Closed berland closed 6 months ago
The same error is emitted if the legacy driver is used. This has worked before, thus something has changed on the cluster
Commented on slack: It's picky : Just split that into -l nodes=1:ppn=1 -l mem=10gb That is a legacy allocation style limitation . For the newer v style, this is valid: –l select=2:ncpus=16:mem=20g,walltime=11:00
With
QUEUE_OPTION TORQUE MEMORY_PER_JOB 10gb
in an ert config, the generated qsub line becomes:
qsub -rn -NDROGON-159 -q hb120 -l nodes=1:ppn=1:mem=10gb -- /prog/komodo/bleeding-py38-rhel7/root/lib64/pythonsite-packages/equilibrium/bin/komodo_job_dispatch /lustre1/users/havb/havb/01_drogon_design_test/realization-159/iter-0
which gives an error message: "
failed with exit code 241 and error message: qsub: node(s) specification error
"This error from qsub goes away if "
:mem=10gb
" is removed from the resource specification.