[GE users] Killing stale processes after a job

Peter Eriksson peter at ifm.liu.se
Tue Nov 8 08:44:08 GMT 2005


    [ 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. ]

On Mon, 7 Nov 2005, Clements, Brent M (SAIC) wrote:

> This is our situation, hopefully someone has a solution(We've thought of
> a few but want your opinions):
>
> We have a cluster running SGE 5.3
>
> We have a user who runs a PVM based application which doesn't appear to
> be tightly integrated with SGE
>
> When the job ends, or the user deletes the job, the applications
> processes are still running on the cluster compute nodes which he
> submitted his job to.
>
> Our configuration is such that a user can submit multiples jobs to the
> same set of nodes, so doing an skill -u uid in a cleanup script would
> not work well in our situation.
>
> I'm assuming others have run into this problem before, how have those of
> you who have, solved the issue?

This sounds like an ideal candidate for Solaris 10 'contracts', but since 
you speak of "skill", which I think is a Linuxism, I think you lose out...

Sorry :-)

-- 
Peter Eriksson <peter at ifm.liu.se>            Phone:    +46 13  28 2786
Computer Systems Manager/BOFH                Cell/GSM: +46 705 18 2786
Physics Department, Linköping University     Room:     Building F, F203
SE-581 83 Linköping, Sweden                  http://www.ifm.liu.se/~peter


    [ Part 2: "Attached Text" ]

---------------------------------------------------------------------
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