Open ekluzek opened 2 weeks ago
@ekluzek , this test fails for me due to the nag netcdf library problem. I didn't see this in your list above.
SMS_Ld10_D_Mmpi-serial.CLM_USRDAT.I1PtClm60Bgc.izumi_nag.clm-default--clm-NEON-HARV (RUN)
Hmmm, I don't see that one in my ctsm5.3.009 testing, maybe that test is just on b4b-dev? In any case -- anything that's izumi_nag with mpi-serial is going to fail so belongs under this issue.
@jedwards4b has a fix to ccs_config that we'll bring in that will solve this for the latest tags when we bring in the submodule update with it. So latest dev tags will be OK.
For older tags I worked with Joseph and we pushed a simple safe change of a symlink for the netcdf mpi-serial library that allows these tests to work.
I'll do more tests with older tags to ensure this is the case -- but this should handle this both for old and new tags.
Brief summary of bug
The Nag compiler mpi-serial tests are now failing on Izumi after the hardware rebuild. This is due to having trouble finding the Nag NetCDF shared library at runtime.
General bug information
CTSM version you are using: ctsm5.2.009 (likely applies to all other baselines before it as well) Does this bug cause significantly incorrect results in the model's science? No
Configurations affected: nag mpi-serial cases (use mvapich for MPI to get around this)
Details of bug
The Nag compiler mpi-serial tests are now failing for baselines created after the hardware update. So the original baselines worked fine -- but now they fail with a runtime error.
Here's the list of tests that now fail:
ERS_D_Ld5_Mmpi-serial.1x1_vancouverCAN.I1PtClm50SpRs.izumi_nag.clm-CLM1PTStartDate (RUN)
ERS_D_Mmpi-serial_Ld5.1x1_brazil.I2000Clm50FatesRs.izumi_nag.clm-FatesCold (RUN)
SMS_D_Ld1_Mmpi-serial.f45_f45_mg37.I2000Clm50SpRs.izumi_nag.clm-ptsRLA (RUN)
SMS_D_Mmpi-serial_Ld5.5x5_amazon.I2000Clm60FatesRs.izumi_nag.clm-FatesCold (RUN)
SMS_Ld10_D_Mmpi-serial.CLM_USRDAT.I1PtClm60SpRs.izumi_nag.clm-default--clm-NEON-TOOL (RUN)
Important output or errors that show the problem
Here's the cesm log: