Segmentation fault when reading grib

[roughnemezis]

What happened?

Using xarray.open_dataset(path, engine="cfgrib") on my grib2 file, I get a segmentation fault.

I tried in a virtualenv, installing only xarray and cfgrib pip install xarray cfgrib, same result

Versions installed are:

> pip freeze | grep -E "cfgrib|xarray|eccodes"
cfgrib==0.9.15.1
eccodes==2.44.0
eccodeslib==2.44.0.5
xarray==2025.6.1

Manually downgrading eccodes pip install eccodes==2.38.1 solves the issue, my file is properly opened

I join a sample grib file

What are the steps to reproduce the bug?

In a fresh virtualenv:

pip install xarray cfgrib

import xarray
path = 'path/to/my/grib'
xarray.open_dataset(path, engine="cfgrib")

-> segmentation fault

pip install eccodes==2.38.1

import xarray
path = 'path/to/my/grib'
xarray.open_dataset(path, engine="cfgrib")

-> no seg fault

Version

0.9.15.1

Platform (OS and architecture)

Ubuntu 22.04.5 LTS x86_64

Relevant log output

Accompanying data

https://filesender.renater.fr/?s=download&token=a3ee63af-c99b-4ace-907c-eed4c5d2217f

Organisation

MeteoFrance

[etienneschalk]

Hello, I am experimenting the same issue. I tested with the file mentioned in the README to ensure it was not my own GRIB file that was incorrect. (https://github.com/ecmwf/cfgrib?tab=readme-ov-file#usage). Thanks for the downgrading tip, it unblocked me. I downgraded eccodes from 2.44.0 to 2.38.1 and now I can read my GRIB files.

More info:

Common code snippet used for testing:

import cfgrib 
import eccodes
import xarray as xr 

print()
print()
print()
print(f"{cfgrib.__version__=}")
print()
print()
print()
print(f"{eccodes.__version__=}")
print()
print()
print()
xr.show_versions()
print()
print()
print()
xr.open_dataset(
    'era5-levels-members.grib',
    engine="cfgrib", 
    backend_kwargs={"indexpath": ""},
)

Test 1 - Not working

Dependencies:

    "cfgrib>=0.9.15.1",
    "eccodes>=2.44.0",

$ ipython
Python 3.13.7 (main, Sep 18 2025, 19:47:49) [Clang 20.1.4 ]
Type 'copyright', 'credits' or 'license' for more information
IPython 9.7.0 -- An enhanced Interactive Python. Type '?' for help.
Tip: You can use `%hist` to view history, see the options with `%history?`

   ...: print(f"{cfgrib.__version__=}")
   ...: print()
   ...: print()
   ...: print()
   ...: print(f"{eccodes.__version__=}")
   ...: print()
   ...: print()
   ...: print()
   ...: xr.show_versions()
   ...: print()
   ...: print()
   ...: print()
   ...: xr.open_dataset(
   ...:     'era5-levels-members.grib',
   ...:     engine="cfgrib",
   ...:     backend_kwargs={"indexpath": ""},
   ...: )



cfgrib.__version__='0.9.15.1'



eccodes.__version__='2.44.0'




INSTALLED VERSIONS
------------------
commit: None
python: 3.13.7 (main, Sep 18 2025, 19:47:49) [Clang 20.1.4 ]
python-bits: 64
OS: Linux
OS-release: 5.15.0-139-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: ('en_US', 'UTF-8')
libhdf5: 1.14.6
libnetcdf: 4.9.3

xarray: 2025.9.1
pandas: 2.3.3
numpy: 2.3.4
scipy: 1.16.3
netCDF4: 1.7.3
pydap: None
h5netcdf: 1.7.3
h5py: 3.15.1
zarr: None
cftime: 1.6.5
nc_time_axis: None
iris: None
bottleneck: None
dask: 2024.12.1
distributed: None
matplotlib: 3.10.7
cartopy: 0.24.1
seaborn: None
numbagg: None
fsspec: 2025.10.0
cupy: None
pint: None
sparse: None
flox: None
numpy_groupies: None
setuptools: 80.9.0
pip: None
conda: None
pytest: 8.4.2
mypy: 1.18.2
IPython: 9.7.0
sphinx: 8.2.3



Segmentation fault (core dumped)

Test B

Dependencies: (I downgraded to eccodes 2.38.1)

uv pip uninstall eccodes
uv pip uninstall eccodeslib # This was installed too in my venv, with a version 2.44.x.x

    "cfgrib>=0.9.15.1",
    "eccodes==2.38.1",

uv pip install -e .

$ ipython
Python 3.13.7 (main, Sep 18 2025, 19:47:49) [Clang 20.1.4 ]
Type 'copyright', 'credits' or 'license' for more information
IPython 9.7.0 -- An enhanced Interactive Python. Type '?' for help.
Tip: The `%timeit` magic has a `-o` flag, which returns the results, making it easy to plot. See `%timeit?`.

