Commonly used commands on HPC cluster

SGE

List all nodes and memory

qhost

Show current running jobs

qstat

Show current jobs for a user <uni>

qstat -u <uni>

Submit a dummy job of “sleeping for 40 seconds” (/bin/sleep 40) to a particular node requesting 10GB memory:

qsub -l h="node33" -l h_vmem=10G -b y /bin/sleep 40

See here for a list of options and explanations.

slurm

Status of a previous job

eg, how much memory it was using:

$ sacct -j 15215620 --format=jobid,jobname,partition,account,alloccpus,state,maxvmsize

         JobID    JobName  Partition    Account  AllocCPUS      State  MaxVMSize
  ------------ ---------- ---------- ---------- ---------- ---------- ----------
  15215620          RCDAS     sandyb pi-msteph+         16  COMPLETED
  15215620.ba+      batch            pi-msteph+         16  COMPLETED  83752696K

Find and rerun multiple jobs

qstat | grep "Q batch" | cut -f 1 -d " " | xargs sudo qrun

Contact

Gao Wang