What version of the code are you using?
2.1.5
Have you made any changes to files in the source tree?
No changes
Describe every step you took leading up to the problem:
export CASE=run_03_b0000-00
cd
rm -rf $CASEROOT/$CASE
rm -rf $BLDRUN_DIR/$CASE
rm -rf $SHORT_ARCH/$CASE
cd $CIMEROOT/scripts
./create_newcase --case $CASEROOT/$CASE --res f19_g17 --compset B1850 | tee 01_newcase.out
mv 01_newcase.out $CASEROOT/$CASE
cd $CASEROOT/$CASE;
./case.setup | tee 02_setup.out
./preview_run | tee 03_preview.out
./case.build --debug | tee 04_build.out
If this is a port to a new machine: Please attach any files you added or changed for the machine port (e.g., config_compilers.xml, config_machines.xml, and config_batch.xml) and tell us the compiler version you are using on this machine.
Please attach any log files showing error messages or other useful information.
In directory "~/.cime" I have:
config_compilers.xml
config_machines,xml
config_batch.xml
Describe your problem or question:
Background
I had CESM 2.1.4 working a few months ago on Anemone (National Oceanography Centre, UK), but unfortunately the system has been upgraded to Rocky9. As my original setup then failed, I've started from scratch using 2.1.5. I found I had to make a number of changes to config_compilers.xml and config_machines.xml files. I also found that I had to include a <MPILIBS> line in the compilers file.
I am trying to use the intel mpi libraries (see my dot.bashrc file) but do not know whether the python scripts want <MPILIBS> ... </MPILIBS> to include nothing, mpi, intelmpi or impi. Anyway whatever I use, case.build fails at the gptl compilation step. The impi module provides libraries with names libmpi.*, libmpifort.* and a few others - so is <MPILIBS> mpi </MPILIBS>correct?
Details of compilation failure
From the gptl log and checking the bld directory and the Makefile in"CESM_2.1.5/cime/src/share/timing/" , the compilation failure appears to occur because SHAREDPATH is not set to "bld/intel/mpi/nodebug/nothreads". Instead it appears to be "bld/intel/ /nodebug/nothreads".
The first line of the log file shows:
gmake -f /home/djw/Downloads/NCAR/CESM_2.1.5/cime/src/share/timing/Makefile install -C /dssgfs01/scratch/djw/NCAR/CASES/run_03_b0000-00/bld/intel/ /nodebug/nothreads/gptl MACFILE=/dssgf
s01/working/djw/NCAR/CASES/run_03_b0000-00/Macros.make MODEL=gptl GPTL_DIR=/home/djw/Downloads/NCAR/CESM_2.1.5/cime/src/share/timing GPTL_LIBDIR=/dssgfs01/scratch/djw/NCAR/CASES/run_03_
b0000-00/bld/intel/ /nodebug/nothreads/gptl SHAREDPATH=/dssgfs01/scratch/djw/NCAR/CASES/run_03_b0000-00/bld/intel/ /nodebug/nothreads
A number of the parameters here includes "intel/ /nodebug/nothreads". The file manager also shows "itel/unnamed directory/nodebug/nothreads".
Later the script tries to copy a file to "bld/intel/include/", the system assumes "include" is a new file,
Later still the compilation fails because "bld/intel//include" is not a directory.
Looking at a pre-Rocky9 compilation, there an include directory at "bld/intel/mpi/debug/nothreads/include" and the log file shows "SHAREDPATH=/dssgfs01/scratch/djw/NCAR/CASES/cesm_run_02_b0005-08/bld/intel/mpi/debug/nothreads"
So what have I done, that has resulted in SHAREDPATH and the other parameters having the wrong form? Does it come from a poorly defined MPILIBS or is it something else?
All suggestions will be welcome.
Best regards,
David Webb.
2.1.5
Have you made any changes to files in the source tree?
No changes
Describe every step you took leading up to the problem:
export CASE=run_03_b0000-00
cd
rm -rf $CASEROOT/$CASE
rm -rf $BLDRUN_DIR/$CASE
rm -rf $SHORT_ARCH/$CASE
cd $CIMEROOT/scripts
./create_newcase --case $CASEROOT/$CASE --res f19_g17 --compset B1850 | tee 01_newcase.out
mv 01_newcase.out $CASEROOT/$CASE
cd $CASEROOT/$CASE;
./case.setup | tee 02_setup.out
./preview_run | tee 03_preview.out
./case.build --debug | tee 04_build.out
If this is a port to a new machine: Please attach any files you added or changed for the machine port (e.g., config_compilers.xml, config_machines.xml, and config_batch.xml) and tell us the compiler version you are using on this machine.
Please attach any log files showing error messages or other useful information.
In directory "~/.cime" I have:
config_compilers.xml
config_machines,xml
config_batch.xml
Describe your problem or question:
Background
I had CESM 2.1.4 working a few months ago on Anemone (National Oceanography Centre, UK), but unfortunately the system has been upgraded to Rocky9. As my original setup then failed, I've started from scratch using 2.1.5. I found I had to make a number of changes to config_compilers.xml and config_machines.xml files. I also found that I had to include a <MPILIBS> line in the compilers file.
I am trying to use the intel mpi libraries (see my dot.bashrc file) but do not know whether the python scripts want <MPILIBS> ... </MPILIBS> to include nothing, mpi, intelmpi or impi. Anyway whatever I use, case.build fails at the gptl compilation step. The impi module provides libraries with names libmpi.*, libmpifort.* and a few others - so is <MPILIBS> mpi </MPILIBS>correct?
Details of compilation failure
From the gptl log and checking the bld directory and the Makefile in"CESM_2.1.5/cime/src/share/timing/" , the compilation failure appears to occur because SHAREDPATH is not set to "bld/intel/mpi/nodebug/nothreads". Instead it appears to be "bld/intel/ /nodebug/nothreads".
The first line of the log file shows:
gmake -f /home/djw/Downloads/NCAR/CESM_2.1.5/cime/src/share/timing/Makefile install -C /dssgfs01/scratch/djw/NCAR/CASES/run_03_b0000-00/bld/intel/ /nodebug/nothreads/gptl MACFILE=/dssgf
s01/working/djw/NCAR/CASES/run_03_b0000-00/Macros.make MODEL=gptl GPTL_DIR=/home/djw/Downloads/NCAR/CESM_2.1.5/cime/src/share/timing GPTL_LIBDIR=/dssgfs01/scratch/djw/NCAR/CASES/run_03_
b0000-00/bld/intel/ /nodebug/nothreads/gptl SHAREDPATH=/dssgfs01/scratch/djw/NCAR/CASES/run_03_b0000-00/bld/intel/ /nodebug/nothreads
A number of the parameters here includes "intel/ /nodebug/nothreads". The file manager also shows "itel/unnamed directory/nodebug/nothreads".
Later the script tries to copy a file to "bld/intel/include/", the system assumes "include" is a new file,
Later still the compilation fails because "bld/intel//include" is not a directory.
Looking at a pre-Rocky9 compilation, there an include directory at "bld/intel/mpi/debug/nothreads/include" and the log file shows "SHAREDPATH=/dssgfs01/scratch/djw/NCAR/CASES/cesm_run_02_b0005-08/bld/intel/mpi/debug/nothreads"
So what have I done, that has resulted in SHAREDPATH and the other parameters having the wrong form? Does it come from a poorly defined MPILIBS or is it something else?
All suggestions will be welcome.
Best regards,
David Webb.