This change will be in version 2.3.2 and fixes this problems:

diff --git a/src/slurmctld/job_mgr.c b/src/slurmctld/job_mgr.c
index 3c07188..1f51639 100644
--- a/src/slurmctld/job_mgr.c
+++ b/src/slurmctld/job_mgr.c
@@ -5530,7 +5530,8 @@ void pack_job(struct job_record *dump_job_ptr, uint16_t show_flags, Buf buffer,
                packstr(dump_job_ptr->comment, buffer);
                packstr(dump_job_ptr->gres, buffer);
                packstr(dump_job_ptr->batch_host, buffer);
-               if ((show_flags & SHOW_DETAIL) &&
+               if (!IS_JOB_COMPLETED(dump_job_ptr) &&
+                   (show_flags & SHOW_DETAIL) &&
                    ((dump_job_ptr->user_id == (uint32_t) uid) ||
                     validate_slurm_user(uid))) {
                        char *batch_script = get_job_script(dump_job_ptr);



Quoting Lennart Karlsson <[email protected]>:

Hi,

I have set

MinJobAge=43200

meaning that I want to keep job information
for several hours after job termination,
so I can run "scontrol show job" for the
job when debugging problems.

After an upgrade, I think to version 2.3.0,
I got a lot of error lines in the slurmctld.log,
and they are still there in version 2.3.1.
Here are some examples:
[2011-10-28T15:30:01] error: Error opening file /usr/local/slurm-state/job.1522753/script, No such file or directory [2011-10-28T15:32:00] error: Error opening file /usr/local/slurm-state/job.1522753/script, No such file or directory [2011-10-28T15:33:39] error: Error opening file /usr/local/slurm-state/job.1522753/script, No such file or directory [2011-10-28T15:43:20] error: Error opening file /usr/local/slurm-state/job.1522753/script, No such file or directory [2011-10-28T15:43:32] error: Error opening file /usr/local/slurm-state/job.1522753/script, No such file or directory [2011-10-28T15:44:02] error: Error opening file /usr/local/slurm-state/job.1522753/script, No such file or directory [2011-10-28T15:45:01] error: Error opening file /usr/local/slurm-state/job.1522753/script, No such file or directory [2011-10-28T15:48:48] error: Error opening file /usr/local/slurm-state/job.1522753/script, No such file or directory [2011-10-28T15:48:48] error: Error opening file /usr/local/slurm-state/job.1522753/script, No such file or directory

Job number 1522753 is a finished job, so I can understand
that there is no state directory for it.

Quite often some part of slurmctld tries to find a file
that is not there, for a finished job. I think, but am
not sure, that these lines appear after a restart of
slurmctld and disappears after several hours (perhaps
after 43200 minutes, I do not know?).

My simple wish is that I get none of these error lines,
or perhaps one warning only and no errors.

For the most part, I like SLURM very much, though!

Best regards,
-- Lennart Karlsson, UPPMAX, Uppsala University, Sweden
  http://www.uppmax.uu.se





Reply via email to