Failed to Exit a Parallel HPC Application Debugging Task

Symptom

The user fails to exit a parallel HPC application debugging task.

Figure 1 Failure to clear the environment

Figure 2 Too long waiting time for clearing the debugging environment

Figure 3 Failure to exit MPI debugging

Possible Causes

• Too many ranks are started.
• The network connectivity is unstable.

Procedure

1. (Optional) Manually delete the files in the displayed path, that is, xxx.

   1

   rm -f xxx

2. Release process resources.
   1. Check the mpirun process.

      1	ps -ef | grep mpirun

   2. Stop the process. pid indicates the process ID.

      1	kill -15 {pid}

3. Restart the services.

   1 2	systemctl restart gunicorn_framework.service systemctl restart gunicorn_plugin.service