[GE users] SGE 6.2u2_1 CPU accounting

jsrinivasan jay at nersc.gov
Thu May 14 18:12:01 BST 2009


We've noticed that SGE 6.2 seems to account for CPU usage differently 
from what 6.0 was doing. When we run a job, the CPU seems to be scaled 
in the qacct report (as well as the reporting file). We have a cpu scale 
factor set for our nodes:

(for example)

usage_scaling         cpu=2.110818

and qacct for a job running on the node shows:

qsub_time    Thu May 14 09:48:38 2009
start_time   Thu May 14 09:48:55 2009
end_time     Thu May 14 10:01:37 2009
ru_wallclock 762
ru_utime     761.322
ru_stime     0.335
cpu          1616.887
mem          822.852
io           0.000
iow          0.000
maxvmem      1.074G

These are all serial, single threaded, non-array jobs.

In 6.0 we were using the "cpu" number to get the cpu utilization. SHould 
we be using utime+stime now or is there some configuration change b/w 
6.2 and 6.0 that might give the old behaviour back?

Thanks for any help,



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

More information about the gridengine-users mailing list