   ...: print(f"{cfgrib.__version__=}")
   ...: print()
   ...: print()
   ...: print()
   ...: print(f"{eccodes.__version__=}")
   ...: print()
   ...: print()
   ...: print()
   ...: xr.show_versions()
   ...: print()
   ...: print()
   ...: print()
   ...: xr.open_dataset(
   ...:     'era5-levels-members.grib',
   ...:     engine="cfgrib",
   ...:     backend_kwargs={"indexpath": ""},
   ...: )
/home/eschalk/work/trishna/trishna_tip-l2b/.venv/lib/python3.13/site-packages/gribapi/__init__.py:23: UserWarning: ecCodes 2.38.0 or higher is recommended. You are running version 2.31.0
  warnings.warn(



cfgrib.__version__='0.9.15.1'



eccodes.__version__='2.31.0'




INSTALLED VERSIONS
------------------
commit: None
python: 3.13.7 (main, Sep 18 2025, 19:47:49) [Clang 20.1.4 ]
python-bits: 64
OS: Linux
OS-release: 5.15.0-139-generic
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: ('en_US', 'UTF-8')
libhdf5: 1.14.6
libnetcdf: 4.9.3

xarray: 2025.9.1
pandas: 2.3.3
numpy: 2.3.4
scipy: 1.16.3
netCDF4: 1.7.3
pydap: None
h5netcdf: 1.7.3
h5py: 3.15.1
zarr: None
cftime: 1.6.5
nc_time_axis: None
iris: None
bottleneck: None
dask: 2024.12.1
distributed: None
matplotlib: 3.10.7
cartopy: 0.24.1
seaborn: None
numbagg: None
fsspec: 2025.10.0
cupy: None
pint: None
sparse: None
flox: None
numpy_groupies: None
setuptools: 80.9.0
pip: None
conda: None
pytest: 8.4.2
mypy: 1.18.2
IPython: 9.7.0
sphinx: 8.2.3



Out[1]: 
<xarray.Dataset> Size: 5MB
Dimensions:        (number: 10, time: 4, isobaricInhPa: 2, latitude: 61,
                    longitude: 120)
Coordinates:
  * number         (number) int64 80B 0 1 2 3 4 5 6 7 8 9
  * time           (time) datetime64[ns] 32B 2017-01-01 ... 2017-01-02T12:00:00
    step           timedelta64[ns] 8B ...
  * isobaricInhPa  (isobaricInhPa) float64 16B 850.0 500.0
  * latitude       (latitude) float64 488B 90.0 87.0 84.0 ... -84.0 -87.0 -90.0
  * longitude      (longitude) float64 960B 0.0 3.0 6.0 ... 351.0 354.0 357.0
    valid_time     (time) datetime64[ns] 32B ...
Data variables:
    z              (number, time, isobaricInhPa, latitude, longitude) float32 2MB ...
    t              (number, time, isobaricInhPa, latitude, longitude) float32 2MB ...
Attributes:
    GRIB_edition:            1
    GRIB_centre:             ecmf
    GRIB_centreDescription:  European Centre for Medium-Range Weather Forecasts
    GRIB_subCentre:          0
    Conventions:             CF-1.7
    institution:             European Centre for Medium-Range Weather Forecasts
    history:                 2025-11-15T11:46 GRIB to CDM+CF via cfgrib-0.9.1...

[iainrussell]

Many thanks for these reports - we will look into it.

[iainrussell]

Hmmmm... I cannot yet reproduce this issue. I created a venv on two different machines (one is our HPC, the other an Ubuntu 22.04 VM) and pip installed xarray and cfgrib, giving me the same versions as both of you (I tried all the versions of xarray listed in this issue), but it works for me without segfault.

I wonder if you could set this environment variable before running the Python script, just to see where the ecCodes binary is being picked up from (it should, by default come from eccodeslib inside your venv):

export ECCODES_PYTHON_TRACE_LIB_SEARCH=1

[iainrussell]

@roughnemezis, it would be interesting to see your complete environment, specifically eckitlib. Here's mine, which works on my Ubuntu VM:

Package         Version
--------------- -----------
attrs           25.4.0
cffi            2.0.0
cfgrib          0.9.15.1
click           8.3.1
eccodes         2.44.0
eccodeslib      2.44.0.5
eckitlib        1.32.2.5
fckitlib        0.14.0.5
findlibs        0.1.2
numpy           2.3.5
packaging       25.0
pandas          2.3.3
pip             25.0.1
pycparser       2.23
python-dateutil 2.9.0.post0
pytz            2025.2
six             1.17.0
tzdata          2025.2
xarray          2025.11.0

[roughnemezis]

Sorry for late reply.
I have:

merzisenh-> pip freeze | grep -e cfgrib -e eccodes -e eccodeslib -e eckitlib -e fckitlib -e numpy -e xarray
cfgrib==0.9.15.1
eccodes==2.44.0
eccodeslib==2.44.0.7
eckitlib==1.32.3.7
fckitlib==0.14.1.7
numpy==2.2.6
xarray==2025.6.1

My version of python is 3.10.12