je-ik commented on issue #28554:
URL: https://github.com/apache/beam/issues/28554#issuecomment-1742501922
Sorry for late reaction, I was OOO.
I once again walked though the code and it seems to make sense now. Here is
the problem:
a) `flushData` method is wired up to
https://nightlies.apache.org/flink/flink-docs-master/api/java/org/apache/flink/streaming/api/operators/StreamOperator.html#finish--
(https://nightlies.apache.org/flink/flink-docs-release-1.12/api/java/org/apache/flink/streaming/api/operators/AbstractStreamOperator.html#close--
to Flink 1.12)
b) this method is apparently called prior to call to
`notifyCheckpointComplete`, which means we still have buffered data
The only solution then seems to make sure that in call to `flushData` je
clean the buffer (process all buffered data and emit outputs, clearing the
watermark hold). All this should happen immediately *before* the last
checkpoint is completed and thus it should not break the stable input DoFn's
contract.
--
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]
