[torqueusers] Fwd: job dieing immediately, 0 byte output file being produced

Sabuj Pattanayek sabujp at gmail.com
Wed Feb 24 16:21:05 MST 2010


So does output from currently running jobs stop being saved into spool
when the pbs_mom on the node is sent kill -2 ? I'm guessing pbs_mom is
using something like tee to capture output?

On Tue, Feb 23, 2010 at 6:01 PM, David Beer <dbeer at adaptivecomputing.com> wrote:
> Sabuj,
>
> If you upgrade while jobs are running and start again with -p, pbs_mom will monitor the jobs when it comes up again, and it will notify you that they close, but since it will no longer know the pid of the jobs, it won't know how they exited and will automatically assume that they exited correctly. If this is acceptable to you, then yes, you are fine to upgrade. If not, you might look into doing a rolling upgrade: http://www.clusterresources.com/torquedocs21/a.eupgrade.shtml
>
> David


More information about the torqueusers mailing list