Add a low resolution test to mimic GFSv17 cycling as much as possible #3617
Conversation
JessicaMeixner-NOAA
requested review from
TerrenceMcGuinness-NOAA,
WalterKolczynski-NOAA and
DavidHuber-NOAA
as code owners
JessicaMeixner-NOAA
requested review from
guillaumevernieres and
CatherineThomas-NOAA
There was a problem hiding this comment.
I think we need to add an account line in the main yaml, like in PR #3411. Otherwise, my test succeeded on WCOSS2 and all the tasks looks correct.
@CatherineThomas-NOAA - Thanks for catching this. I've added the account. |
There was a problem hiding this comment.
Thanks @JessicaMeixner-NOAA! The account update works in my test. Approve.
|
@CatherineThomas-NOAA Would this experiment, with appropriate changes to the ensemble, be a good basis for working on the reduced ensemble members? |
|
@AndrewEichmann-NOAA - Yes, with changes to |
|
Took a look at the test output provided by @JessicaMeixner-NOAA on Hera and noticed that the final cycle metp jobs (18z offset from end) aren't running because their dependencies are set for a non-existent cycle: Will fire off CI on WCOSS2, look for the same behavior, and see what fix is needed. |
KateFriedman-NOAA
added
the
CI-Wcoss2-Ready
|
@KateFriedman-NOAA - The C96C48_hybatmaerosnowDA.yaml test does not have the extra met+ jobs, although it also has DO_METP=YES as well. Not sure why this test would have the extra met+ jobs that are different from C96C48_hybatmaerosnowDA |
emcbot
added
CI-Wcoss2-Building
|
CI Tests set up to run in /lfs/h2/emc/ptmp/emc.global/PR/PR_3617/RUNTESTS on WCOSS |
aerorahul
added
CI-Hera-Ready
emcbot
added
CI-Gaeac6-Building
|
Experiment C96C48mx500_S2SW_cyc_gfs FAILED on Gaeac6 in Build# 1 in |
emcbot
added
CI-Gaeac6-Failed
|
CI Failed on Gaeac6 in Build# 1 |
|
Some gempak jobs hit their walltimes in the extended test on WCOSS2. I am rerunning them to confirm they ran long due to a machine issue. All other jobs and test cases completed without issue. |
emcbot
added
CI-Hera-Building
|
Found a fix that resolves the gfs_metp dependency issue by making the offset for part of the dependency adjust based on WCOSS2 (BACKUPSYS-C) global-workflow> git diff dev/workflow/rocoto/gfs_tasks.py
diff --git a/dev/workflow/rocoto/gfs_tasks.py b/dev/workflow/rocoto/gfs_tasks.py
index 89ac5e6e8..3b0a99dfd 100644
--- a/dev/workflow/rocoto/gfs_tasks.py
+++ b/dev/workflow/rocoto/gfs_tasks.py
@@ -1938,7 +1938,10 @@ class GFSTasks(Tasks):
dep_dict = {'type': 'cycleexist', 'condition': 'not', 'offset': offset}
deps2.append(rocoto.add_dependency(dep_dict))
- offset = timedelta_to_HMS(-to_timedelta(f'{6*lookback}H'))
+ edate_gfs = self._base['EDATE']
+ edate_metp = edate_gfs.replace(hour=18)
+ edate_metp_diff = edate_metp - edate_gfs
+ offset = timedelta_to_HMS(-to_timedelta(f'{edate_metp_diff}H'))
dep_dict = {'type': 'task', 'name': f'{self.run}_arch_vrfy', 'offset': offset}
deps2.append(rocoto.add_dependency(dep_dict))
deps.append(rocoto.create_dependency(dep_condition='and', dep=deps2))I updated the xml for the The extended CI case on WCOSS2 is still running after I booted the gempak jobs that hit their walltimes (reran fine within time) and it should finish in the next couple hours. |
|
@KateFriedman-NOAA thanks! please let me know if there's any code you want me to add to this branch, etc. And thanks for finding and fixing the bug! I just thought it was a feature ... oops!? |
emcbot
added
CI-Hera-Failed
|
Experiment C96C48mx500_S2SW_cyc_gfs FAILED on Hera in Build# 2 in |
emcbot
added
CI-Hera-Failed
|
CI Failed on Hera in Build# 2 |
Description
This PR has one minor bug fix for the stage IC job when and adds a low resolution test that has all components for cycling that is anticipated to be used in GFSv17.
Resolves #3441
Type of change
Change characteristics
How has this been tested?
This test was run on hera:
RUNTESTS=/scratch1/NCEPDEV/climate/Jessica.Meixner/addlowrestest/testlowres03
(As of posting this PR a few last jobs remained, but 1.5 cycles have completed successfully).
Update: 2.5 cycles succeeded.
Checklist