You can list all your submitted jobs, queued and running, using:
qstat
To get detailed information on a specific job, your or others, use:
qstat -j <job_id>
To see jobs of another user, use:
qstat -u <user>
To see jobs of all users, use:
qstat -u '*'
For example,
$ qstat -u '*'
job-ID prior name user state submit/start at queue slots ja-task-ID
------------------------------------------------------------------------------------------
380725 0.05508 job_qb3.sh alice r 04/20/2018 15:24:13 long.q@mac-inst-hmid1 24
380768 0.16819 proc0009.s carol r 04/20/2018 15:00:28 long.q@msg-hmio3 4
380815 0.05508 job_qb3.sh alice r 04/20/2018 15:59:58 long.q@mac-inst-id1 24
382375 0.05508 job_qb3.sh alice r 04/21/2018 13:37:13 ondemand.q@mac-inst-id3 24
386295 0.06649 se_psev_sc bob r 04/25/2018 08:42:39 long.q@msg-id19 6
386435 0.16819 proc0010.s carol r 04/25/2018 10:54:24 member.q@cdhi-idgpu1 4
386532 0.05055 GATK_SUB_G charlie r 04/25/2018 12:46:54 ondemand.q@mac-inst-id1 24
386533 0.05055 GATK_SUB_G charlie r 04/25/2018 12:46:54 ondemand.q@mac-inst-hmid1 24
386435 0.16819 proc0010.s carol r 04/25/2018 10:54:24 member.q@cdhi-idgpu1 4
386594 0.05055 GATK_SUB_G charlie r 04/26/2018 00:22:30 long.q@cdhi-idgpu1 24
387048 0.17577 C3Db2_REST bob r 04/25/2018 23:02:30 gpu.q@msg-iogpu11 4
387524 0.16566 class3d alice r 04/26/2018 10:34:00 gpu.q@msg-iogpu9 2
387674 0.16566 class3d alice r 04/26/2018 13:40:30 gpu.q@msg-ihgpu3 2
388578 0.05004 pedigree1 bob r 04/27/2018 10:57:45 long.q@msg-ihgpu2 1 2
388578 0.05004 pedigree2 bob r 04/27/2018 10:57:45 long.q@msg-iogpu4 1 3
388806 0.05004 run_cluste bob r 04/27/2018 15:06:33 long.q@qb3-id3 1 20
388693 0.50000 run_0423.s carol qw 04/27/2018 12:53:02 2
388745 0.41908 test carol Eqw 04/27/2018 14:00:42 3
372081 0.16240 DIAGS alice qw 04/12/2018 10:44:20 1
In the output of qstat, there is a state column, which shows the
current state of each job. Commonly seen states are:
q - the job is queued
w - the job is waiting to be launched
h - the job is held on the queue, e.g. waiting for another job
dependency to finish
r - the job is currently running on a compute node
E - there was an error launching the job. See qstat -j <job_id>
for the reason why the job failed
d - the job is being deleted (from calling qdel)
For more details, see man qstat.
When your jobs will be launched depends on your jobs’ current priority on the queue. If one of your jobs is on the top of the priority queue and the resources (CPU, memory, …) you have requested are available, then that job will be launched next. If sufficient resources are not available, then a lower-priority jobs with lower resource may be launched in the meantime.
The priority scores of all jobs can be seen in column prior in the qstat -u '*' output (*). The “priority scores” are constantly recalculated as a function of all users’ jobs currently queued and running on the cluster. They are a function of:
There is no memory, that is, what you, your group, or others have run in the past does not matter.
(*) The priorities of already running jobs (those with an r in column state) are irrelevant.