There are no dedicated tables with all the dependencies between jobs, the full state of the dependencies is only visible with squeue and is not available with sacct.
The workaround is to parse the submit_line from the job_table to get the dependencies of this job with a single row request.
To get the jobs that depend on a specific job, a select similar to select * from beluga_job_table where submit_line like '%--depend=afterok:%42%' and id_user='1000'; will fetch the various jobs and reuse the MySQL index by user to prevent a full table scan.
There are no dedicated tables with all the dependencies between jobs, the full state of the dependencies is only visible with
squeue
and is not available withsacct
.The workaround is to parse the
submit_line
from thejob_table
to get the dependencies of this job with a single row request.To get the jobs that depend on a specific job, a select similar to
select * from beluga_job_table where submit_line like '%--depend=afterok:%42%' and id_user='1000';
will fetch the various jobs and reuse the MySQL index by user to prevent a full table scan.