[GE users] sge_shepherd staying around

Rayson Ho rayrayson at gmail.com
Fri Jun 22 00:09:42 BST 2007


    [ The following text is in the "ISO-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

If you cannot kill the processes with "kill -9", then it is usually
something wrong with the OS or the NFS setup...

What state are those sge_shepherd in?? If they are "running" normally,
you should be able to kill them easily.

Also, you can use strace to find out what they are doing... and also,
you can use gdb to find out what functions they are currently
executing...

Rayson



On 6/21/07, Margaret Doll <Margaret_Doll at brown.edu> wrote:
> The sge_shepherd jobs stay around after a user has qdel their jobs.
>
> The first five jobs indicated by the number after "_shepherd" are
> shown as running in qmon. (status is "r")
> The first four jobs are actually taking up 100% of the CPU for their
> slot.
> The last two jobs were "qdel" by the user over a day ago.
>
>
>
> sge       6281  4225  0 Jun19 ?        00:00:00 sge_shepherd-482 -bg
> sge       6355  4225  0 Jun19 ?        00:00:00 sge_shepherd-484 -bg
> sge       6429  4225  0 Jun19 ?        00:00:00 sge_shepherd-488 -bg
> sge       6503  4225  0 Jun19 ?        00:00:00 sge_shepherd-490 -bg
> sge      11285  4225  0 Jun20 ?        00:00:00 sge_shepherd-508 -bg
> nanguyen 11286 11285  0 Jun20 ?        00:00:00 sge_shepherd-508 -bg
> nanguyen 11292     1  0 Jun20 ?        00:00:00 sge_shepherd-511 -bg
> nanguyen 11311     1  0 Jun20 ?        00:00:00 sge_shepherd-516 -bg
>
> I have only been able to get rid of the extraneous sge_shepherd jobs
> by rebooting the compute node.
>
> Obviously I need to learn a lot about sge_shepherd.
>
> Any one know what is causing the problem.  I believe that the
> extraneous sge_shepherd jobs count as a job filling a slot.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list