LocalMapdlPool with a docker image

[chart-magnum]

Hello,

one of my collaborators works on MacOS. Out of convenience, he runs ANSYS APDL from docker image with license server provided by his institute. Now, he would like to parallelize some computation with LocalMapdlPool. Thus, I'd like to ask whether it would be possible to execute LocalMapdlPool with a docker image and his setup.

Thank you in advance!

[germa89]

I will have a look at this. I shall come back to you in 3-4 days. Thank you!

[germa89]

Hi @chart-magnum

Apologies for the delay.

I believe the current implementation of PyMAPDL does not support out-of-box LocalMapdlPool on a docker image. The main limitation currently is that the pool class does not support the connection to other Mapdl process. Right now it just create the requested number of processes (workers). For your collaborator setup (and any other docker configuration) it is needed that you can connect to existing Mapdl images.

It should not be very difficult to implement this, but, it will take more time than the initial promised 3-4 days hence my answer.
I will create an issue from this and see if we can implement this asap, as we are very interested in this too.

Thank you.

[germa89]

This request is being tracked in: #706

[chart-magnum]

Hi @germa89,

thank you for coming back to me. I see, that was my impression. Please let me know if I could be of any help in the development/testing.