DB2 Issue

[RickyJ]

Hi Folks,

I've been facing a DB2 issue for a week, so I hope someone can help me...
Customer reported the following error output:

BACKUP: DATABASE[EDB] MODE[ONLINE] LEVEL[DB] TYPE[FULL] OPTIONS[INCLUDE_LOGS] BUFFERS[0] TARGET TYPE[TSM] PARALLELISM[4] PRIORITY[100] SESSIONS[4] MGMT CLASS
[] NODENAME[] OWNER[]
Started at Sat Aug 22 02:00:02 2009
SQL2025N An I/O error "9" occurred on media "TSM".

In the file "dsmrc.h" I found the following entry for this return code:

DSM_RS_ABORT_DATA_OFFLINE
which means (according to TSM API book):
Explanation: For the restore or retrieve operation, one or
more of the requested files must be recalled from offline
storage media (generally tape). The wait time depends on
your site’s offline storage management policies.

System Action: TSM waits for offline storage media to
become available and then continues.

I looked on the TSM server if there were any media mounts but the actlog indicates no mount activities for the timeframe of the report. I also tried to extend the search for the whole day, but found nothing usable...

Is there anything that could be checked from client/API side or did I miss something to check on TSM server?

Thanks a lot in advance for any help!
Cheers,
Ricky

 

[cfhoffmann]

Has this ever worked before or is it a new TSM client/API installation ?
And are you actually running a backup and if so what's the command you're using ?
Have you tried running 'db2adutl query' to see if you can connect to TSM ?

 

[DanGiles]

If you just search the actlog for "mount", you won't find anything since TSM won't even try to mount an off-line volume. Find the client session number and search the actlog for that to see what errors are appearing on the TSM server.

 

[RickyJ]

It turned out today that the db2 node had a successful backup on 29 August. I really have no clue what happened a week before. What I found on tsm is that the node had a backup session with ~65MB data transfer and that session was successful. I got this from an SQL query from summary... How is this possible if the error log says there were some kind of problems?

BTW another question...what do you think if a node starts different backup/archive sessions on 2 TSM servers? In dsm.sys I see separated stanzas for the same nodenames (it is also a DB2 node). Does it make any sense?

Thanks,
Ricky