Troubleshooting

License not Authorized for User

Failed to checkout DSLS license for: QAE
ERROR 1A000035: License not authorized for this user
...

If you receive this error when trying to run Abaqus, then you first need to register for a university Abaqus license - speak to your supervisor about how to do this.

Out of Memory

If you do not request sufficient memory for your job, then it may be terminated by the job scheduler with the following error message (check the log output files):

slurmstepd: error: Detect 1 oom-kill event(s) in StepId=xxxxxx.batch. Some of your processes may have been killed by the cgroup out-of-memory handler.

This may also be accompanied by the following Abaqus error message (or similar):

Abaqus Error: The executable explicit_dp
aborted with system error "Termination signal" (signal 9)

You can address this issue by increasing the requested --mem-per-cpu in your job submission script.

Problem During Linking

The compilation and linking of large complex user subroutines may require much more memory than the actual Abaqus job that you intend to run.

If you encounter an error similar to the following:

ifort: error #10106: Fatal error in ld
Abaqus Error: Problem during Linking ...

Then you should check that your requested memory limit is not too low (<500M) and try increasing it.

Node Configuration not Available

sbatch: error: Batch job submission failed: Requested node configuration is not available

This error message can occur when you request too many resources (e.g. cpus, memory) than is available on the cluster nodes.

Note

Your total memory request is the --mem-per-cpu parameter multiplied by the number of cpus you have requested. You can find out how much RAM is available per node in the ACRC HPC documentation here (UoB only).

Job Terminates Immediately

If your job ends immediately after starting, make sure you have specified the interactive keyword for Abaqus in your job script.

Example:

abaqus job=<job-name> cpus=${SLURM_NTASKS} user=<usub-file> mp_mode=threads double=both interactive

Job Terminates Unexpectedly

If your jobs ends before the allotted maximum time, make sure you are running the job in your scratch space (/user/work/<username>) and not your $HOME space (/user/home/<username>).

Make sure also that you have enough space remaining in your scratch space; you can check with the user-quota command.