Decommission an instance

[Paradox667]

Hi all,

Bit of a strange question - is there anything special required when a TSM instance is decommissioned?
Particularly with shared library / DRM ?

I am currently exporting the node data that I need to retain, once done I will be deleting node filespaces / removing nodes from the instance that is going - what other steps do I need to build into my process?

Not just to make sure the server is gone, but that the tapes are reusable for the remaining servers?

Cheers for any replies, as always much appreciated.

 

[rgg]

hrmm....

so, I assume this is a library client...

any client data on tapes it owned should be gone once you delete the nodes and filesystems, so that should take care of most of it.

I guess you might want to do some delete volhist type stuff to send backup db tapes back to scratch for the LM to divvy out and such. the last backup db volume can't be deleted with this, so maybe you'll need to identify that tape and do a relabel or something on the LM....or run a "temporary" backup db to a file devclass to make it the most current one.

please post back what you find, this is a pretty interesting question

 

[Paradox667]

Thanks RGG - that's the sort of thing I was thinking of and it certainly is a library client.

What I'm thinking from your comment is:

1. Delete all the node filespaces after export is complete (This is going to free up a bucket of scratch media...500+ TB :/ )
2. Drop DB expiration days down to 1
3. Take final DB backup to file devclass
4. Delete vol hist to current day (make sure all tapes are gone)
5. Decommission the instance

I think that should give the best chance for all tapes being returned to LM for reuse.

Will report back the exact process I go through when it happens - the export is going to take a little while (500TB server to server transfer isn't very quick and it's all legacy data that I have to keep for a bit longer)