[Ncep.list.fv3-announce] Update on FV3 issues on Theia and Luna

Samuel Trahan - NOAA Affiliate samuel.trahan at noaa.gov
Thu Oct 5 17:11:12 UTC 2017

Hi all,

There may be some NEMS FV3 developers not on the NEMS list.  This is for
their benefit.  If you don't use NEMSfv3gfs on Theia or Luna, you can
safely ignore this email.

There are known issues on Luna and Theia due to recent machine upgrades.
Many issues will impact FV3 developers.  Below is the status report on
fixing those issues, and any fixes or workarounds already available.  None
of these issues are specific to the FV3; they affect other dynamical cores
as well.

Sam Trahan

---------- Forwarded message ----------
From: Samuel Trahan - NOAA Affiliate <samuel.trahan at noaa.gov>
Date: Thu, Oct 5, 2017 at 1:02 PM
Subject: Update on NEMS issues on Theia and Luna
To: NEMS developers <ncep.list.nems.announce at lstsrv.ncep.noaa.gov>

Hi all,

Updates on the Theia and Luna NEMS issues and related dev tool issues.

Three dynamical cores are having trouble running in NEMS on Theia: MOM (5 &
6), FV3, and GSM.  We have some ideas of why this may be happening, and
we're trying to narrow down the possibilities and fix the problem.  Any
further updates will be sent out as soon as we know more.

A few people have complained about trouble compiling or using dev tools on
WCOSS Cray after the filesystem move.  These are the known issues and fixes:

1. Wide variety of error messages about modulefiles or inability to find
ftn.  Log out and log back in and this should be fixed.  If not, please
submit a ticket to the wcoss helpdesk with as much information as you can

2. The emacs 24.5 installed in /gpfs/hps3/emc/hwrf/noscrub/soft.  When
emacs is installed, the path to the installation area is hard-coded several
places.  I have installed emacs 25.3 in that area.  If you use "module load
emacs" or "module load emacs/25.3" you should get a new version that works.

3. The git 2.6.2 in /gpfs/hps3/emc/hwrf/noscrub/soft/ won't work.
Apparently git also has hard-coded paths.  The latest release, 2.14.2, is
installed in that area as of ~12:30 PM Eastern time today.  I'm testing
that now.  If you have time, please test it and let me know if you run into

4. Another git issue: you may find hard-coded paths to /gpfs/hps/emc/ (or
cpc or swpc or...) in your ~/.gitconfig, ~/.emacs, ~/.profile, and other
~/. files.  If so, you will need to fix those.

5. ESMF-linked executables.  Some versions of esmf use the -rpath option to
insert hard-coded library paths in the executable.  If that happens, you
will need to recompile.  Nobody  has reported this problem yet, but if you
do get runtime linker errors (not compile time) about esmf being missing,
try recompiling.

6. Some /nems/noscrub regression test files are missing.  This is because
/nems/noscrub has hit quota.  We're waiting for the main offenders to
delete files so we can finish the rsync from Gyre to Tide.  Until then,
some of the apps will be unable to run their compsets, or will be able to
run but will be unable to verify against the (missing) baseline.

I know some people like good news.  ESMF compile time should work.  I did
replace /gpfs/hps/emc with /gpfs/hps3/emc in all of the esmf makefiles and
include files in emc.nemspara's areas, ESMF should be happy at compilation
and compile-time linker stages.  All other software in emc.nemspara's soft/
directory should work too.

Sam Trahan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://www.lstsrv.ncep.noaa.gov/pipermail/ncep.list.fv3-announce/attachments/20171005/0463de4f/attachment.html 

More information about the Ncep.list.fv3-announce mailing list