Convert ncdiag to IODA is now broken

[rtodling]

Not clear whether the changes the obs group has been talking about associated w/ the format of the sat bias files has come into JEDI-dev ... but that is now braking in swell and the converters.

[Dooruk]

Can you share the error message please?

[rtodling]

Is there an auto test for the converters? I mean is the ncdiag converter in the tier 1 set of tests?

[Dooruk]

Tier 2 tests does run convert ncdiag. So it should fail tonight and I would be able to see the error message then.

Can you see this regarding CI-workflows please, I could use your input.

[rtodling]

Tier 2 tests does run convert ncdiag. So it should fail tonight and I would be able to see the error message then.

Can you see this regarding CI-workflows please, I could use your input.

ok, if so, then it should have failed last night, no?

[rtodling]

This is the error:

Diag.toGeovals(self.cycle_dir())
File "/discover/nobackup/rtodling/SwellExperiments/swell-convert_ncdiags/jedi_bundle/build/lib/python3.10/pyiodaconv/gsi_ncdiag.py", line 805, in toGeovals
"nlevs", self.df.dimensions["atmosphere_pressure_coordinate_arr_dim"].size)
KeyError: 'atmosphere_pressure_coordinate_arr_dim'
2024-05-06T21:18:43Z CRITICAL - failed/ERR

The code is looking for a variable that does not exist - perhaps JEDI/EMC added a variable we don't have in the NCdiag files! Although that's a very general field!

[gmao-jjin3]

Don't see that failure in creating Geovals in my swell-ufo test after updating my JEDI.

[Dooruk]

@rtodling Tier2 tests are running convert-ncdiags successfully, I can point you to the folder. Are you sure you didn't change anything locally?

[Dooruk]

Closing this as the issue was not reproducible by other users.