[GE users] Subjob or parent/child semantics in SGE?

soeren_g soeren.gerlach at quanteam.de
Tue May 19 15:26:23 BST 2009


I am currently in progress of configuring a newly setup SGE based cluster to match it some very very specific requirements from energy traders regarding their applications ,-) So although the cluster is not too big at the moment I need to learn still a lot about SGE as the requirements are quite scary...

Is there anything like a subjob or parent/child semantics available in SGE? Having worked quite long with LSF/Symphony this was a handy feature, e.g. if the parent job had been killed all the subsequent child jobs have been killed automatically as well if necessary. This of course is only possible if the middleware knows about this relationship between the jobs.
Is something similar available in SGE? If not what would be the recommended practice to cope with this problem.

My first thought would be to introduce some kind of rigid job naming scheme so that at least one could see the relationship from the job name.

Please note, that I'm not talking about array jobs, but only simple "one slot" jobs.

Thanks & regards,
Soeren Gerlach


To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].

More information about the gridengine-users mailing list