Porting regression tests to Orion

[MinsukJi-NOAA]

Issue #77

• This branch has merged in the latest develop
• I will run 3 separate regression tests and add the result files when completed:
  1. NEMSCompsetRun on Hera
  2. rt.sh on Hera
  3. rt.sh on Orion

[junwang-noaa]

Sorry, I confused myself, need a cup of coffee. For Orion porting, @ DeniseWorthen Minsuk' branch updates all the subcomponents, since current FV3 and NEMS have CEMPS related updates, do we need extra commits on CICE/MOM6 for s2s to run with latest FV3/NEMS? If *NO*, then let's update s2s to point to the latest version of all the components. If *YES*, then let's keep FV3 at aac9ff9 <https://github.com/NOAA-EMC/fv3atm/tree/aac9ff9da6914792f018f78c21d769c6336b8920> (removal ccpp dynamic build) and NEMS at 6e36027 <https://github.com/NOAA-EMC/NEMS/tree/6e36027d643e5153fc42694263bc7826dce2df08> with debug test changes, we will only make porting changes (updating mom6, cice5 + s2s) in this commit. Hope it is clear now. On Mon, May 4, 2020 at 9:57 AM Jun Wang - NOAA Federal <jun.wang@noaa.gov> wrote:

…

I think at last Tuesday's meeting, the plan is to get Orion porting in first, then we have a ocean-wave coupling commit (is it tested and ready for commit yet?, it will be running on hera and orion. It can be run on phase2 when phase2 porting is finished). Then we will have several commits for CMEPS transition. For Orion porting, @ DeniseWorthen Minsuk' branch updates all the subcomponents, since current FV3 and NEMS have CEMPS related updates, do we need extra commits on CICE/MOM6 for s2s to run with latest FV3/NEMS? If yes, then let's update s2s to point to the latest version of all the components. If not, then let's keep FV3 at aac9ff9 <https://github.com/NOAA-EMC/fv3atm/tree/aac9ff9da6914792f018f78c21d769c6336b8920> (removal ccpp dynamic build) and NEMS at 6e36027 <https://github.com/NOAA-EMC/NEMS/tree/6e36027d643e5153fc42694263bc7826dce2df08> with debug test changes, we will only make porting changes (updating mom6, cice5 + s2s) in this commit. On Mon, May 4, 2020 at 9:40 AM Denise Worthen ***@***.***> wrote: > I'm now really confused---I made a new PR for MOM6 and CICE5 to be > updated with the orion ports but w/o FV3 updated. The last commit on FV3 > now for the PR (#80 > <#80>) to ufs-s2s is > the removal of the CCPP dynamic build. I thought this would allow you to > use the existing hera baselines since I had updated my original port to > orion ufs-s2s-model branch to the FV3 with the Chsp physics update. > > My original feature/orion port had updated FV3 because of an earlier > email from Jun asking me to update my branch and I think I must have > misunderstood which components I was being asked to update. > > — > You are receiving this because your review was requested. > Reply to this email directly, view it on GitHub > <#78 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AI7D6TMYV5SRKD6X3MX4NADRP3ATRANCNFSM4MXKY5RA> > . >

[DeniseWorthen]

MOM6 and CICE5 do not need to be updated to run with the latest FV3/NEMS (the cmeps related changes).

The original feature/orion branch of s2s-model https://github.com/DeniseWorthen/ufs-s2s-model/tree/feature/orion has MOM6,CICE5 updated to develop in each case, which includes the orion ports. It also has FV3/NEMS updated to the current develop in each case (the cmeps related changes).

For CMEPS, I need to commit the restart features to CICE (1 commit to ufs-s2s, new baseline required). It can happen prior to the actual CMEPS commit; I have only been waiting until after the porting commits, the rt.sh commit and the dropping of ipd.

To bring in CMEPS, a second commit would then be required to ufs-s2s (new baseline required) with the CMEPS related tests using rt.sh.

[junwang-noaa]

Thanks for the clarification. So here is what we have: the latest revisions of all the subcomponents have porting changes already, s2s can run when pointing to the latest revisions of the subcomponent and RT results will change because of the FV3 physics updates and ww3 code changes in develop branch. For this Orion porting commit, we need to point to a ww3 revision with porting updates, and it will change RT results anyway. So I would suggest to use Minsuk's feature/orion branch (it already points to the latest revisions of all the subcomponents), create new baseline on hera and Orion and run rt.sh for RT and then commit the code. Minsuk, can you reopen your PR? Thanks

…

On Mon, May 4, 2020 at 10:17 AM Denise Worthen ***@***.***> wrote: MOM6 and CICE5 do not need to be updated to run with the latest FV3/NEMS (the cmeps related changes). The original feature/orion branch of s2s-model https://github.com/DeniseWorthen/ufs-s2s-model/tree/feature/orion has MOM6,CICE5 updated to develop in each case, which includes the orion ports. It also has FV3/NEMS updated to the current develop in each case (the cmeps related changes). For CMEPS, I need to commit the restart features to CICE (1 commit to ufs-s2s, new baseline required). It can happen prior to the actual CMEPS commit; I have only been waiting until after the porting commits, the rt.sh commit and the dropping of ipd. To bring in CMEPS, a second commit would then be required to ufs-s2s (new baseline required) with the CMEPS related tests using rt.sh. — You are receiving this because your review was requested. Reply to this email directly, view it on GitHub <#78 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AI7D6TJ7AKVY4A2MUZQ5GBLRP3E6BANCNFSM4MXKY5RA> .

