Primary information
Username: briel
Cluster: yggdrasil
Description
Submitting a job array that writes to nfs-isilon.astro.unige.ch:/ifs/astro/projects/posydon results in the jobs immediately FAILING with an exit code 0:53, when ran from the private-astro-cpu partition.
I’m still able to access and create files on this filesystem from the login node.
Moreover, it’s also still possible to write to nfs-isilon.astro.unige.ch:/ifs/astro/projects/posydon from the debug-cpu`.
These jobs ran from the private-astro-cpu partition used to work fine, but now suddenly fail.
Steps to Reproduce
- Create a slurm submission file with the
private-astro-cpupartition onnfs-isilon.astro.unige.ch:/ifs/astro/projects/posydonand submit it from there. - The submitted run fails.
Example
#!/bin/bash
#SBATCH -N 1
#SBATCH --array=0
#SBATCH --partition=private-astro-cpu
#SBATCH --ntasks-per-node 1
#SBATCH --time=0-00:05:00
#SBATCH --job-name="mesa_grid_\${SLURM_ARRAY_TASK_ID}"
#SBATCH --output=mesa_grid.%A_%a.out
#SBATCH --mail-type=ALL
#SBATCH --mail-user=max.briel@unige.ch
echo "test"
Expected Result
I expect the test submit to create a file with mesa_grid_SLURM_ARRAY_TASK_ID.out with ‘test’ written in it.
Actual Result
I get no .out file and the job nearly immediately fails with a 0:53 error code.