[MITgcm-devel] testreport on glacier

[Jean-Michel Campin]

Hi Gael,

Since we move Matt "global_oce_biogeo_bling" in the main verification/ dir
to become a standard tested experiment, it's causing little problems
for your tests on glacier:

1) your are testing all the experiments that match "gloval_oce_*"
 (to get both: global_oce_cs32 and global_oce_llc90) but this new one
 is also matching: "gloval_oce_*"
 so it's getting in the way.

2) when you run testreport, you don't specify a "command" to run
 (and it's OK since the default MPI command works). But when you run
 the restart, using ../tools/do_tst_2+2, you specify:
  -exe 'mpirun -np 24 ./mitgcmuv' 
 and this breaks the restart for "global_oce_biogeo_bling" (just because it
 compile and run on only 4 procs and not 24). So my guess is that if
 you did not use this "-exe" option there, just like this:
  ../tools/do_tst_2+2 -t 'global_oce_*' -mpi -a jm_c at mitgcm.org
 it would just work normally.

3) I am guessing that all the 4 testreports are run one after the other,
 withing the same MITgcm copy, and there is no "cleaning" after
 the restart test is done. And since the 1 restart test fails (see above)
 for experiment global_oce_biogeo_bling, then the next testreport 
 to run also fails because the "run" dir has not been cleaned-up.
 The way to fix this would be to run:
  ../tools/do_tst_2+2 -clean
 before the 2nd testreport.

Cheers,
Jean-Michel