Hi, Can we configure s_vmem to send an email rather than sending signals? How?
Thanks, Ondrej > -----Original Message----- > From: Reuti [mailto:re...@staff.uni-marburg.de] > Sent: Friday, March 24, 2017 8:49 AM > To: Ondrej Valousek <ondrej.valou...@s3group.com> > Cc: William Hay <w....@ucl.ac.uk>; sge-discuss@liv.ac.uk <sge- > disc...@liverpool.ac.uk> > Subject: Re: Customizing emails send from GridEngine (was: RE: [SGE- > discuss] Enforcing memory limit for job) > > > Am 23.03.2017 um 18:35 schrieb Ondrej Valousek: > > > Hi, > >> Is this execution of the application the last command in the job > >> script? I always try to safe the return code of the application before > >> doing > some cleanup and then reuse this for the overall joscript. I mean: > >> joberror=$? > >> do_housekeeping > >> exit $joberror > > > > Ok, I'll try that - I was thinking about it, too. Thing is that engineers > > would > possibly have to rewrite lots of job scripts... Anyway, I'll see. > > > >> SGE does not see any error. It would be necessary for the OS to take > >> action, after a request for the memory allocation was denied to an > application. > > > > Not sure If I understand. Some while ago you said that there are two > > factors which ensures an application is within the limits > > - OS (by its shell limits) > > - SGE > > Correct. > > OS is per process > SGE all processes belonging to the job > > > > I am asking for a way to define limit the way so that hardlimit is not > enforced by setrlimit() but something (be it either OS or SGE) would rather > kill the job at once. > > I undestand this can be achieved by using the s_ limits right? > > In case there is a lower limit in s_vmem, which in total across all processes > indicates already that the main application will fail, it can be used this > way. > > Pitfall: I just realized, that s_vmem will also set the soft ulimit for the > job. This > renders it a little bit useless, unless you set it back to the hard limit in > the > jobscript again. This seems to work and you have SGE s_vmem now active in > case there is only a single processes which consumes all the memory (and > ulimit and *_vmem get both the same information). > > > > What is I use Control Groups and h_vmem? Will CGroups with its memory > handler be used to implement h_vmem limit? > > I have no experience in this area. > > -- Reuti ----- The information contained in this e-mail and in any attachments is confidential and is designated solely for the attention of the intended recipient(s). If you are not an intended recipient, you must not use, disclose, copy, distribute or retain this e-mail or any part thereof. If you have received this e-mail in error, please notify the sender by return e-mail and delete all copies of this e-mail from your computer system(s). Please direct any additional queries to: communicati...@s3group.com. Thank You. Silicon and Software Systems Limited (S3 Group). Registered in Ireland no. 378073. Registered Office: South County Business Park, Leopardstown, Dublin 18. _______________________________________________ SGE-discuss mailing list SGE-discuss@liv.ac.uk https://arc.liv.ac.uk/mailman/listinfo/sge-discuss
