Runtime error with simple GNSS RO test run

JEDI
1

I tried running my own simple test run of hofx for GNSS RO Bending Angle but failed and need some advice on what went wrong! Please help

Here are the details of my setup which follows that of the online tutorial linked here except using my own inputs and model geometry.

1) environment
running on a HPC server but using the following container supplied by JCSDA: jedi-gnu-openmpi-dev_latest.sif

2) inputs
a) some GFS FV3 output files that where lying around from an experiment run by the NCEP guys on our server
b) an ioda format GNSS RO file that I created myself using the ioda converters (see my other post about compiling that). This file seems perfectly fine to me and has data within it that fell within the DA window

3) error message
a) It complains about the ioda data file, specifically the geoid_height_above_reference_ellipsoid@MetaData variable. However I can dump that variable and it has quite reasonable numbers in it.

GnssroBndNBAM vars: 1 variables: bending_angle
Exception:      Reason: An exception occurred inside ioda while opening a variable.
        name:   geoid_height_above_reference_ellipsoid@MetaData
        source_column:  0
        source_filename:        /cw3e/mead/projects/cwp106/projects/hiaper/mjmurphy/modeling/jedi/src/fv3-bundle/ioda/src/engines/ioda/src/ioda/Has_Variables.cpp
        source_function:        virtual ioda::Variable ioda::detail::Has_Variables_Base::open(const string&) const
        source_line:    107

b) I’m not sure these matter as they are warnings

WARNING: There was an error initializing an OpenFabrics device.

  Local host:   comet-14-57
  Local device: mlx4_0
--------------------------------------------------------------------------
--------------------------------------------------------------------------
WARNING: Could not generate an xpmem segment id for this process'
address space.

The vader shared memory BTL will fall back on another single-copy
mechanism if one is available. This may result in lower performance.

  Local host: comet-14-57
  Error code: 2 (No such file or directory)

I appreciate any help you can provide,
mjm

2

Hi mjm,

Is this the post you referred to in your latest post (from Mon Oct 25). If so, has this been resolved? I can help if this is still an issue.

Steve

3

Hi @stephenh,

This Runtime error is not resolved. Not sure what is the source of the error, it seems to be the GNSS RO data file but I don’t see any issue with the file when I examine it. Please advise.

Michael

4

Hi @mjm,

I think the issue is that the ioda file that you created with the converter is still in the old ioda version 1 format, whereas the ioda reader wants the file in ioda version 2 format.

Most of the converters are still writing files in idoa v1 (version 1) format and we supplied an “upgrader” application with the ioda v2 release. The upgrader can be run on the file you created to covert it to ioda v2 format which will hopefully fix this issue.

The build should have produced the application in:

your_build_area/bin/ioda-upgrade.x

If you didn’t get this binary executable, then make sure you are cd’d into your_build_area and run make there.

Once you have the upgrader executable, then to convert your file, run:

your_build_area/bin/ioda-upgrade.x input_iodav1_file output_iodav2_file

You can look at your ioda files using ncdump. In the version 1 file, you should see a list of variables that look like latitude@MetaData, geoid_height_above_reference_ellipsoid@MetaData, etc. In the version 2 file, these variables should appear in groups where the MetaData group contains variable latitude, and the variable geoid_height_above_reference_ellipsoid, etc.

By the way, we have started the effort this week to update the converters to directly write out ioda v2 format so that the upgrader application would no longer be needed.

1 Like