WRF compilation error on a CRAY machine

[acast]

Dear users,
I am having issues compiling WRF on an NOAA HPC (Gaea) using a Cray machine. uname -a produces "Linux gaea12 4.12.14-197.99-default"
I am new to this kind of supercomputer so I would like to know if someone here can help. I already contacted the HPC system administrators and they did not provide much help.
When compiling WRF I try with option 46 (CRAY CCE (ftn $(NOOMP)/cc): Cray XE and XC) and I get the following:
" Not sure how to figure out the version of this compiler ftn"

I attach my .log file with all the errors and details.
Any help on how to start debugging these errors is appreciated.
Thanks,

Alma

 

[kwerner]

Hi,
Can you issue a 'clean -a' and then reconfigure and recompile with the command

./compile em_real -j 1 >& compile.log

and then attach that full compile.log file, along with the configure.wrf file? Thanks!

 

[acast]

Hi Kelly,

I attach my compile.log and configure.wrf files.
Thanks,
Alma

 

[kwerner]

Alma,
Thanks for sending those. I, myself, have never worked with a Cray system, nor do I have access to one, but one of the first error messages is

ftn -hnoomp -c -s integer32 -s real`expr 8 \\* 4`  -O1 -Ofp1 -Oipa0 -Onomodinline -N1023 -f free -h byteswapio   -I../dyn_em -I../dyn_nmm  -I/ncrc/home1/Alma.Castillo.Trujillo/models/COAWST/WRF/external/esmf_time_f90  -I/ncrc/home1/Alma.Castillo.Trujillo/models/COAWST/WRF/main -I/ncrc/home1/Alma.Castillo.Trujillo/models/COAWST/WRF/external/io_netcdf -I/ncrc/home1/Alma.Castillo.Trujillo/models/COAWST/WRF/external/io_int -I/ncrc/home1/Alma.Castillo.Trujillo/models/COAWST/WRF/frame -I/ncrc/home1/Alma.Castillo.Trujillo/models/COAWST/WRF/share -I/ncrc/home1/Alma.Castillo.Trujillo/models/COAWST/WRF/phys -I/ncrc/home1/Alma.Castillo.Trujillo/models/COAWST/WRF/wrftladj -I/ncrc/home1/Alma.Castillo.Trujillo/models/COAWST/WRF/chem -I/ncrc/home1/Alma.Castillo.Trujillo/models/COAWST/WRF/inc -I/opt/cray/pe/netcdf/4.6.3.2/CRAYCLANG/9.0/include -I  -I/usr/local/mct/include module_state_description.f90\
ftn-2301 crayftn: ERROR in command line\
  The "-I" option must be followed by an include directory.\
make[2]: [../configure.wrf:562: module_state_description.o] Error 1

You can see in the line above the ERROR statement, there is a single "-I" without a path connected to it. Can you check your environment settings to see if you have an extra "-I" that needs to be removed?

 

[acast]

Hi Kelly,
So I just managed to compile WRF but not sure if what I did will produce any errors later. The first Error on the compilation.log that I sent you was produced because the compiler did not understand the flags as shown below:

which FC
which: invalid option -- 'h'
which: invalid option -- 'n'
which: invalid option -- 'o'
which: invalid option -- 'o'
which: invalid option -- 'm'
which: invalid option -- 'p'

So I modified the WRF/arch/configure.defaults file and changed DM_FC = ftn $(NOOMP) to DM_FC = ftn. The compilation was successful and I created the appropriate .exe files.

Now, as expected, I was not able to compile WPS with grib2 after compiling WRF.
This is the error I get:
/opt/cray/pe/cce/9.1.1/binutils/x86_64/x86_64-pc-linux-gnu/bin/ld: cannot find -ljasper^M
/opt/cray/pe/cce/9.1.1/binutils/x86_64/x86_64-pc-linux-gnu/bin/ld: cannot find -lpng^M
make[1]: [Makefile:19: ungrib.exe] Error 1 (ignored)^M

I already set my JASPERLIB and JASPERINC environmental variables so not really sure what other things I can check to debug this error. I attach my full compile wps log file here.
echo $JASPERLIB /usr/lib64
echo $JASPERINC /usr/include
I set those variables because the sys admin told me the libraries are there and they have not helped much after that. Do you want me to add this to a separate post?

Thanks

 

[kwerner]

Alma,

That's great to hear you were able to get WRF compiled! As long as it compiled and you have the executables in the main/ directory, there shouldn't be any future issues due to the compilation.

As for the WPS error, yes, please post that question in the WPS section of the forum. We like to keep the inquiries separate to help searchability and readability in the future. Thanks!