crooney@pdx_edu
New Member
I am attempting to port CESM to our HPC for a user. Currently scripts_regression_tests.py still produces 13 failures and 4 errors.
Hopefully someone here is more experienced with CESM and can identify where I have gone wrong with porting more quickly than I have been able to.
Our cluster uses SLURM and I am attempting to use gcc 6.3.0, mvapich2 2.2, and Python 2.7.13.
I have edited config_machines.xml, and config_batch.xml. I have not edited config_compilers.xml. All three xml files validate when tested with xmllint.All of the failures are in the phase SHAREDLIB_BUILD
I have attached the output of scripts_regression_tests, and both edited xml files. The machine I have added is called 'coeus'. The variable CESMDATAROOT=/scratch/cesm is set. If the logs of individual tests are needed I can provide those, or any other necessary information.
Thank you to anyone who is able to help.
___________________________________________________________________________________________________________________________________________________
UPDATE:After updating the config_machines and config_compilers xml files my previous errors have resolved. Netcdf was not being properly located or linked. I needed to add NETCDF_PATH to and the shared library -L${NETCDF_PATH}/lib config_compilers.xml, I am attaching the updated files in case someone else is experiencing similar issues.
Unfortunately I am now running into more errors. One of the output files makes it seem like a fortran file is being read like a bash script:
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 1: program: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 2: use: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 3: use: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 4: use: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 6: use: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 8: implicit: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 10: integer: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 11: syntax error near unexpected token `len=1024'
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 11: ` character(len=1024) :: arg = '' ! cmd-line argument'
I will attach one of the TestStatus.log files in full, this is the very end of that file which led me to the above output file:
/scratch/cesm/crooney/scripts_regression_test.20190203_184904/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205/run/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205.cpl.hi.0001-01-05-00000.nc.base did NOT match /scratch/cesm/crooney/scripts_regression_test.20190203_184904/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205/run/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205.cpl.hi.0001-01-05-00000.nc.da
cat /scratch/cesm/crooney/scripts_regression_test.20190203_184904/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205/run/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205.cpl.hi.0001-01-05-00000.nc.base.cprnc.out
___________________________________________________________________________________________________________________________________________________
UPDATE:
jedward's response solved the above issue. I built cprnc and edited config_machines.xml to contain the path to the executable:/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc
I am now encountering the error "Test compiler changeable at build time". If I gain any insight on why this error is happening I will post it here, but for now I plan on alerting users that they should create a new case if they have to switch compilers so that the use of this feature is avoided.
-Colleen
Hopefully someone here is more experienced with CESM and can identify where I have gone wrong with porting more quickly than I have been able to.
Our cluster uses SLURM and I am attempting to use gcc 6.3.0, mvapich2 2.2, and Python 2.7.13.
I have edited config_machines.xml, and config_batch.xml. I have not edited config_compilers.xml. All three xml files validate when tested with xmllint.All of the failures are in the phase SHAREDLIB_BUILD
I have attached the output of scripts_regression_tests, and both edited xml files. The machine I have added is called 'coeus'. The variable CESMDATAROOT=/scratch/cesm is set. If the logs of individual tests are needed I can provide those, or any other necessary information.
Thank you to anyone who is able to help.
___________________________________________________________________________________________________________________________________________________
UPDATE:After updating the config_machines and config_compilers xml files my previous errors have resolved. Netcdf was not being properly located or linked. I needed to add NETCDF_PATH to and the shared library -L${NETCDF_PATH}/lib config_compilers.xml, I am attaching the updated files in case someone else is experiencing similar issues.
Unfortunately I am now running into more errors. One of the output files makes it seem like a fortran file is being read like a bash script:
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 1: program: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 2: use: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 3: use: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 4: use: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 6: use: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 8: implicit: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 10: integer: command not found
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 11: syntax error near unexpected token `len=1024'
/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc.F90: line 11: ` character(len=1024) :: arg = '' ! cmd-line argument'
I will attach one of the TestStatus.log files in full, this is the very end of that file which led me to the above output file:
/scratch/cesm/crooney/scripts_regression_test.20190203_184904/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205/run/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205.cpl.hi.0001-01-05-00000.nc.base did NOT match /scratch/cesm/crooney/scripts_regression_test.20190203_184904/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205/run/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205.cpl.hi.0001-01-05-00000.nc.da
cat /scratch/cesm/crooney/scripts_regression_test.20190203_184904/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205/run/DAE.f19_f19.A.coeus_gnu.fake_testing_only_20190203_191205.cpl.hi.0001-01-05-00000.nc.base.cprnc.out
___________________________________________________________________________________________________________________________________________________
UPDATE:
jedward's response solved the above issue. I built cprnc and edited config_machines.xml to contain the path to the executable:/vol/apps/hpc/src/cesm-2.1.0/cime/tools/cprnc/cprnc
I am now encountering the error "Test compiler changeable at build time". If I gain any insight on why this error is happening I will post it here, but for now I plan on alerting users that they should create a new case if they have to switch compilers so that the use of this feature is avoided.
-Colleen