Have you considered MPI - Gather and distribute a function to all processes (That I can evaluate at a point) - #4 by dokken
I would also suggest making a reproducible example; use a built in mesh, add all required imports and make a code that reproduce the error you show above.
It would also be of help to others if you add your implementation of: Problem with evaluation at a point in parallel - #6 by MiroK
which hangs, such that people can verify that it happens for them as well.