Restoring Database to new HW Config

[mccleld]

I'm still not sure why when the tape it wanted wasn't available it didn't request the copy pool tape. Is there somewhere I need to set in TSM to restore client data from copy pool, rather than backup pool?

I'm pretty sure that the last time I did this, I made the no-longer-available-in-DR-location primary storage pools (i.e. the DISK pool and the primary TAPE pools) unavailable with an `UPDATE STG <STGPOOL_NAME> ACC=UNAVAILABLE` - any subsequent restore requests would then get fetch data directly from the checked-in copy pool volumes.

Should be as simple as that...

(edit)
In fact, that's one or two options - here are both articulated in a Technote from IBM: "Restore of client files from offsite volumes" http://www-01.ibm.com/support/docview.wss?uid=swg21248881
(/edit)

HTH,
_______________________
David McClelland
London, UK

 

[hjeckman]

I actually just did that before I even read your post
I'm still getting "Data was not available on server and was skipped" on my client.
I'm going to try restarting the TSM Server and if it still isn't working I'm going to try a different client or a different set of files.

 

[mccleld]

I'm still getting "Data was not available on server and was skipped" on my client.

Have a look at exactly which files you have there that aren't restoring - is it for all of them? For a client recently, they had a disk pool for directories only (e.g., DIRPOOL) - which is great, but they weren't creating any storagepool backup volumes for it. So, whilst the data objects were restoring from the copy stgpool vols, their directories were not.

 

[hjeckman]

Yes, it's for all of them. I don't have anything special set up for dirs or anything either, just straight incremental backups forever.

I'm getting

ANR1430W Retrieval request denied for storage pool HBGTAPE - access mode="unavailable". (SESSION: 13)

And

ANR0836W No query restore processing session 3 for node IGFILESRVR1 and \\igfilesrvr1\s$ failed to retrieve file \BACKUPTEST\EXCEL 2000\EXCEL PRACTICE.XLS - file being skipped. (SESSION: 3)

HBGTAPE is one of my primary storage pools at the main site. I disabled it and left the only option for hte server to retrieve files as the copypool, it doesn't seem to want to go there.
Also, before I set the HBGTAPE pool to unavailable, it kept trying to grab a tape that was actually at the main site, which if there was a true disaster would not be available.

Edit:
Still not working for different nodes first time trying to restore it tries to grab the production sgtpool tape and gives

ANR8351E 003: Mount request for volume A00104L3 has timed out.
Aug 3, 2010 11:08:20 AM
ANR8313E Volume A00104L3 is not present in library HBGTS3310. (SESSION: 31)
Aug 3, 2010 11:08:20 AM
ANR1402W Mount request denied for volume A00104L3 - volume unavailable. (SESSION: 31)
Aug 3, 2010 11:08:20 AM
ANR1410W Access mode for volume A00104L3 now set to "unavailable". (SESSION: 31)
Aug 3, 2010 11:08:21 AM
ANR1420W Read access denied for volume A00104L3 - volume access mode = "unavailable". (SESSION: 31)
Aug 3, 2010 11:08:21 AM
ANR1420W Read access denied for volume A00104L3 - volume access mode = "unavailable". (SESSION: 31)
Aug 3, 2010 11:08:21 AM
ANR0565W Retrieve or restore failed for session 31 for node IGDBSERVER3 (WinNT). The storage volume A00104L3 is inaccessible. (SESSION: 31)

But it's not going to the COPYPOOL for the data once that fails.

 

[dms19]

Did you check in the volumes that you have on hand?

Its been a while since I've done a DR/test [thankfully] so I'm trying to remember the steps.

I recall checking the DRM tapes in and assigning them to a pool which was then used for the restore.

 

[hjeckman]

Did you check in the volumes that you have on hand?

Its been a while since I've done a DR/test [thankfully] so I'm trying to remember the steps.

I recall checking the DRM tapes in and assigning them to a pool which was then used for the restore.

Hmm..
No I didn't.
I was under the impression that I would be able to monitor the activity log and see what volume was being requested and just insert that volume for that particular restore.

