[Ncep.list.nems.announce] NEMSAppBuilder problems and fixes

Samuel Trahan - NOAA Affiliate samuel.trahan at noaa.gov
Wed Feb 7 18:03:32 UTC 2018


Hi all,

I apologize, but I wasn't able to finish testing these changes before the
WCOSS downtime.  I am tasked to do three hundred eighteen things right
now.  I am going to finish testing on Theia and Jet so that I can test on
WCOSS as soon as it is available again.  If this commit is needed as a dire
emergency, we may be able to test on the operational machine, but I would
rather not do that except as a last resort.

Sincerely,
Sam Trahan

On Mon, Feb 5, 2018 at 11:32 AM, Samuel Trahan - NOAA Affiliate <
samuel.trahan at noaa.gov> wrote:

> Hi all,
>
> I'm including one more NEMSAppBuilder update in my upcoming commit.
> Jessica Meixner gave me an update needed to build newer versions of WW3.
> Testing is underway.  I'm doing this under the "underscoreww3" branch in
> NEMS VLAB Git.
>
> Sincerely,
> Sam Trahan
>
> On Thu, Feb 1, 2018 at 3:08 PM, Samuel Trahan - NOAA Affiliate <
> samuel.trahan at noaa.gov> wrote:
>
>> Hi all,
>>
>> Two problems in the NEMSAppBuilder are causing a lot of trouble right
>> now.  I held off fixing these issues because it will be disruptive to
>> development, and we've had too many development disruptions already.
>>
>> Specifically:
>>
>> 1. The NEMSAppBuilder $FULL_MACHINE_ID used to find modulefiles and conf
>> files has underscores on WCOSS Cray (wcoss_cray) but dots on the IBM parts
>> of WCOSS (wcoss.phase1, wcoss.phase2).  This was done to match the strange
>> mix of conventions the various apps were using when the NEMSAppBuilder was
>> last updated.
>>
>> 2. The NEMSAppBuilder does not delete the configuration files before
>> running NEMS/src/configure.  This can cause it to inadvertently use old
>> configuration information if NEMS/src/configure fails.
>>
>> It turns out, for several weeks now, the NEMSAppBuilder has not been able
>> to build the NEMSfv3gfs on WCOSS phase 1 and phase 2 due to bug #1 listed
>> above.  The reason nobody noticed this (including the nightly tests) is
>> that the NEMSfv3gfs compile.sh was run first, for earlier compsets in the
>> same test suite.  That compile.sh left behind valid modules.nems,
>> externals.nems, and ESMFVersionDefine.h.  The NEMSAppBuilder was run next,
>> to test the fv3_appbuilder compset.  It failed to run configure because it
>> was looking for configure.fv3.wcoss.phase2 (or phase1) which did not
>> exist.  However, the NEMSAppBuilder didn't delete the configuration
>> information compile.sh left behind, so the build continued as if nothing
>> had gone wrong (bug #2 listed above).
>>
>> It is possible that all apps will need to make minor changes to their
>> compsets and appbuilder files as a result of fixing these two bugs.  I'm
>> going to try this myself on the supported apps first and report the results.
>>
>> Sincerely,
>> Sam Trahan
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://www.lstsrv.ncep.noaa.gov/pipermail/ncep.list.nems.announce/attachments/20180207/2ed4f7b7/attachment.html 


More information about the Ncep.list.nems.announce mailing list