ESMValGroup / ESMValTool

ESMValTool: A community diagnostic and performance metrics tool for routine evaluation of Earth system models in CMIP
https://www.esmvaltool.org
Apache License 2.0
215 stars 126 forks source link

`recipe_russell18jgr.yml` fails to run #3478

Closed schlunma closed 8 months ago

schlunma commented 8 months ago

3422 changed the provenance tracking for NCL diagnostics so that invalid plot paths raise an error. With this change, issues that have been present in the diagnostic now surface in recipe_russell18jgr.yml: For many diagnostics, no plot file has been written (at least since v2.9.0):

I also found this warning in the logs, which sounds very concerning and should probably be investigated:

2023-12-15 12:17:32,268 UTC [359555] WARNING Converted cube units from 'm-3' to 'm3'
(for file /work/bd0854/DATA/ESMValTool2/CMIP5_DKRZ/NOAA-GFDL/GFDL-ESM2M/historical/fx/ocean/fx/r0i0p0/v20130514/volcello/volcello_fx_GFDL-ESM2M_historical_r0i0p0.nc)

The raw data appears to be already in the correct units, so ESMValCore shouldn't convert these!

katjaweigel commented 8 months ago

Yes, I think it is broken since several releases, but it never surfaced because it said it completed successfully before the provenance change and every time missed to produce the same figures, so also the comparison to the release before looked ok at a first glance. But I have unfortunately no idea what is wrong with (and switched to use recipe_ocean_example.yml instead of trying to fix it, sorry!). It would be good, to get it running again, but I guess not worth/possible for the current release (since the missing figures were unnoticed/unreported for quite some time now).

axel-lauer commented 8 months ago

I think this commit https://github.com/ESMValGroup/ESMValTool/commit/548954cb4fb6fe9021078a0d78d9de99e46ec988 should fix the remaining provenance / no plot issues with the russell18jgr diagnostics. The fixes cover all of the above mentioned diagnostics: