ModuleNotFoundError: No module named 'distributed'

Not specified by the original author.

TL;DR: distributed is a separate package that dask does not install by default. Run pip install "dask[distributed]" and the Client import works.

The error

ModuleNotFoundError: No module named 'distributed'

Fix it

1. Install distributed alongside dask:

   pip install "dask[distributed]"

   Expected: installs the distributed package.

2. Verify:

   python -c "from dask.distributed import Client; print('ok')"

   Expected: prints ok.

3. Conda equivalent: conda install dask distributed.

When this applies

Use this when from dask.distributed import Client or import dask.distributed raises the missing-module error.

When it does not apply

If you only use dask's threaded or single-machine schedulers, you do not need distributed at all. If distributed is installed but the import still fails, check you are in the right environment.

Compatibility

All dask versions; distributed has always been a separate package.

Root cause

dask core ships the schedulers and collections; the distributed scheduler (Client, clusters) lives in the separate distributed package so single-machine users stay light. Importing dask.distributed without it installed raises this error by design.

Edge cases

dask and distributed versions must match closely; installing via the dask[distributed] extra keeps them in sync. A stale distributed install can raise a different import error; upgrade both together.

The Dask forum thread (dask.discourse.group/t/3857) reports ModuleNotFoundError: No module named 'distributed' when doing from dask.distributed import Client even with dask installed. The verified answer: distributed is a separate package that dask does not hard-require, so the fix is pip install distributed (or pip install dask[distributed] / conda install dask distributed), after which the Client import works. Thread: https://dask.discourse.group/t/installing-distributed/3857