What version of the code are you using?
release-cesm2.2.2-0-g779b0a3
Have you made any changes to files in the source tree?
None
Describe every step you took leading up to the problem:
I am working on porting CESM 2.2.2 to the Rorqual machine of the Digital Research Alliance of Canada (Rorqual - Alliance Doc). I have significantly modified the config_*.xml files, and have attached them to this post. My config_compilers.xml file is based on the GNU compiler template included with cesm 2.2.2
My process is as follows:
./create_newcase --case $SCRATCH/cases/aqua_Ptest_v3 --res f19_f19_mg17 --compset QPC6
./case.setup
./case.build --clean-all
./case.build
Which then fails at the Building cesm stage with error:
make: *** [/scratch/calvinc/cases/aqua_Ptest_v3/Tools/Makefile:985: /scratch/calvinc/cime_output/aqua_Ptest_v3/bld/cesm.exe] Error 1
However, there were several -Wmissing-include-dirs warnings in the mct build log that looked like this:
f951: Warning: Nonexistent include directory ‘/scratch/calvinc/cime_output/aqua_Ptest_v2/bld/gnu/openmpi/nodebug/nothreads/mct/mct/noesmf/c1a1l1i1o1r1g1w1i1e1/include’ [-Wmissing-include-dirs]
f951: Warning: Nonexistent include directory ‘/scratch/calvinc/cime_output/aqua_Ptest_v2/bld/gnu/openmpi/nodebug/nothreads/mct/finclude’ [-Wmissing-include-dirs]
I have confirmed that the first of those two directories does exist, but the second one does not. I have attached both the cesm build log and the mct build log.
Describe your problem or question:
I believe there is an issue with the way I have defined my environment variables in my config_machines.xml file, or with my linker flags in my config_compilers.xml file. I don't have much experience with compilation issues such as this, and I couldn't find anything similar to this on the forums. I am using the GNU compilers as there is an interesting quirk in the Rorqual toolchain in which our BTL protocol is only compiled with GNU, and I would prefer not to mix the Intel compilers with a GNU compiled BTL protocol. I look forward to hearing from someone and discussing this build issue further! I am more than happy to provide additional context or files if necessary.
Thank you!
release-cesm2.2.2-0-g779b0a3
Have you made any changes to files in the source tree?
None
Describe every step you took leading up to the problem:
I am working on porting CESM 2.2.2 to the Rorqual machine of the Digital Research Alliance of Canada (Rorqual - Alliance Doc). I have significantly modified the config_*.xml files, and have attached them to this post. My config_compilers.xml file is based on the GNU compiler template included with cesm 2.2.2
My process is as follows:
./create_newcase --case $SCRATCH/cases/aqua_Ptest_v3 --res f19_f19_mg17 --compset QPC6
./case.setup
./case.build --clean-all
./case.build
Which then fails at the Building cesm stage with error:
make: *** [/scratch/calvinc/cases/aqua_Ptest_v3/Tools/Makefile:985: /scratch/calvinc/cime_output/aqua_Ptest_v3/bld/cesm.exe] Error 1
However, there were several -Wmissing-include-dirs warnings in the mct build log that looked like this:
f951: Warning: Nonexistent include directory ‘/scratch/calvinc/cime_output/aqua_Ptest_v2/bld/gnu/openmpi/nodebug/nothreads/mct/mct/noesmf/c1a1l1i1o1r1g1w1i1e1/include’ [-Wmissing-include-dirs]
f951: Warning: Nonexistent include directory ‘/scratch/calvinc/cime_output/aqua_Ptest_v2/bld/gnu/openmpi/nodebug/nothreads/mct/finclude’ [-Wmissing-include-dirs]
I have confirmed that the first of those two directories does exist, but the second one does not. I have attached both the cesm build log and the mct build log.
Describe your problem or question:
I believe there is an issue with the way I have defined my environment variables in my config_machines.xml file, or with my linker flags in my config_compilers.xml file. I don't have much experience with compilation issues such as this, and I couldn't find anything similar to this on the forums. I am using the GNU compilers as there is an interesting quirk in the Rorqual toolchain in which our BTL protocol is only compiled with GNU, and I would prefer not to mix the Intel compilers with a GNU compiled BTL protocol. I look forward to hearing from someone and discussing this build issue further! I am more than happy to provide additional context or files if necessary.
Thank you!