E3SM icon indicating copy to clipboard operation
E3SM copied to clipboard
verified publisher icon E3SM-Project

MPAS: Add instance string to output filenames

Open mkstratos opened this issue 3 years ago • 2 comments

To solve issue #5122:

  • Create a stream file for multiple instances of MPAS Ocean and Sea Ice
  • Adds the instance string to streams.ocean_* and streams.seaice_* output and restart filenames

mkstratos avatar Aug 10 '22 17:08 mkstratos

@akturner please review

rljacob avatar Aug 25 '22 17:08 rljacob

@akturner please review.

rljacob avatar Sep 15 '22 17:09 rljacob

@akturner please review.

rljacob avatar Oct 06 '22 17:10 rljacob

Compiled stand-alone MPAS-Ocean with gnu debug and intel debug. Passes nightly suite for both. This is not surprising because ocn_comp_mct.F is the only file altered in this PR that affects stand-alone. I tested

SMS_D_Ln9.T62_oQU120_ais20.MPAS_LISIO_TEST.cori-haswell_gnu

and it failed the run step, but the error was not clear. Since cori is nearing its end, I'll test on other machines.

mark-petersen avatar Oct 12 '22 15:10 mark-petersen

telecon notes: Needs another review.

rljacob avatar Nov 03 '22 17:11 rljacob

passes:

  • ERS.ne11_oQU240.WCYCL1850NS.chrysalis_intel
  • SMS_D_Ld3.T62_oQU120.CMPASO-IAF.chrysalis_intel
  • SMS_D_Ld1.ne30pg2_EC30to60E2r2.WCYCL1850.chrysalis_intel.allactive-wcprod

merged to next

jonbob avatar Dec 06 '22 23:12 jonbob

re-merged to next with fix for branch and hybrid tests. Now passes:

  • SMS_Ld1.ne30pg2_EC30to60E2r2.WCYCLSSP370.chrysalis_intel.allactive-wcprodssp
  • ERS.ne30pg2_EC30to60E2r2.WCYCLSSP370.chrysalis_intel.allactive-wcprodssp

jonbob avatar Dec 08 '22 17:12 jonbob

The current NCK test failures are related to the mpas files now having instance strings in their names, as they should, and not having baselines for those filenames.

jonbob avatar Dec 12 '22 17:12 jonbob

merged to master and NCK test DIFFs blessed

jonbob avatar Dec 12 '22 17:12 jonbob

Do we need to now bless the NCK test on all machines?

ndkeen avatar Dec 16 '22 02:12 ndkeen

@ndkeen -- I did bless them, but apparently the baselines need to be regenerated instead of simply blessed. @jgfouca and I worked on this yesterday and he came up with some commands to add to the bless in order to do so. So I made new requests for that yesterday, at least for chrysalis. The other machine reporting this failure is perlmutter and it's still down.

jonbob avatar Dec 16 '22 16:12 jonbob

@ndkeen - on today's cdash for chrysalis-next, it now appears as a NML DIFF instead of a regular DIFF. I submitted a request to bless the NML DIFF as well

jonbob avatar Dec 16 '22 16:12 jonbob

OK great -- you're on it. I just didn't want to lose the fact that I tracked down an issue to this PR and we had to re-investigate later.

Note, I learned you can add commenter:@me into github search to show issues/PR's that you commented on (which doesnt count as a mention).

ndkeen avatar Dec 16 '22 16:12 ndkeen