How to close all APDL tasks

[natter1]

I have a script that runs pymapdl repeatedly. Sometimes one of the simulations causes an error. I try to catch that and make sure to close the mapdl-task.

    a.mapdl.exit()
    time.sleep(wait_s)

Often this is enough, but sometimes an MAPDL-task keeps running (I still don't know what exactly is responsibl for this). This prevents the next simulation to start ("unable to remove lock-file"). Is there a way to force a shutdown of all running MAPDL-tasks via python?

[germa89]

Hi @natter1

Are you using windows or linux? I'm aware that on Windows it might leave some child processes still running.

[natter1]

Yes, I'm under windows.

[germa89]

On Windows, exiting leave some orphan processes. Still investigating that.

I am opening an issue about that.

[germa89]

This should be fixed after #1700.

[mikerife]

Hi @natter1 Can I ask why your script is existing/starting MAPDL repeatedly? I thought that one of the big benefits of MAPDL as a Service is that we can just /clear and send commands for next model. Unless you need to change some command line options (number of cpu cores etc)?
Mike

[natter1]

Right now I'm working on a lokal PC. Hmm, I guess I could also clear and send new commands. Closing and starting again seemed to be the "safest" method to me. Is MAPDL still in a stable state after raising an exception, that stops the solution? After catching an exception in python I simply haven't considered to continue without restarting MAPDL so far.

[germa89]

I agree with Mike. Although closing the MAPDL instance is "safer" regarding cleaning, it also add quite a few extra time to the simulation.

Regarding the exception, well... if the MAPDL instance hasn't crashed with the exception, it should be OK to reuse.