SGE Account filed format


     accounting – Grid Engine accounting file format


     An accounting record is written to the Grid Engine  account-

     ing  file  for each job having finished. The accounting file

     is processed by qacct(1) to derive accounting statistics.


     Each job is represented by a line in  the  accounting  file.

     Empty  lines  and  lines which contain one character or less

     are ignored.  Accounting record  entries  are  separated  by

     colon  (‘:’)  signs.  The  entries  denote in their order of



     Name of the cluster queue in which the job has run.


     Name of the execution host.


     The effective group id of the job owner when  executing  the



     Owner of the Grid Engine job.


     Job name.


     Job identifier – job number.


     An account string as specified by the qsub(1)  or  qalter(1)

     -A option.


          Priority value assigned to the job corresponding to the

          priority  parameter  in  the  queue  configuration (see



     Submission time (GMT unix time stamp).  For slave  tasks  of

     tightly  integrated  parallel  jobs,  the submission_time is

     always set to 0.


     Start time (GMT unix time stamp).


     End time (GMT unix time stamp).


     Indicates the problem which occurred in case a job could not

     be  started on the execution host (e.g. because the owner of

     the job did not have a valid account on  that  machine).  If

     Grid  Engine  tries  to start a job multiple times, this may

     lead to multiple entries in the accounting file  correspond-

     ing to the same job ID.


     Exit status of the  job  script  (or  Grid  Engine  specific

     status in case of certain error conditions).


     Difference between end_time and start_time (see above).

     The remainder of the accounting entries follows the contents

     of  the  standard  UNIX  rusage  structure  as  described in

     getrusage(2).  Depending on the operating system  where  the

     job was executed some of the fields may be 0.  The following

     entries are provided:



















     The project which was assigned to the job.


     The department which was assigned to the job.


     The parallel environment which was selected for that job.


     The number of slots which were dispatched to the job by  the



     Array job task index number.


     The cpu time usage in seconds.


     The integral memory usage in Gbytes cpu seconds.


     The amount of data transferred in input/output operations.


     A string specifying the job category.


     The io wait time in seconds.


     If this identifier is set the task was part  of  a  parallel

     job  and  was  passed  to  Grid Engine via the qrsh -inherit



     The maximum vmem size in bytes.


     sge_intro(1), qacct(1),  qalter(1),  qsub(1),  getrusage(2),



     See sge_intro(1) for a full statement of rights and  permis-



슈퍼컴퓨팅 전문 기업 클루닉스/ 상무(기술이사)/ 정보시스템감리사/ 시스존 블로그 운영자

You may also like...

페이스북/트위트/구글 계정으로 댓글 가능합니다.