Slurm Useful Command

Summary of useful command

Posted by Jerry Liu on July 3, 2021

Job status related command

  • Waiting time:
    Use squeue to see estimated start dates assuming new jobs with higher priority are not submitted, and assuming that jobs run to the wall time given on submission.
squeue -a --start -j <job_id>
  • Submitted jobs info:
    The scontrol command can be used to display information about submitted jobs, running jobs, and very recently completed jobs.
scontrol show job <job_id>
  • Finished jobs info:
    To see job usage for a finished job, issue:
seff <job_id>
  • Running job info:
    to list status info for a currently running job, issue:
sstat --format=AveCPU,AvePages,AveRSS,AveVMSize,JobID -j <jobid> --allsteps
  • More details when connet to the nodes
    ssh nova056
    top
    

Partition status

  • All partition info:
    To report the status of the available partitions and nodes, issue:
sinfo
  • More details:
    For more details on partitions limits, issue:
scontrol show partitions

Group info

sprio can be used to display a breakdown of the priority components for each job, e.g. sshare shows the two level fair-share hierachy data.

Install R packages

module load python/anaconda  
conda create -p /workspace/[username]/myEnvironment anaconda  
conda install p /workspace/[username]/myEnvironment anaconda your-module-name

See cheat sheet

  • If modlue not exist: try to search it in conda with conda search -t conda your-module-name.
  • If nltk.download function is needed, use python -m nltk.downloader <collection|package|all>
source activate /workspace/[username]/myEnvironment

External Links

see more tutorials here.