netcdf-c: Error accessing GOES-16 data over opendap

I’ve run into another issues hitting opendap from the TDS:

ncdump -h https://thredds.unidata.ucar.edu/thredds/dodsC/satellite/goes/east/grb/ABI/CONUS/Channel13/current/OR_ABI-L1b-RadC-M6C13_G16_s20200302236114_e20200302238499_c20200302238530.nc

gives

ncdump: https://thredds.unidata.ucar.edu/thredds/dodsC/satellite/goes/east/grb/ABI/CONUS/Channel13/current/OR_ABI-L1b-RadC-M6C13_G16_s20200302236114_e20200302238499_c20200302238530.nc: https://thredds.unidata.ucar.edu/thredds/dodsC/satellite/goes/east/grb/ABI/CONUS/Channel13/current/OR_ABI-L1b-RadC-M6C13_G16_s20200302236114_e20200302238499_c20200302238530.nc:
NetCDF: Not a valid data type or _FillValue type mismatch

If I download that same file and open locally, it works fine. I’m not sure if this is a recurrence of the issue in #1316 or not. This is happening for me on netCDF-C 4.7.3, installed from conda-forge on macOS 10.15.3. I should note that I’m not having problems with OPeNDAP on any other datasets on thredds.unidata.ucar.edu.

About this issue

  • Original URL
  • State: closed
  • Created 4 years ago
  • Comments: 27 (16 by maintainers)

Commits related to this issue

Most upvoted comments

I don’t understand how this file violates the fill value rules.

However, I believe the fill value checks are a bug and should be removed. As @dopplershift points out, these rules make files and software that used to work, fail.

I think that fill value type checking is a good idea that came 25 years too late. Now netcdf is stuck with all the existing files and software that did not follow the fill value rules.