Consider removing scripts_regression_tests workflow, likely replacing it with something else

[billsacks]

The srt.yml workflow has been failing recently (due to ESMCI/cime#4965). This raised the question for me as to why we are running scripts_regression_tests from CMEPS. Should we remove this? Perhaps we can replace it with running some small CMEPS-focused test list?

It looks like this was originally added by @fischer-ncar and @jedwards4b .

[fischer-ncar]

I believe Jim added this since the CIME scripts_regression_tests depended on CMEPS. I'm not sure if this is still true or not.

[billsacks]

Thanks, @fischer-ncar - that makes sense and it's good to know the background.

My feeling is that we should get rid of running the full scripts_regression_tests and instead run a select set of tests more tailored to exercising cmeps that would hopefully pick up any issues that would impact scripts_regression_tests or any other testing that requires cmeps (which is really all testing in the CESM context).

[mvdebolskiy]

Btw, tmate keeps running after failed srt, so you have wallclock time of 6h on every commit to an open PR branch, I would suggest to elevate the priority of this issue.

[jedwards4b]

I don't understand why you allow this issue to continue to exist. Either fix the test or comment out the tmate section in the srt.

[billsacks]

@jedwards4b - I don't want to pull you too far into this, but since you commented: do you see any strong reason to keep the scripts_regression_tests run in place in CMEPS? I'm tempted to just remove it: I can see running a few system tests as part of the CMEPS GitHub action, but I'm not sure it's worth trying to keep scripts_regression_tests running from within CMEPS.

[jedwards4b]

I think that if you remove it you should replace it with some other testing.

[billsacks]

Based on discussion with @mvertens , I am temporarily addressing this by disabling the srt workflow - see #652 . I want to leave this issue open for a better long-term solution that will involve adding some new GitHub testing to CMEPS; at that point we can consider completely deleting the srt workflow rather than just disabling it.