open_mfdataset segfaults when using engine="netcdf4" and Prallel=Tru

What happened?

When opening a mf_dataset using netcdf4 engine and parallel=True a segmentation fault arises. The whole open_mfdataset and all operations are wrapped in a class inheriting from xr.Dataset. I would like to know whether netcdf4 engine is compatible with parallel=True, or if there are other ways of doing this. Don't mind the preprocess steps, it changes a dimension name.

PS : Using h5netcdf engine with parallel=True increases the reading time compared to using the serial netcdf4 engine

What did you expect to happen?

No segmentation fault.

Minimal Complete Verifiable Example

import os

os.environ["HDF5_USE_FILE_LOCKING"] = "FALSE"

import re
from typing import List, Optional, Any

import numpy as np
import xarray as xr
import dask

mf_dataset = xr.open_mfdataset(
  self.data_paths, engine="netcdf4", preprocess=preprocess, combine="by_coords", mask_and_scale=False, decode_cf=False, parallel=False,
)

Steps to reproduce

No response

MVCE confirmation

• Minimal example — the example is as focused as reasonably possible to demonstrate the underlying issue in xarray.
• Complete example — the example is self-contained, including all data and the text of any traceback.
• Verifiable example — the example copy & pastes into an IPython prompt or Binder notebook, returning the result.
• New issue — a search of GitHub Issues suggests this is not a duplicate.
• Recent environment — the issue occurs with the latest version of xarray and its dependencies.

Relevant log output

Anything else we need to know?

No response

Environment

Start with xr.open_mfdataset using engine="netcdf4" and parallel=True, then verify the behavior against the reported Python, xarray, netCDF4, HDF5, and Dask environment. The report does not include a complete verifiable example, traceback, or test location, so first establish a reproducible case and compare serial and parallel opening. Done means the reproducible case no longer segfaults or the incompatibility is clearly documented.