Catch issue with partial mpi4py installation. (#21353)

* Catch issue with partial mpi4py installation.

Depending on precise install, mpi4py may be installed without `MPI`  being present. Catch that case in the detect() function,
and treat it as if mpi4py is not installed.

* update

* update

* update

---------

Co-authored-by: deependujha <[email protected]>
Co-authored-by: Justus Schock <[email protected]>

Author: 3 people

src/lightning/fabric/CHANGELOG.md

@@ -25,13 +25,11 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 ### Fixed
 
-- Fixed `EADDRINUSE` errors in distributed tests with port manager and retry logic ([#21309](https://github.com/Lightning-AI/pytorch-lightning/pull/21309))
-
+- Fixed issue in detecting MPIEnvironment with partial mpi4py installation ([#21353](https://github.com/Lightning-AI/pytorch-lightning/pull/21353))
 
 - Learning rate scheduler is stepped at the end of epoch when `on_train_batch_start` returns -1 ([#21296](https://github.com/Lightning-AI/pytorch-lightning/issues/21296)).
 
 
-
 ---
 
 ## [2.5.5] - 2025-09-05

src/lightning/fabric/plugins/environments/mpi.py

@@ -73,7 +73,11 @@ def detect() -> bool:
         if not _MPI4PY_AVAILABLE:
             return False
 
-        from mpi4py import MPI
+        try:
+            # mpi4py may be installed without MPI being present
+            from mpi4py import MPI
+        except ImportError:
+            return False
 
         return MPI.COMM_WORLD.Get_size() > 1