Unix loop over output of cut

Question

I have a bunch of slurm jobs that produce standard output and error files with this format:

<string>.<string>.<string>.<job_id>.ERR

where job_id is the job id assigned by slurm.

So to get these job ids I can:

cut -f 4 -d "." *.ERR

I'd like to pipe the output of this command to a loop that will run sacct -j <job_id> and grep which jobs have failed, using:

sacct -j <job_id> | grep "FAILED"

Can this be done in one command?

codeforester · Accepted Answer · 2017-03-03 01:12:31Z

6

You could use a Bash read loop, with process substitution:

while read -r job_id; do
  sacct -j "$job_id" | grep "FAILED"
done < <(cut -f4 -d. *.ERR)

You could also use xargs, which is more succint:

cut -f4 -d. *.ERR | xargs -n 1 sacct -j | grep "FAILED"

answered Mar 3, 2017 at 1:05

codeforester

43.8k21 gold badges122 silver badges159 bronze badges

Sign up to request clarification or add additional context in comments.

1 Answer 1