Am 10.12.2011 um 13:23 schrieb mahbube rustaee: > Hi , > I use cpu accounting, one record is: > . > > qsub_time Sat Nov 19 13:44:39 2011 > start_time Sat Nov 19 13:44:43 2011 > end_time Sat Nov 19 13:44:44 2011 > slots 200 > failed 100 : assumedly after job > exit_status 137 > ru_wallclock 1 > ru_utime 0.030 > ru_stime 0.023 > ru_maxrss 0 > ru_ixrss 0 > ru_ismrss 0 > ru_idrss 0 > ru_isrss 0 > ru_minflt 3008 > ru_majflt 0 > ru_nswap 0 > ru_inblock 0 > ru_oublock 0 > ru_msgsnd 0 > ru_msgrcv 0 > ru_nsignals 0 > ru_nvcsw 39 > ru_nivcsw 8 > cpu 6.310 > mem 4.654 > io 0.013 > iow 0.000 > maxvmem 309.693G > > > 1) must cpu be equal to (ru_utime + ru_stime) ?
No. In the above case you killed the job and the shepherd never got any values for the child tasks. Only reliable is then the CPU time. > 2) How can I describe cpu, ru_utime and ru_stime values? CPU time is collected by the shepherd by the additonal group id attached to all kids of a job. The ru_* values are the one by a system call `getrusage` by the shepherd. -- Reuti _______________________________________________ users mailing list [email protected] https://gridengine.org/mailman/listinfo/users
