Something that was used this morning as TSM died with logvolume being 100% used and no DB backup possible as library is offline.
Determining if the log is full
Start TSM (dsmserv) from the command line so that the startup messages can be seen. A message stating that log space cannot be allocated will be seen if the server is out of log space just prior to dsmserv dumping core.
So your log is full
From a command line on the TSM server, execute:
dsmfmt –m –log /tsmlog01/log911.tsm 1000
dsmserv extend log /tsmlog01/log911.tsm 1000
The log will be extended by 1000 meg. This is very little room (room=time) to fix the problem.
Start the TSM server in quiet mode. If TSM fails to start, see section Determining if the log is full.
From a TSM command prompt execute:
This command will stop ALL client activity.
TSM is up, now what?
Now that TSM is running, determine the cause of the log becoming full, most likely causes are no scratch tapes, no tape drives, and a large archive conversion. Remember, you only have 1000 megs of log space. (Q log f=d)
Now that everything is cool again
After the server is running normally, you must remove the 911 log file. Failure to do this will render the system unrecoverable should the log become full again.
1. Change the log mode to normal
2. Reduce the log size by 1000 meg – (REDUCE LOG 1000)
3. DELETE LOGVOLUME /tsmlog01/log911.tsm (the file is not deleted, just the TSM reference)
4. Change the log mode to Rollforward, this will cause a DB backup
5. From a TSM command line run: enable sessions
thanks to adsm.org for the procedure