Ghepardo

Thanks Mikael. You are absolutely right that I was running this on a node where I'd opened a session using srun. However, I have now retried this by connecting to...

Thanks Mikael. I'm afraid nothing seems to make any difference: I have tried your sbatch script, and I have even tried turning the Slurm daemon off on our build machine,...

Noted (and sorry) re response method. I've created that .env file and sourced it, and I've found a directory /home/apps/eb/build/netCDF/4.9.0/gompi-2022a. This contains no files and 2 subdirectories, easybuild_obj and netcdf-c-4.9.0....

Hmm, I'm pretty lost here. The log file shows: == 2022-08-09 12:52:18,019 run.py:233 INFO running cmd: OMPI_MCA_rmaps_base_oversubscribe=1 make test ...before the start of the list of the 227 tests. I...

OK thanks. My first attempt to run "make test" manually did in fact produce slightly more information than appeared in the Easybuild log file: it suggested I reran with some...

Sorry, I didn't address a couple of the points you made: >Like checking if there were any other tests that ran MPI and somehow worked, seeing what actually happens on...

I have noticed that this build will succeed for me if I comment out the following line in the easyconfig: ```runtest = 'test'``` I have no way of telling what...

I have discovered that 'runtest' is just an easyconfig keyword, and "runtest = 'test'" just says to run 'make test' after 'make'. I guess its absence from previous versions of...

Thanks @Micket that's really helpful. I reran the failing test with the timeout you suggested. The test now ran to conclusion - but, in taking over 4 hours to complete,...

@Micket running this rebuild myself now. This looks very promising - the testing for iteration #0 took 0:18:35 instead of the 4:37:23 previously. I think the testing for iteration #...