Misleading failure message from ERI test

[mnlevy1981]

For a recent CESM alpha tag, we had the following failure in TestStatus:

FAIL ERI.TL319_t232.G_JRA.derecho_intel.cice-default COMPARE_base_hybrid

However, the issue was that one part of MOM6 was not initializing from a restart in a branch run... the hybrid run was fine.

[billsacks]

Is your expectation that this would say COMPARE_branch_hybrid? If so, I think (but haven't looked super carefully or tested) that this could be accomplished by adding suffix="branch" here:

cime/CIME/SystemTests/eri.py

Line 271 in 511bf57

self.run_indv()

and then change these lines to have branch instead of base:

cime/CIME/SystemTests/eri.py

Lines 301 to 302 in 511bf57

	self._component_compare_test("base", "hybrid")
	self._component_compare_test("base", "rest")

Does that sound like the right fix to you? If so, would you have time to confirm this fix, through testing and/or looking through the relevant code?

[mnlevy1981]

@billsacks I don't really understand the ERI test (I thought about asking about the difference among the ${TEST}, ${TEST}.ref1, and ${TEST}.ref2 directories, and whether refN could be replaced with something more descriptive), but I suspect the test that is failing is comparing base and branch rather than branch and hybrid. So that's still the correct line to edit, but maybe it's the second argument to change?

I haven't had a chance to dig into _component_compare_test() to see how it knows which tests to compare, though... that function might need to be updated to figure out what to test given the updated arguments? I'm happy to add this to my list, but it'll be a pretty low priority. There's a chance I have time the week of Thanksgiving, but sometime next month is more likely.

[billsacks]

I mostly understand the ERI test. I'd be happy to sit down with you at some point and work through this together if you'd like.