[MinsukJi-NOAA]

On Hera, if we generate Baseline using rt.sh, and run NEMSCompsetRun RT against it, there will be a file missing error for REGTEST-FINGERPRINT.md because rt.sh does not create this file when generating Baseline. Either we need to remove verify_fingerprint in NEMS/tests/rtgen, or stick to generating Baseline using NEMSCompsetRun.

[DeniseWorthen]

Looks good.

[junwang-noaa]

Let's remove that. It tried to track the code changes with baseline. But it affects model portability (I got problems on phase2 when there is no matching baseline on the platform).

…

On Tue, May 5, 2020 at 7:07 AM Minsuk Ji ***@***.***> wrote: On Hera, if we generate Baseline using rt.sh, and run NEMSCompsetRun RT against it, there will be a file missing error for REGTEST-FINGERPRINT.md because rt.sh does not create this file when generating Baseline. Either we need to remove verify_fingerprint in NEMS/tests/rtgen, or stick to generating Baseline using NEMSCompsetRun. — You are receiving this because your review was requested. Reply to this email directly, view it on GitHub <#78 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AI7D6TPR25U272XXTIWRX3LRP7XO7ANCNFSM4MXKY5RA> .

[MinsukJi-NOAA]

I will open an issue in NEMS repo regarding this.

[junwang-noaa]

May I ask why we comment out these two debug tests?

[MinsukJi-NOAA]

Debug tests are not implemented in rt.sh yet. Since the new baselines (20200504) were generated using rt.sh, I had to comment those lines.

[junwang-noaa]

I see, let's make a separate commit for the debug test then. At this time we will keep running NEMSCompsetRun and rt.sh until the debug tests are in rt.sh.

…

On Tue, May 5, 2020 at 8:46 AM Minsuk Ji ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In compsets/all.input <#78 (comment)> : > @@ -23,8 +23,8 @@ run cpld_fv3_ccpp_384_mom6_cice_1d_bmark_rt @ fv3, coupledapp, warm384, physics, run cpld_fv3_ccpp_384_mom6_cice_ww3_cold_bmark_rt @ fv3, coupledapp, coldBM, physics, bm, ww3 run cpld_fv3_ccpp_384_mom6_cice_ww3_1d_bmark_rt @ fv3, coupledapp, warm384, physics, bm, ww3 -run cpld_fv3_ccpp_mom6_cice_cold_debug @ fv3, coupledapp, cold, physics, debug -run cpld_fv3_ccpp_mom6_cice_6h_debug @ fv3, coupledapp, warm, physics, debug +#run cpld_fv3_ccpp_mom6_cice_cold_debug @ fv3, coupledapp, cold, physics, debug Debug tests are not implemented in rt.sh yet. Since the new baselines (20200504) were generated using rt.sh, I had to comment those lines. — You are receiving this because your review was requested. Reply to this email directly, view it on GitHub <#78 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AI7D6TK4PYGCZO3JHG667FLRQADBPANCNFSM4MXKY5RA> .

[MinsukJi-NOAA]

For this PR, would it be safer if we generated the new baselines on Hera using nemscompsetrun? And start using rt.sh for baselines when the NEMS issue (verify fingerprint) & debug cases are taken care of?

[junwang-noaa]

Yes, that would be good, this way we will keep the 2 debug test results.

…

On Tue, May 5, 2020 at 8:55 AM Minsuk Ji ***@***.***> wrote: For this PR, would it be safer if we generated the new baselines on Hera using nemscompsetrun? And start using rt.sh for baselines when the NEMS issue (verify fingerprint) & debug cases are taken care of? — You are receiving this because your review was requested. Reply to this email directly, view it on GitHub <#78 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AI7D6TONWYLCFTCJTHNZZ2DRQAEEZANCNFSM4MXKY5RA> .

[DeniseWorthen]

Is this PR ready to merge?

[MinsukJi-NOAA]

Is this PR ready to merge?
RT using rt.sh is pending

[DeniseWorthen]

OK, great. I will wait for confirmation all tests you intend to run are complete.

[MinsukJi-NOAA]

@DeniseWorthen , all tests are now complete!