Hera gnu build broken #831
Comments
|
Opened issue JCSDA/spack-stack#1483 to request a new GNU installation of spack-stack on Hera. |
|
Based on comments from @RatkoVasic-NOAA in GSI-utils #66, I made the following changes to a working copy of develop The first GSI build attempt failed with As a test I turned off the MGBF build and tried again. The build got further but failed again Line 107 of src/gsi/hybrid_ensemble_isotropic.F90 is in the block We do not use MGBF in any of our regional or global ctests. We do not use MGBF in the operational GDAS. Nonetheless, it seems we must compile MGBF in order to build gsi.x. My guess is that the gnu build will work once the gnu MGBF issues are resolved. Tagging @TingLei-NOAA and @ShunLiu-NOAA to confirm that we must compile MGBF in order to build gsi.x and enkf.x. If true we need to
|
|
Russ
I will check why it is a problem for gnu as soon as possible.
For mgbf in gsi that issue/pr was created because rtma is planning to use
it
Ting
…______________________________
Ting Lei
Physical Scientist, Contractor with Lynker in support of
EMC/NCEP/NWS/NOAA
5830 University Research Ct., Cubicle 2765
College Park, MD 20740
***@***.***
301-683-3624
On Fri, Jan 31, 2025 at 9:16 AM RussTreadon-NOAA ***@***.***> wrote:
Based on comments
<NOAA-EMC/GSI-utils#66 (comment)>
from @RatkoVasic-NOAA <https://github.com/RatkoVasic-NOAA> in GSI-utils
#66 <NOAA-EMC/GSI-utils#66>, I made the
following changes to a working copy of develop
modulefiles/gsi_hera.gnu.lua
-prepend_path("MODULEPATH", "/contrib/spack-stack/spack-stack-1.6.0/envs/gsi-addon-dev-rocky8/install/modulefiles/Core")
+prepend_path("MODULEPATH", "/scratch4/NCEPDEV/stmp/role.epic/spack-stack/spack-stack-1.6.0/envs/gsi-addon-dev-rocky8/install/modulef
iles/Core")
--Needed for openmpi build
-prepend_path("MODULEPATH", "/scratch1/NCEPDEV/jcsda/jedipara/spack-stack/modulefiles")
+--prepend_path("MODULEPATH", "/scratch1/NCEPDEV/jcsda/jedipara/spack-stack/modulefiles")
local python_ver=os.getenv("python_ver") or "3.11.6"
local stack_gnu_ver=os.getenv("stack_gnu_ver") or "9.2.0"
-local stack_openmpi_ver=os.getenv("stack_openmpi_ver") or "4.1.5"
+local stack_openmpi_ver=os.getenv("stack_openmpi_ver") or "4.1.6"
The first GSI build attempt failed with
[ 0%] Building Fortran object src/mgbf/CMakeFiles/mgbf.dir/jp_pkind2.f90.o
/scratch1/NCEPDEV/da/Russ.Treadon/git/gsi/develop/src/mgbf/jp_pkind2.f90:22:4:
22 | use mpi
| 1
Fatal Error: Cannot open module file 'mpi.mod' for reading at (1): No such file or directory
/scratch1/NCEPDEV/da/Russ.Treadon/git/gsi/develop/src/mgbf/jp_pkind.f90:22:4:
22 | use mpi
| 1
Fatal Error: Cannot open module file 'mpi.mod' for reading at (1): No such file or directory
compilation terminated.
As a test I turned off the MGBF build and tried again. The build got
further but failed again
79%] Building Fortran object src/gsi/CMakeFiles/gsi_fortran_obj.dir/cplr_regional_io.f90.o
[ 80%] Building Fortran object src/gsi/CMakeFiles/gsi_fortran_obj.dir/hybrid_ensemble_isotropic.F90.o
/scratch1/NCEPDEV/da/Russ.Treadon/git/gsi/develop/src/gsi/hybrid_ensemble_isotropic.F90:107:6:
107 | use mg_intstate
| 1
Fatal Error: Cannot open module file 'mg_intstate.mod' for reading at (1): No such file or directory
compilation terminated.
make[2]: *** [src/gsi/CMakeFiles/gsi_fortran_obj.dir/build.make:2610: src/gsi/CMakeFiles/gsi_fortran_obj.dir/hybrid_ensemble_isotropic.F90.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:203: src/gsi/CMakeFiles/gsi_fortran_obj.dir/all] Error 2
make: *** [Makefile:146: all] Error 2
Line 107 of src/gsi/hybrid_ensemble_isotropic.F90 is in the block
! For MGBF
use mg_intstate
use mg_timers
We do not use MGBF in any of our regional or global ctests. We do not use
MGBF in the operational GDAS. Nonetheless, it seems we must compile MGBF in
order to build gsi.x.
My guess is that the gnu build will work once the gnu MGBF issues are
resolved.
Tagging @TingLei-NOAA <https://github.com/TingLei-NOAA> and @ShunLiu-NOAA
<https://github.com/ShunLiu-NOAA> to confirm that we must compile MGBF in
order to build gsi.x and enkf.x. If true we need to
1. fix MGBF so that it compiles with gnu, or
2. modify the GSI build so that MGBF is an optional component
—
Reply to this email directly, view it on GitHub
<#831 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/APEFS7HQHRIXQFQJTTRN5ZL2NOAR7AVCNFSM6AAAAABWFTIBD2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMRXGQ2TONBWHA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
Thank you for reminding me of GSI issue #765. This issue remains open. |
|
@RussTreadon-NOAA It seems the mpi dependency is not specified in the mgbf cmake file. Seems stack_gnu_ver and so on are not specified correctly. |
|
@TingLei-NOAA , did you make the following changes in your working copy of I made the above changes to a working copy of GSI
|
|
@RussTreadon-NOAA Thanks. Yes, I directly copied your gsi_hera.gnu.lua and resolved my problem. |
|
@TingLei-NOAA , thank you for fixing the mgbf gnu build. Please open a PR to get the changes from feature/mgbf4gnu into |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The
gsi_hera.gnu.luafromdevelopat 27c03e8 is missing-rocky8on the theprepend_pathline. Even after fixing this in a working copy ofdevelopthe gnu build fails with the message
This issue is opened to document the fact that the Hera gnu build is broken. This issue will remain open until the Hera gnu build is fixed or the decision is made to not support a GSI gnu build on Hera.
The text was updated successfully, but these errors were encountered: