I'm trying to find a covenient way to peek through the jobs I have submited on an HPC system. Currently, I have the function

peek() {ccc_mpeek "$1" | less;}

defined in my .bashrc, where ccc_mpeek is an HPC specific function to see the output file of a running job and the $1 argument is the JOBID.

However, this is still a little incovenient, since I have to manually copy and paste the JOBID (7 digits) when I want to run the function above.

I'm looking for a way in which I can toggle through JOBID's as the argument to the previous function, based on the output of the squeue -u user_name, which has the following header:

         JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)

CodePudding user response：

The squeue command has an option to choose the information that is displayed. In you case, you would run

squeue -h -t R --me -o %i

to remove the header (-h) , filter only "running" jobs (-t R) and only display the job ID (-o %i).

Assuming you have only one job running, you can then define

peek() {ccc_mpeek "$(squeue -t R --me -o %i)" | less;}

If you have multiple jobs running, you will need to adapt it to either show only the first job ( ... | head -1) or iterate through the jobs with a for loop.