Remove --impersonate_service_account whenever PipelineOptions are s…

…erialized (#32031)

* Remove the impersonate_service_account pipeline option during serialization.

* Update Changes.md

CHANGES.md

@@ -79,7 +79,7 @@
 
 ## Bugfixes
 
-* Fixed X (Java/Python) ([#X](https://github.com/apache/beam/issues/X)).
+* Fixed incorrect service account impersonation flow for Python pipelines using BigQuery IOs ([#32030](https://github.com/apache/beam/issues/32030)).
 
 ## Security Fixes
 * Fixed (CVE-YYYY-NNNN)[https://www.cve.org/CVERecord?id=CVE-YYYY-NNNN] (Java/Python/Go) ([#X](https://github.com/apache/beam/issues/X)).
@@ -526,6 +526,7 @@ as a workaround, a copy of "old" `CountingSource` class should be placed into a
 ## Known Issues
 
 * Long-running Python pipelines might experience a memory leak: [#28246](https://github.com/apache/beam/issues/28246).
+* Python pipelines using the `--impersonate_service_account` option with BigQuery IOs might fail on Dataflow ([#32030](https://github.com/apache/beam/issues/32030)). This is fixed in 2.59.0 release.
 
 
 # [2.48.0] - 2023-05-31

sdks/python/apache_beam/options/pipeline_options.py

@@ -247,6 +247,20 @@ def __init__(self, flags=None, **kwargs):
         self._all_options[option_name] = getattr(
             self._visible_options, option_name)
 
+  def __getstate__(self):
+    # The impersonate_service_account option must be used only at submission of
+    # a Beam job. However, Beam IOs might store pipeline options
+    # within transform implementation that becomes serialized in RunnerAPI,
+    # causing this option to be inadvertently used at runtime.
+    # This serialization hook removes it.
+    if self.view_as(GoogleCloudOptions).impersonate_service_account:
+      dict_copy = dict(self.__dict__)
+      dict_copy['_all_options'] = dict(dict_copy['_all_options'])
+      dict_copy['_all_options']['impersonate_service_account'] = None
+      return dict_copy
+    else:
+      return self.__dict__
+
   @classmethod
   def _add_argparse_args(cls, parser):
     # type: (_BeamArgumentParser) -> None

website/www/site/content/en/blog/beam-2.49.0.md

@@ -52,6 +52,7 @@ For more information on changes in 2.49.0, check out the [detailed release notes
 
 * Long-running Python pipelines might experience a memory leak: [#28246](https://github.com/apache/beam/issues/28246).
 * Python SDK's cross-language Bigtable sink mishandles records that don't have an explicit timestamp set: [#28632](https://github.com/apache/beam/issues/28632). To avoid this issue, set explicit timestamps for all records before writing to Bigtable.
+* Python pipelines using the `--impersonate_service_account` option with BigQuery IOs might fail on Dataflow ([#32030](https://github.com/apache/beam/issues/32030)). This is fixed in 2.59.0 release.
 
 
 ## List of Contributors