User doesn't know why their job exited early

Occasionally a user will report that their job exited before completing and they don't know why. There are several possible reasons for this:

• Their job ran out of time. Look in the server's accounting file to see which service node their job ran on. Then look in the MOM's log file on that node. If the MOM killed the job because it ran over it's walltime allocation, you will see a message like walltime used (24:30:52) exceeds limit (24:00:00). There is also a message in their .e file telling them that PBS killed their job.

• One of their processes exited with an error or a signal. They should be able to determine this by looking at their .o and .e files, since yod will report such an error. Alternatively, look in the server's accounting file to determine what node their job ran on, then look in the userlog file written by yod for an entry about their job. The entry will indicate if one of their processes terminated with an error code or a signal.

• The MOM refused to run the job that was handed to it by the server. Again, look in the server's accounting file to determine which node that job ran on. Then look in the MOM's log file on that node. If the MOM refused to run the job, it will log that fact to the file, along with a reason. This happens rarely, and if it does the best course of action is to restart the server and all the MOMs.