Monitoring job execution

AMEHPCJobHandler is a monitoring facility, it shows the progress and the status of the submitted Simcenter Amesim study.

Copy

usage: AMEHPCJobHandler.py [-h] [--sched {SLURM,UGE,LSF,PBS,NATIVE}]
                           [--jobid JOBID] [--jobdir JOBDIR]
                           {sched_job_status,sched_job_cancel,job_progress}

positional arguments:
  {sched_job_status,sched_job_cancel,job_progress}
                        Command to execute

optional arguments:
  -h, --help            show this help message and exit
  --sched {SLURM,UGE,LSF,PBS,NATIVE}
                        Batch job scheduler
  --jobid JOBID         Job id
  --jobdir JOBDIR       Job directory (required for sched_job_status and
                        job_progress command)

More precisely, when the AMEHPCJobHandler is called with the job_progress option, it returns five types of information, namely:

• job_progress_ratio (percentage of completed jobs)

• job_nb_runs

• job_simulated_runs

• job_failed_runs

• job_aborted_runs

When the AMEHPCJobHandler is called with the sched_job_status option with a jobid as an argument, it returns the status of a particular job. The different statuses available are:

PENDING	The job is being queued waiting for execution. It might stay pending, waiting for resources to become available.
RUNNING	The job is currently being executed.
COMPLETED	The job is finished, results are available.
CANCELED	The job has been canceled by the job’s owner or the system administrator.
RUN_ERROR	The scheduler has reported an error state for the job. This is not related to the simulation status.
UNKNOWN	The Simcenter Amesim HPC framework cannot retrieve status information from the scheduler. For instance, the command might be unknown or there may be an installation issue.
SUSPENDED	Please check with your system administrator (not used for SLURM scheduler).

Note that the tool provides an instantaneous status for the job. In order to follow the evolution of the job please run it multiple times in a row (for example with the "watch" command on Linux).