aIbrahiim commented on PR #38287:
URL: https://github.com/apache/beam/pull/38287#issuecomment-4359735323

   > It's still failing
   > 
   > ```
   > testBundleFinalizationOccursOnUnboundedSplittableDoFn 
(org.apache.beam.sdk.transforms.SplittableDoFnTest) failed
   > 
   > 
runners/google-cloud-dataflow-java/build/test-results/validatesRunnerLegacyWorkerTestStreaming/TEST-org.apache.beam.sdk.transforms.SplittableDoFnTest.xml
 [took 15m 19s]
   > ```
   
   
   
   > It's still failing
   > 
   > ```
   > testBundleFinalizationOccursOnUnboundedSplittableDoFn 
(org.apache.beam.sdk.transforms.SplittableDoFnTest) failed
   > 
   > 
runners/google-cloud-dataflow-java/build/test-results/validatesRunnerLegacyWorkerTestStreaming/TEST-org.apache.beam.sdk.transforms.SplittableDoFnTest.xml
 [took 15m 19s]
   > ```
   
   I investigated further and the ConcurrentHashMap change helps the 
callback/thread safety path but the run is still failing through a different 
path
   
   based on what i observed in logs on gcp console
   
   Can't verify serialized elements of type BoundedSource have well defined 
equals method ... on step "PAssert$81"
   Dataflow job ... terminated in state RUNNING but did not return a failure 
reason.
   
   so my current thought is
   
   issue #1 is test/assertion behavior (PAssert + BoundedSource equality)
   issue #2 is runner-side streaming failure reporting during async 
timeout/cancel
   
   so i guess the fix is  to uypdate the failing assertion path to avoid direct 
equality on BoundedSource (or assert on deterministic derived values)
   and update TestDataflowRunner to wait briefly for terminal state after 
cancel and report clearer non terminal/cancelled state


-- 
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]

Reply via email to