-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am 30.04.2017 um 06:27 schrieb Ben De Luca:

> I see this with processes that spawn another processes, when you kill the 
> parent the children go free. 

In case they no longer share the same process group you can try:

$ qconf -sconf
…
execd_params                 ENABLE_ADDGRP_KILL=TRUE

- -- Reuti


> On 29 April 2017 at 20:37, Reuti <[email protected]> wrote:
> Hi,
> 
> Am 28.04.2017 um 08:57 schrieb Guillermo Marco Puche:
> 
> > I'm expecting a weird behavior when I reschedule any job on my work cluster.
> >
> > Suppose I've a job running with id 1000, I reschedule it with the following 
> > command:
> >
> > qmod -r 1000
> > Pushed rescheduling of job 1000 on host compute-0-3.local
> >
> > Then if I login into compute-0-3.local and do a top/htop I still see the 
> > process running and consuming resources even if job is now running on 
> > another compute node after rescheduling.  I'm experiencing this on jobs 
> > that use a parallel environment like "mpe" and multi-threading.
> 
> Are they still bound to the sge_shepherd or did they jump out of the process 
> tree?
> 
> -- Reuti
> 
> 
> 
> 
> _______________________________________________
> users mailing list
> [email protected]
> https://gridengine.org/mailman/listinfo/users
> 

-----BEGIN PGP SIGNATURE-----
Comment: GPGTools - https://gpgtools.org

iEYEARECAAYFAlkFpS0ACgkQo/GbGkBRnRq66wCgth0t5j84l4izs9rmR6geGPZY
EokAoLgQ1oUjOPH9+zB1IVwUqmx5Anfp
=rNqH
-----END PGP SIGNATURE-----

_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users

Reply via email to