[GE users] accounting

reuti reuti at staff.uni-marburg.de
Wed Nov 4 16:40:22 GMT 2009


Hi,

Am 04.11.2009 um 15:53 schrieb mlmersel:

> Hi:
>
> When I get the accounting data for a particular queue I get the  
> following:
>
> CLUSTER QUEUE     WALLCLOCK         UTIME         STIME           CPU
> ======================================================================
> a.q              3017647             8             0       1019688
>
>
> do these numbers look okay?
> Why are the user time(UTIME) and system time (STIME) so low
> relative to the cpu time and wallclock time.

are these normal serial jobs, or any special applications which jump  
out of the process tree? The CPU time is measured by SGE due to the  
attached additonal group_id (whether they jump out of the process  
tree or not), while the s/u_time is just the time the shepherd gets  
from the returned process (like `time my_application`).

When the information is not send back, e.g. because the job (and so  
the shepherd) was killed by a qdel or reached the h_rt, then there  
are now values to report.

-- Reuti

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=225042

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



More information about the gridengine-users mailing list