pabloem commented on a change in pull request #14113:
URL: https://github.com/apache/beam/pull/14113#discussion_r592802990
##########
File path: sdks/python/apache_beam/io/gcp/bigquery_file_loads.py
##########
@@ -857,6 +972,8 @@ def _load_data(
of the load jobs would fail but not other. If any of them fails, then
copy jobs are not triggered.
"""
+ singleton_pc = p | "ImpulseLoadData" >> beam.Create([None])
Review comment:
It'll be best to have separate singletons for each of these paths. The
issue was because by having one single PCollection reused in all these paths
was concentrating all of these various paths into a single stage with ~10 side
inputs, and this complicates firing of triggers for the stage, because we need
to wait for ~10 stages to advance their watermarks to start the 'singleton'
stage.
If we have multiple singleton pcollections, then we have multiple stages
with only one side input each.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
