From Documentation
Jump to: navigation, search
Note: This page's content only applies to the SHARCNET's legacy systems; it does not apply to Graham.


sqjobs - show jobs


   sqjobs [-a|--all][-u user]
       [-q][-r][-z][-d][--queue queuename][--command command-substring]

   -a, --all     show jobs of all users and states
   -u user       show jobs for the given user
   -q            show queued jobs
   -r            show running jobs
   -z            show suspended/preempted jobs
   -d            show done/killed/failed jobs
   --queue       show jobs in a specific queue
   --command     show jobs whose command includes this string
   -l            show more detail on jobs
   -L            show even more detail (per-process)
   -n, --none    show one-line status of the cluster
   -s, --summary show a brief summary of cluster jobs
   --sortbytime  sort by time in current state
   --sortbysize  sort by number of cpus
   -h, --help    show usage
   --man         show man page
   jobid...      one or more jobids to examine


sqjobs displays a succinct tabular listing of jobs in the SQ system.

by default, sqjobs shows your own jobs - it can show specific jobids, or of other users, or of all users. "sqjobs -aq", for instance, shows queued jobs of all users.

by default, one job is shown per line. more details of a job (or jobs) can be shown by adding the "-l" switch. adding "-L" causes sqjobs to query the nodes running a job and collect current, per-process information.

When showing all users, the listing for a user who has many jobs will be truncated. To disable this truncation, use -a twice.

job status is summarized:

   Q - queued, pending, waiting to be run
   Q* - queued, but waiting on completion of some other job
   R - running
   Z - suspended, probably preempted by a test job
   D - done (completed, failed, aborted) (visible for a short time after completion)

on the one-line-per-job summary, the time shown is elapsed time since the last major transition. for queued jobs, it's time since submission. for running jobs, since starting.

jobs are normally sorted by jobid (equivalent to submission order). it can be handy to use --sortbytime instead, to see how long jobs have been waiting/running.