I remember seeing something like this before... (~like~) OK, so by your log, the client rolled straight to tape at some point (probably because the disk pool filled or got full enough to where the preallocation couldn't be made). What is the maxnummp for your client? In the past there was something funky where if you were writing to tape and filled a tape and needed another tape (scratch) to continue but another drive wasn't available (for whatever reason and can't remember if it was because it treated the dismounting drive as an allocated drive and client was limited to maxnummp of 1) but anyway, things would die. Then what was funny... every time we retried... it would do the same thing... because the action that it failed on would be backed out and when we would retry, that same tape would be mounted again, reach EOT marker, call for another scratch, get messed up, fail, the transaction would be backed out, etc.....
Long shot but check you maxnummp on your client, then look in the log to see exactly what was going on with the tapes it was using... did the tape it was using hit EOT and was a scratch requested? stuff like that. Dwight E. Cook Systems Management Integration Professional, Advanced Integrated Storage Management TSM Administration (405) 253-4397 (877) 625-4186 T/L 349-4361
