PandaRoot jobs on GSI farm [message #6862] |
Mon, 09 June 2008 14:37 |
Bertram Kopf
Messages: 110 Registered: March 2006
|
continuous participant |
From: *ep1.ruhr-uni-bochum.de
|
|
Dear experts for the PandaRoot event production,
there are still 106 PandaRoot jobs running on the GSI machines. Most of them (82) have been started on 28th of May or even earlier.
The remaining 25 jobs have been started the 25th of June. Moreover they only used a few seconds of CPU time so far. It seems that this are endless jobs with the consequence that the CPU resources for the other groups are very limited right now.
Therefore I would kindly like to ask you to check whether everything is going well and in case of endless jobs to kill them.
Many thanks in advance,
Bertram.
|
|
|
|
|
|
|
|
|
|
|
Re: PandaRoot jobs on GSI farm [message #6944 is a reply to message #6934] |
Mon, 16 June 2008 15:17 |
Kilian Schwarz
Messages: 91 Registered: June 2004 Location: GSI, Darmstadt
|
continuous participant |
From: *gsi.de
|
|
Hi Johan,
yes, AliEn automatically restarts the Job agents so that they are always on the nominal value, which is supposed to be 20 running and 20 queueing.
But AliEn submits in bunches of 100, so this is one of the problems, I suppose, why the nominal value can be exceeded if such small numbers are configured.
The Grid is designed for large numbers of jobs
The CE seems to count correctly, though. At the moment it finds 20 queueing jobs and submits nothing more.
I will try to keep things under control.
Cheers and sorry for the inconvenience.
Cheers,
Kilian
|
|
|