If I check the volumes in the COPYPOOL all of the volumes show up, just not physically in the tape drive, so I figured it would call them up if no other stgpool was available.

 

[Kelvin]

Hmm..
No I didn't.
I was under the impression that I would be able to monitor the activity log and see what volume was being requested and just insert that volume for that particular restore.

If I check the volumes in the COPYPOOL all of the volumes show up, just not physically in the tape drive, so I figured it would call them up if no other stgpool was available.

Are you sure that backup stg is completing successfully?

 

[hjeckman]

I've never gone into the volumes to make sure that there's data on them, but they show as full or filling, I never have any failures or errors on my maintenance script, etc. and TSM choses them to be sent off-site daily... So I'm assuming that they are..

 

[Kelvin]

I've never gone into the volumes to make sure that there's data on them, but they show as full or filling, I never have any failures or errors on my maintenance script, etc. and TSM choses them to be sent off-site daily... So I'm assuming that they are..

Oh ok..u can run a q content or use select statements to check for data..Also I usually set the primary pools to destroyed..

 

[hjeckman]

Oh ok..u can run a q content or use select statements to check for data..Also I usually set the primary pools to destroyed..

Running q content 'volumename' returns data. File names.

I can try setting the stgpools to destroyed, but according to IBM documentation, unavailable works just as well.

 

[Kelvin]

Try updating the primary pool volumes as destroyed or unavail (personally, ive never tried unavail - destroyed has never failed me though) and make sure the copypool vols are readonly..otherwise youre missing a step.

 

[Kelvin]

In fact, I dont even do anything with the stgpools themselves, just the volumes in the stgpool...upd vol * wherestgpool=<stgpoolname> is what i use. - also make sure the db youre restoring from is the backup of the db after copy stgpools have finished (i've restored the wrong db before, wasnt fun starting from scratch)

 

[hjeckman]

Ok.. Tried just making the volume that it was looking for as destroyed. Now the backup/archive client is throwing the error:

ANS5122E The specified filespace does not exist on the server. The filespace might have been deleted by another client or an administrator.

I don't get it

 

[Kelvin]

Ok.. Tried just making the volume that it was looking for as destroyed. Now the backup/archive client is throwing the error:

I don't get it

It could be that your database backups start before copystgpool finishes..

 

[hjeckman]

It could be that your database backups start before copystgpool finishes..

Wow.. I never thought of that. I'll check on it tomorrow morning. That should all be in the maintenance script or admin schedules, correct?
If it's part of the maint script, as long as copystgpool is first, the dbbackup won't start until after that finishes, right?

 

[Kelvin]

Wow.. I never thought of that. I'll check on it tomorrow morning. That should all be in the maintenance script or admin schedules, correct?
If it's part of the maint script, as long as copystgpool is first, the dbbackup won't start until after that finishes, right?

I am not really familiar with maint script - I usually run out of AIX/Unix or TSM in itself with backup stg wait=yes, then backup db...you have to make sure the backup db runs after copy stg or the DR wont work. Also check for unavail copypool volumes on your production server and ensure backup stg was a success on the day of "disaster"

 

[hjeckman]

Ok, had some other things going on here, but I'm back on this now.
Easiest way to make sure that my copy pool back up and dbbackup are running in the correct order?

 

[Kelvin]

Check your admin schedules and the times that the processes kick off

 

[hjeckman]

Ok, that's what I thought, wanted to make sure.
The script is running now and won't allow me to view it until after it's finished. I think everything is good to go w/ it though.
If this is all good and you guys don't have any other suggestions, I may have to open a ticket w/ IBM.

 

[hjeckman]

Ok, I've looked at the maint script and it appears the active pools are being backed up to the copy pool as the first step, the DB Backup is towards the end. I don't see anyway of telling exactly when each process started and stopped, but I'm pretty sure it will only handle on process of the maint script at a time. So I should be good to go on that front.
Any suggestions as to why my secondary server WILL NOT look to the copy pool for client restores?