Hi Peter
On Torque 4.2.5 I have the same behavior.
Note the PID and PPID.
The various pbs_mom processes are spawned in sequence by the first (2199),
and end gracefully when the job ends.
Why the extra pbs_mom instances are needed for interactive jobs is not clear,
but it has been working for us.
Gus Correa
Dear list,
ps -ef | grep pbs_mom
root 2199 1 0 Jan03 ? 00:04:22 /usr/sbin/pbs_mom -p -d /var/lib/torque
qsub -I -X
ps -ef | grep pbs_mom
root 2199 1 0 Jan03 ? 00:04:22 /usr/sbin/pbs_mom -p -d /var/lib/torque
root 29619 2199 0 11:14 pts/1 00:00:00 /usr/sbin/pbs_mom -p -d /var/lib/torque
root 29620 29619 0 11:14 pts/1 00:00:00 /usr/sbin/pbs_mom -p -d /var/lib/torque
pipeuser 29623 29621 0 11:14 pts/1 00:00:00 /usr/sbin/pbs_mom -p -d /var/lib/torque
Execution host after pipeuser has exited the shell of the interactive job
ps -ef | grep pbs_mom
root 2199 1 0 Jan03 ? 00:04:22 /usr/sbin/pbs_mom -p -d /var/lib/torque
Is this expected behaviour?
Thanks in advance for any hint. I have checked the changelog up to torque 6.0.0 and searched the internet and this list's archives but have not found anything related. We run a cluster with 10x RHEL6.7 nodes with TORQUE 4.2.10 and MAUI 3.3.1. The behaviour is seen directly after a fresh reboot. When a batch job is run on the same execution host, no additional pbs_mom processes appear in ps -ef. We do not use any of the multi-mom functionality of TORQUE.
Best regards,
Peter
_______________________________________________
torqueusers mailing list
http://www.supercluster.org/mailman/listinfo/torqueusers