PYTHON-4703 MongoClient should default to connect=False on FaaS environments (#1844)

doc/changelog.rst

@@ -42,6 +42,12 @@ PyMongo 4.9 brings a number of improvements including:
 - Fixed a bug where PyMongo would raise ``InvalidBSON: date value out of range``
   when using :attr:`~bson.codec_options.DatetimeConversion.DATETIME_CLAMP` or
   :attr:`~bson.codec_options.DatetimeConversion.DATETIME_AUTO` with a non-UTC timezone.
+- The default value for ``connect`` in ``MongoClient`` is changed to ``False`` when running on
+  unction-as-a-service (FaaS) like AWS Lambda, Google Cloud Functions, and Microsoft Azure Functions.
+  On some FaaS systems, there is a ``fork()`` operation at function
+  startup.  By delaying the connection to the first operation, we avoid a deadlock.  See
+  `Is PyMongo Fork-Safe`_ for more information.
+
 
 Issues Resolved
 ...............
@@ -49,6 +55,7 @@ Issues Resolved
 See the `PyMongo 4.9 release notes in JIRA`_ for the list of resolved issues
 in this release.
 
+.. _Is PyMongo Fork-Safe : https://www.mongodb.com/docs/languages/python/pymongo-driver/current/faq/#is-pymongo-fork-safe-
 .. _PyMongo 4.9 release notes in JIRA: https://jira.mongodb.org/secure/ReleaseNote.jspa?projectId=10004&version=39940
 
 

pymongo/asynchronous/mongo_client.py

@@ -720,6 +720,10 @@ class AsyncMongoClient(common.BaseObject, Generic[_DocumentType]):
 
         .. versionchanged:: 4.7
             Deprecated parameter ``wTimeoutMS``, use :meth:`~pymongo.timeout`.
+
+        .. versionchanged:: 4.9
+            The default value of ``connect`` is changed to ``False`` when running in a
+            Function-as-a-service environment.
         """
         doc_class = document_class or dict
         self._init_kwargs: dict[str, Any] = {
@@ -803,7 +807,10 @@ class AsyncMongoClient(common.BaseObject, Generic[_DocumentType]):
         if tz_aware is None:
             tz_aware = opts.get("tz_aware", False)
         if connect is None:
-            connect = opts.get("connect", True)
+            # Default to connect=True unless on a FaaS system, which might use fork.
+            from pymongo.pool_options import _is_faas
+
+            connect = opts.get("connect", not _is_faas())
         keyword_opts["tz_aware"] = tz_aware
         keyword_opts["connect"] = connect
 

pymongo/synchronous/mongo_client.py

@@ -263,7 +263,8 @@ class MongoClient(common.BaseObject, Generic[_DocumentType]):
             aware (otherwise they will be naive)
         :param connect: If ``True`` (the default), immediately
             begin connecting to MongoDB in the background. Otherwise connect
-            on the first operation.
+            on the first operation.  The default value is ``False`` when
+            running in a Function-as-a-service environment.
         :param type_registry: instance of
             :class:`~bson.codec_options.TypeRegistry` to enable encoding
             and decoding of custom types.
@@ -719,6 +720,10 @@ class MongoClient(common.BaseObject, Generic[_DocumentType]):
 
         .. versionchanged:: 4.7
             Deprecated parameter ``wTimeoutMS``, use :meth:`~pymongo.timeout`.
+
+        .. versionchanged:: 4.9
+            The default value of ``connect`` is changed to ``False`` when running in a
+            Function-as-a-service environment.
         """
         doc_class = document_class or dict
         self._init_kwargs: dict[str, Any] = {
@@ -802,7 +807,10 @@ class MongoClient(common.BaseObject, Generic[_DocumentType]):
         if tz_aware is None:
             tz_aware = opts.get("tz_aware", False)
         if connect is None:
-            connect = opts.get("connect", True)
+            # Default to connect=True unless on a FaaS system, which might use fork.
+            from pymongo.pool_options import _is_faas
+
+            connect = opts.get("connect", not _is_faas())
         keyword_opts["tz_aware"] = tz_aware
         keyword_opts["connect"] = connect
 

test/lambda/mongodb/app.py

@@ -8,6 +8,7 @@ from __future__ import annotations
 
 import json
 import os
+import warnings
 
 from bson import has_c as has_bson_c
 from pymongo import MongoClient
@@ -18,6 +19,9 @@ from pymongo.monitoring import (
     ServerHeartbeatListener,
 )
 
+# Ensure there are no warnings raised in normal operation.
+warnings.simplefilter("error")
+
 open_connections = 0
 heartbeat_count = 0
 streaming_heartbeat_count = 0

tools/synchro.py

@@ -108,7 +108,8 @@ replacements = {
 docstring_replacements: dict[tuple[str, str], str] = {
     ("MongoClient", "connect"): """If ``True`` (the default), immediately
             begin connecting to MongoDB in the background. Otherwise connect
-            on the first operation.""",
+            on the first operation.  The default value is ``False`` when
+            running in a Function-as-a-service environment.""",
     ("Collection", "create"): """If ``True``, force collection
             creation even without options being set.""",
     ("Collection", "session"): """A