[Ncep.list.nems.announce] upcoming NEMS and FMS commit

Wed Nov 14 16:30:58 UTC 2018

Hi all,

The target commit time is now Thursday between noon and 2:30 PM Eastern
time.  The delay is because some of my files were deleted from WCOSS before
I could see the results.

Long version...

Last night, I was almost done with the final tests of the aforementioned
NEMS commit, but something deleted all of my files under emc.nemspara for
WCOSS Phase 1 & 3 before I could get the test results.  The files were less
than twelve hours old, so this was either from the scrubber going crazy, or
someone manually deleting the files.  Regardless, I'm going to have to redo
the vanished tests.  Being a pessimist, I'm planning the commit tomorrow,
Thursday November 15, between noon and 2:30 PM Eastern time.  I need the
unusually long time window (2.5 hours) because three framework repositories
are being updated (produtil, NEMS, and FMS).

As promised, the gory details of the commit are below.

Sincerely,
Sam Trahan

*Updates to add WCOSS Phase 3 (Dell) support in NEMS, NEMSfv3gfs, and
produtil*

All of these were from Sam Trahan:

1. produtil.cluster can detect WCOSS Phase 3, and report whether it is the
operational side
2. produtil.run's automatic MPI chooser is updated to find IMPI on WCOSS
Phase 3.
3. NEMS/tests/rtgen is updated to know how to autodetect the dell1/2/3 ptmp
& stmp with the most available space.
4. NEMS/src/incmake is updated for WCOSS Phase 3 awareness.
5. NEMSfv3gfs/compsets is updated to have baseline locations, Rocoto
information, and node sizes, for WCOSS Phase 3.

*FMS Updates*

All of these updates are from Dom Heinzeller.  This is verbatim from Dom's
pull request to the GFDL FMS master:

"This PR adds missing/modifies existing/removes no longer needed
portability features from the GFS-FMS branch with the goal to minimize the
drift from NOAA-GFDL FMS master:

   - "mpp/include/mpp_gather.h: bugfix/workaround for Intel18 compiler,
   already in NOAA-GFDL/FMS master and required to run the fv3_stretched_nest
   tests with Intel18
   - "mpp/affinity.c: use empty functions to set and get affinity on MacOSX
   platforms instead of removing functions entirely; this is a modified
   version of the changes to mpp/affinit.c that was created based on the
   request by Rusty, currently prepared for merging into NOAA-GFDL/FMS master
   - "mpp/include/mpp_domains_define.inc: revert workaround previously
   introduced for gfortran, no longer required; this change removes a
   workaround that was needed several months back to mitigate a GNU crash when
   starting up the model; it turns out that this is no longer required, most
   likely because the crash in that particular file was only a symptom for an
   issue that had to do with the ESMF version used back then (pre 7.1.0); that
   workaround never made it into NOAA-GFDL/FMS master"

*Other Updates*

These updates come from various individuals, so they're detailed in
parentheses.

1. Baselines are generated for NEMSfv3gfs on all Jets, GAEA, and all four
WCOSS phases.  NEMSfv3gfs compsets are updated with new baseline
locations.  (This update is from Sam Trahan.)

2. The t/u Jet now runs the fv3_wrtGauss_nemsio_c768.  (This update was
modified from Terry McGuinness's FV3 GFS workflow port to Jet.)

3. Some invalid spaces were replaced with a tab in the GSDCHEM compilation
logic in NEMS (This update is from Raffaele Montuoro.)

4. Configuration file dependencies in NEMS/src/incmake/configure*.mk files
are corrected (This is Sam Trahan's resolution of bugs identified by Dusan
Jovic.)

5. Some bash-specific logic in Makefiles was replaced with POSIX sh
compliant code.  (This update is from Dusan Jovic.)

6. The temporary file components.mk is now deleted before the
NEMS/GNUmakefile exits.  (This is Sam Trahan's resolution of bugs
identified by Dusan Jovic.)

7. The tests-results.mk is handled as a regular configuration file (like
configure.nems or modules.nems) and is cleaned up by "make unconfigure."
(This is Sam Trahan's resolution of bugs identified by Dusan Jovic.)

8. Some "echo" statements were replaced by "@echo" so that the echo command
is not printed.  (This update is from Dusan Jovic.)

On Fri, 9 Nov 2018 at 15:11, Samuel Trahan - NOAA Affiliate <
samuel.trahan at noaa.gov> wrote:

> Hi all,
>
> There is a batch of NEMS and FMS commits coming in soon, focused on WCOSS
> Phase 3 support and build system bug fixes.  The changes have already been
> fully tested with all apps before today's FV3 commit.  I'm going to have to
> re-run the NEMSfv3gfs tests before I can commit, and I don't like Friday
> commits.  Thus, the commit will happen sometime next week, probably
> Wednesday.
>
> The branch with the changes is dell-produtil in NEMSfv3gfs, NEMS, and
> NCEPLIBS_pyprodutil.  Overview of the changes:
>
> 1. WCOSS Phase 3 (Dell) support for NEMSCompsetRun and NCEPLIBS_pyprodutil
> (tested with NEMSfv3gfs)
>
> 2. Portability changes in FMS (WCOSS Phase 3 support requires these)
>
> 3. Various build system bug fixes in NEMS/src/incmake
>
> 4. Enable an additional NEMSfv3gfs regression test on t & u Jets
>
> I will refrain from giving the detailed, point-by-point, breakdown until
> after I retest with the new FV3.
>
> Sincerely,
> Sam Trahan
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://www.lstsrv.ncep.noaa.gov/pipermail/ncep.list.nems.announce/attachments/20181114/76c771e1/attachment.html 