Hi Matthew and Tamas, Taras Shapovalov <[email protected]> writes:
> Re: [slurm-dev] Fwd: Slurm Question > > Hi Matthew, > > We have not seen this error before. > > I suggest to stop slurmd on dena1 and start it in a terminal with > debug messages and system calls traces (you will see what files it > opens and hopefully some details about errors): > > strace /path/to/slurmd -D -vvv > > then submit a job to dena1 again. > > Also check that uid of gm1 user are the same on submit host and dena1. > > Best regards, > Taras > > > > > On Wed, Aug 21, 2013 at 9:37 PM, Matthew Russell > <[email protected]> wrote: > > > Hi, > > I upgraded slurm on my Bright 5.2 Cluster from 2.2.7 o > 2.4.2. Recently I've been having issue running slurm processes > though. I've read some postings ( like this ) but I still > couldn't get it working. > > > My slurm job is: > > gm1@dena:~$ cat slurm_batch_test.sh > #!/home/gm1/ > #SBATCH -D /home/gm1 > #SBATCH --export=NONE > #SBATCH -o /home/gm1/test_001.10470.out.o > #SBATCH -e /home/gm1/test_001.10470.out.e > #SBATCH -J test_001.10470 > #SBATCH --time=3600 > #SBATCH --partition=matt > #SBATCH > #SBATCH -c 4 > #SBATCH -t 4 > #SBATCH > #SBATCH > > > module load slurm > mpirun hello.sh > > > > > I run it with, and get the results: > > gm1@dena:~$ rm test_001.10470.out.*; sbatch slurm_batch_test.sh; sleep 1; > cat test* > Submitted batch job 1577 > slurmd[dena1]: execve(): > /cm/local/apps/slurm/2.4.2/spool/job01577/slurm_script: Permission denied > > > > In my log file, I get: execve(): > /cm/local/apps/slurm/2.4.2/spool/job01576/slurm_script: Permission denied > > > > > > > > My script is executable: > > gm1@dena:~$ ls hello.sh > -rwxr-xr-x 1 gm1 gm 37 Aug 21 11:55 hello.sh > > > > slurmd is being run by root, > > > root 18462 0.0 0.0 159384 1868 ? S 14:24 0:00 > /cm/shared/apps/slurm/current/sbin/slurmd > > > > I think it's running in /var/run/slurm > > > [root@dena1 2.4.2]# cat etc/slurm.conf |grep run > SlurmctldPidFile=/var/run/slurm/slurmctld.pid > SlurmdPidFile=/var/run/slurm/slurmd.pid > > > Which is owned by slurm, > > > [root@dena1 2.4.2]# ls /var/run/ | grep slurm > drwxr-xr-x 2 slurm slurm 4.0K Aug 21 14:24 slurm > > > And the ,pid file is owed by root, > > > [root@dena1 2.4.2]# ls /var/run/slurm/ > total 4.0K > -rw-r--r-- 1 root root 6 Aug 21 14:24 slurmd.pid > > > I'm not sure how to continue. > > > Can anyone help? Thanks. > I'd be interested in hearing more about how this goes because we currently have Bright Cluster Manager 5.2 and SLURM 2.2.7 and are looking to update SLURM. Maybe Tamas can say something about which configurations are supported by Bright. As this is rather Bright-specific, maybe it would be better to discuss this off-list. Cheers, Loris -- Dr. Loris Bennett (Mr.) ZEDAT, Freie Universität Berlin Email [email protected]
