Failing landIce_FO_MMS_FO tests on some platforms about albany HOT 8 CLOSED

I reverted the uvm-free stuff in ResponseGLFlux and the tests are now passing on my local build on weaver. Looking at the diff, it's not really clear to me what could be causing the issue though. Nightlies should be clean at least while I debug.

from albany.

Ah, I didn't see this had an issue opened for it. Yes, these are the tests I mentioned in the other issue that started failing after I fixed an issue with the GLFlux evaluator. I'm going to revert all my changes to this evaluator to see if it fixes and then debug from there.

from albany.

@mcarlson801 longshot: could this be a similar issue to the one you fixed in some responses a caouple of days ago? I recall that issue had very large diffs in the response/sensitivities.

from albany.

Thanks for working on this @mcarlson801 !

from albany.

These are back to normal, I can confirm. Should I close this, or keep it open? It seemed like you might still be investigating the root cause @mcarlson801 .

from albany.

This should be fixed now. If no FO_GIS tests start failing tomorrow, we can close this.

from albany.

landIce_FO_GIS_AdjointSensitivity_Thickness is failing again. It's working on GPU now at least. I'll do some debugging today on cee machines to see if I can find a solution.

from albany.

These tests are passing now, closing this and moving to the AdjointSensitivity_Thickness issue.

from albany.