Look at your slurmd log file on bengal30 I would also recommend upgrading. Slurm v2.2 is two years old.
Quoting Mads Boye <[email protected]>: > Hi. > I've got slurm-2.2.0 running, and until yesterday it worked fine. > When it try to run my test script "test.sh": > > #!/bin/sh > > ### Job name > #SBATCH --job-name=mads > ### Name of queue to submit to > ### #SBATCH --partition munchkin > ### Time limit of total run time of the job > #SBATCH --time=2:00 > ### Number of nodes required for the job > #SBATCH --nodes=1 > ### Number of processor cores > #SBATCH --ntasks=1 > ### Amount of memory per core > #SBATCH --mem-per-cpu=200 > > #SBATCH --output=slurm-%j.out > #SBATCH --error=slurm-%j.err > > > srun hostname > echo "Hello World" > > exit 0 > > I get the following error: > > srun: WARNING: We will use a much slower algorithm with > proctrack/pgid, use Proctracktype=proctrack/linuxproc or > Proctracktype=proctrack/rms with Job accounting gather LINUX plugin > srun: error: Task launch for 8900.0 failed on node bengal30: > Unspecified error > srun: error: Application launch failed: Unspecified error > srun: Job step aborted: Waiting up to 2 seconds for job step to finish. > srun: error: Timed out waiting for job step to complete > Hello World > > I have the following partitions: > lion up infinite 1 alloc* lion1 > lion up infinite 19 alloc lion[2-12,14-21] > lion up infinite 4 idle lion[13,22-24] > bengal* up infinite 8 drain bengal[22-23,27-29,81-83] > bengal* up infinite 10 alloc bengal[20-21,24-26,44-45,78-80] > bengal* up infinite 65 idle bengal[1-19,30-43,46-77] > persian up infinite 16 idle persian[14-29] > munchkin up infinite 13 idle munchkin[1-13] > > The only think I've change since yesterday is that I've change > persian1-13 to be named munchkin[1-13] but I fail to see this being > the issue, as it also failes on the Bengal partition, which main. > > Best regards > > Mads. >
