tvalentyn commented on code in PR #37350:
URL: https://github.com/apache/beam/pull/37350#discussion_r2708653075
##########
sdks/python/apache_beam/io/gcp/bigquery_test.py:
##########
@@ -723,7 +723,7 @@ def test_read_export_exception(self, exception_type,
error_message):
'Insert') as mock_query_job, \
mock.patch('time.sleep'), \
self.assertRaises(Exception) as exc,\
- beam.Pipeline() as p:
+ beam.Pipeline(runner='DirectRunner') as p:
Review Comment:
Generally, a DirectRunner is a runner that executes the pipeline locally. We
have multiple implementations of DirectRunner, see
https://github.com/apache/beam/blob/600bd612dc2ef495fef08675398361159f7a03df/sdks/python/apache_beam/runners/direct/direct_runner.py.
The DirectRunner should already be a default (when a runner is not
explicitly specified). However, which implementation is used, should be defined
by the logic in direct_runner.py.
Originally, it should be set to 'SwitchingDirectRunner' , meaning that an
underlying implementation will be defined based on which DirectRunner
implementation supports the pipeline.
https://github.com/apache/beam/blob/600bd612dc2ef495fef08675398361159f7a03df/sdks/python/apache_beam/runners/direct/direct_runner.py#L626
The BundleBasedDirect Runner is the oldest implementation and is expected to
be used the least. I think these days we should be using either the FnAPI
runner or the PrismRunner.
grpc errors is a problem that we have in the FnAPI runner, and possibly also
in the PrismRunner - I am not sure. But ultimately, both FnAPI runner and
PrismRunner use grpc and we need to find a way how to use these runners in our
test suites without grpc flakes.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
