trying to do.
What exactly does this mean?
"... as long the server name remains the same... ".
Do you want to recover master server catalog from standalone to new clustered master server?

If so, this is not supported (as can be seen in the other posts that you found ). The catalog restore will fail.

Also not sure that I quite understand what you are saying about the tapes.... sorry ...

[vazkr]

Hi Marianne,

Apologies as unable to revert earlier.

I intend to keep the name of the clustered virtual server the same as the old non-clustered master/media NBU server.

Further, using a new library to import the non catalogued tapes. Can you please explain what is on the tape header in the sense, if it references the old library (I guess not) as I noticed few users had trouble with the prefix on importing tapes?

Finally, is there anything else I need to keep in mind when moving from clustered to non clustered instance on WSFC?

Truly appreciate your kind advice!

[Marianne · Accepted solution]

Since you won't be able to recover NBU catalog on clustered master server, neither the hostname nor tape labels is of any importance.
You will start from scratch with everything new and do phase 1 and 2 import of all tapes.

The only other solution is to pay for catalog manipulation services to transfer standalone catalog to clustered config.

[vazkr]

Apologies, Marianne! Forgot to clarify a couple of related points. Happy to start a new thread unless you don't mind replying. :)

1. Since the catalog can't be directly imported from a non clustered to clustered instance, is there a simple way to 'copy' the configuration, specially the policy or settings?

2. This may be a no brainer but do we need new licenses (e.g. shared library/storages) or separate installer for a clustered instance? If yes, is there a doc for the same?

I understand ports requirements stay the same.

Thanks for helping out!