Problem: When submitting batch analysis jobs, either locally or to an external cluster, the job fails with an error such as:
Job IDs could not be parsed from submit output.
ERROR:coca.commander:Command failed with error code 1: /bin/bash /cluster/jobs/uuid/run_job.sh
stdout: PREPROCESSING: 1
ARRAY: stderr: sbatch: error: Batch job submission failed: Invalid job array specification
ERROR:create_cluster_job:Job was not submitted.
Cause: During batch analysis SImA submits the jobs to SLURM (Job Scheduler) as part of an array, where the number of tasks in the array is equivalent to the number of wells in the dataset, e.g. so that a dataset containing 1536 wells would be submitted as an array including 1536 tasks.
The default maximum array size in SLURM is 1000, and is controlled via the MaxArraySize
parameter in SLURMs's configuration file (slurm.conf
).
If the number of wells in a dataset exceeds the value of MaxArraySize set in SLURM, the batch analysis job fails with the error show above.
Solution: Increase SLURM's MaxArraySize value so that it is greater than the number of wells in any given dataset.
The MaxArraySize
value of SImAs internal SLURM cluster is set to 5001 via the slurm_template.conf
file found in the config
directory of the SImA installation, so its unlikely that this problem will manifest when submitting local batch analysis jobs. The problem is more likely to occur when submitting batch analysis jobs to an external SLURM cluster, where the default MaxArraySize
value has not be increased.
Comments
0 comments
Please sign in to leave a comment.