[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Condor-users] wishlist for notification messages



Here is the beginning part of an automatic Condor e-mail message:

-----
This is an automated email from the Condor system
on machine "jdc.math.uwo.ca".  Do not reply.

Your condor job /home/spin/integrate/lorentz.condor -i 15 -w 3 -x 1000000000 -d 9 -f /home/spin/integrate/test-lorentz/7/7 -g -w 0 -y 7 7 7 7 7 7 7 7 7 7 was killed by signal 11.


Submitted by: jdc@xxxxxxxxxxx
-----

I was wondering if there could be a few changes to the format/content
to make it easier for both people and machines to parse.  Here is
an example of what I mean:

-----
This is an automated email from the Condor system
on machine "jdc.math.uwo.ca".  Do not reply.

Job: 207.40
Status: killed by signal 11.
Command: /home/spin/integrate/lorentz.condor -i 15 -w 3 -x 1000000000 -d 9 -f /home/spin/integrate/test-lorentz/7/7 -g -w 0 -y 7 7 7 7 7 7 7 7 7 7
Last run host: xxx.xxx.xxx.xxx  (or hostname)

Submitted by: jdc@xxxxxxxxxxx
Submit host: yyy.yyy.yyy.yyy  (or hostname)
Submit directory: /home/spin/integrate/test-lorentz
-----

The changes are:

1) Make the "Your condor job" line more easily parsable.  The format 
   I suggest above makes it easier to find the "killed by ..." or
   "exited normally with exit code ..."  information, since it'll
   always be near the left column.  (Also, I intend to write a script
   which when fed such an e-mail message, resubmits the job).
2) Add last run host.  When things go wrong, one of the first
   questions is where it was running, to see if there is a 
   pattern to the failures.  This information is in the log file,
   but extracting it is an extra step.
3) Add submit host.  Probably the same as the machine sending the
   e-mail?  But nice to have it explicit.
4) Add the working directory that condor_submit was run from.
   Some paths could be relative to this directory, so it could
   be important to know this.
5) Repeat cluster.process in body (not that important, but nice
   to have the info in one spot).

The exact format doesn't matter to me, but it'd be great to have
something including the extra information and which is easily
parsable.

Thanks,

Dan