mpas3p75 tests fail in ctsm5.4.044 due to CDEPS version
- Dominant language
- Fortran
- Stars
- 352
- Forks
- 361
- Avg merge
- 6d 6h
- Merged PRs (30d)
- 8
Description
mpasa3p75 tests fail somewhere between ctsm5.4.021 and ctsm5.4.044 due to an update in CDEPS. They pass if I go back to cdeps1.0.91 in ctsm5.4.044. ctsm5.4044 uses cdeps1.0.96.
Here's the text in the PR comment that lead to this:
The uhr_decomp_init test had one test run as expected, one that's in the queue and this one failed:
SMS_Ln1_PL.mpasa3p75_mpasa3p75_mt13.I2000Clm45Sp.derecho_intel.clm-for_testing_fastsetup_bypassrun--clm-mpasa3p75 ( RUN)
It ran out of the hour wallclock time it was given. This is concerning because based on previous timings an hour should have been overly generous. From looking at log files it looks like it failed in datm and likely hung so it didn't die until it reached the wallclock limit.
_Originally posted by @ekluzek in https://github.com/ESCOMP/CTSM/issues/3666#issuecomment-4883058043_
Contributor guide
Research direction
Start with the uhr_decomp_init test and the SMS_Ln1_PL.mpasa3p75... test run described in the issue. Compare logs from ctsm5.4.021 through ctsm5.4.044, especially the datm portion, and reproduce with CDEPS 1.0.96 versus 1.0.91. Done means identifying the CDEPS change that causes the hang or timeout and confirming the mpasa3p75 test completes successfully.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- fortran
- Domain
- testing
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100