Dear all,

I am currently running optimization workflows (with OptimizeFW, not wf_structure_optimization to be able to change the workflow name) for a large number of LLZO structures and some of the workflows fizzled while the VASP calculations finished successfully (files zipped, OUTCAR looks ok). In the output file, I could find this error :
repeated 5 times which stops the workflow:
custodian.custodian.MaxCorrectionsPerJobError: Max errors per job reached: 5.

I checked custodian which indicates the MaxForceErrorHandler is deprecated. Is it possible to deactivate the MaxForceErrorHandler ? Since the structures are partially randomly generated, it is certainly be a problem with initial POSCAR (I checked however I don’t submit structures with atoms too close) but since VASP calculations finished, I am not sure.

Dear Florian,

Apologies for the delay responding to your message. I agree that it’s surprising this error handler is in use when it has been deprecated. I’ll look into that.

In the meantime, you can disable the MaxForceErrorHandler by setting the max_force_threshold argument in the OptimizeFW initialisation to None.