Skip to content

squeue#

Useful options#

  • -j <job_id1,job_id2> : to restrict to one or more jobs (comma separated)
  • --name=<jobname> : to restrict to job with a given name
  • -p <partition name>: to restrict to a partition
  • --qos=<qos name> : to restrict to a specific QoS
  • --state=<STATE1,STATE2> : to restrict to jobs in a specific state (comma separated), that is PENDING or PD, RUNNING or R
  • -u <login>: to restrict to jobs belonging to a specific user
  • --Format=<field1:length,field2:length>: to display only desired fields with the adequate length

Examples#

To see all jobs belonging to you

Code Block (bash)

login@maestro-submit ~ $ squeue -u <yourlogin>

The output looks like

Code Block (bash)

             JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)
           6886972    common       sh    login  R   00:40:40      1 maestro-1024

To see the jobs that are waiting to start in partition dedicated with QoS fast, you can do:

Code Block (bash)

login@maestro-submit ~ $ squeue --state=PD --qos=fast -p dedicated --Format=jobid:10,name:20,username:10,partition:10,qos:10,statecompact:3,reason:22,numcpus:6

The output looks like

Code Block (bash)

JOBID     NAME                USER     PARTITION QOS       ST REASON                CPUS
16327366  jobname1            login1   dedicated fast      PD Priority              4     
16326698  jobwithalongname    login1   dedicated fast      PD BeginTime             12    
16116959  script.sh           login2   dedicated fast      PD Resources             1

3 jobs are pending:

  • 16116959: because Resources are not availalble
  • 16327366: because other pending jobs have a higher Priority
  • 16326698: because the owner has specified to start the job later with--begin option or the job has just been requeued (REASON: BeginTime). That job was probably running on a node belonging to a research unit  and was killed because the unit launched its own jobs.

You can focus on a particular user such as yourself with -u   option:

Code Block (bash)

login@maestro-submit ~ $ squeue --Format=jobid:10,name:15,username:10,partition:10,qos:10,statecompact:3,reason:20,numcpus:6,numnodes:6,minmemory:11,starttime:22,timeused:12,nodelist:15  -u login1

Then the output looks like

Code Block (bash)

JOBID     NAME        USER    PARTITION QOS       ST REASON              CPUS  NODES MIN_MEMORY START_TIME            TIME        NODELIST       
33487321  breseq      login1  common    normal    PD Priority            8     1     25000M     2018-04-30T23:09:25   0:00                       
33487319  breseq      login1  common    normal    PD Resources           8     1     25000M     2018-04-30T22:35:49   0:00                       
33521872  breseq      login1  common    normal    PD Priority            8     1     25000M     2018-04-30T23:09:25   0:00                       
33487317  breseq      login1  common    normal    R  None                8     1     25000M     2018-04-30T18:00:04   27:19       maestro-1023

sprio: check jobs priority#

Useful options#

  • -l  option is recommended to see your name and find your jobs easily,
  • -p   is convenient to reduce the list to the jobs submitted in one or more partitions (comma separated), for example the partition on which you submitted your own jobs,
  • --sort=-y  is convenient to sort by priority instead of job id by default. If you want to have jobs with higher priority at the bottom instead of the top, change to --sort=+y
  • -u allows you to restrict the list to jobs of some users (comma separated logins)

Example#

To see the list of jobs pending on the common partition in descending order of priority, you can do the following:

Code Block (bash)

[login@maestro-submit ~]$ sprio -l -p common --sort=-y | less
         JOBID PARTITION     USER   PRIORITY       SITE        AGE      ASSOC  FAIRSHARE    JOBSIZE  PARTITION        QOS        NICE                 TRES
       38765414 common      user3        448          0         35          0        403          0          0         10           0                     
       38765415 common      user3        437          0         25          0        403          0          0         10           0                     
       38781051 common      user2        406          0          5          0        392          0          0         10           0                   
       38718124 common      user1        372          0        351          0         11          0          0         10           0   
       38750272 common      user1        154          0        133          0         11          0          0         10           0                     
       38757150 common      user1        154          0         43          0         11          0          0        100           0

As you can see, user1 has 2 jobs with exactly the same priority of 154. The second one has been waiting for less time but it is compensated by the fact that it was submitted with the fast QoS which has a priority of 100 instead of 10 for the default normal QoS.

To change the priority of one or more of your jobs, see the dedicated paragraphe.

Related articles appear here based on the labels you select. Click to edit the macro and add or change labels.

false5FAQAfalsemodifiedtruepagelabel in ("monitoring","squeue") and type = "page" and space = "FAQA"squeue monitoring

true
Related issues