BUG: WAVECAR deletion

[JaGeo]

Describe the bug
In my current tests of the MPVaspLobsterMaker, the WAVECARs are not correctly deleted. It's a purely jobflow-based run. We would need to investigate this in detail.

@naik-aakash have you seen issues like this?

[naik-aakash]

Umm, I did not see this. But I can check it 😃

[JaGeo]

I ran the workflow with jobflow on supermuc.

Could you also try running the workflow with jobflow remote? I get weird key errors when the lobster job is saved. It has nothing to do with size but with invalid keys.

[naik-aakash]

Is it related specifically to the Bandoverlaps data key or any specific field? And any specific compound?

[JaGeo]

Mg3Sb2. The rest is not clear as the outputs are not saved. It would need further testing

[naik-aakash]

Thanks will look into it. 😄

[naik-aakash]

Hi @JaGeo ,Checked it with jobflow remote and it seems to originate from band_overlaps field. I will test with the recent version of our Schema (open PR), I think that should have fixed this issue already as we do not store it as dict anymore but as data object.

Seems earlier pydantic version did not have such checks implemented ,so never encountered this issue before.

[JaGeo]

I tried it with the recent version as well (also your branch). It also failed for me

[naik-aakash]

Yeah , I just finished checking the same. Seems I need to update keys of bandoverlaps object in pymatgen to make it work with pydantic checks...