The GitHub Actions job "Flink Connector Kafka CI" on 
flink-connector-kafka.git/FLINK-33484 has failed.
Run started by GitHub user dalelane (triggered by dalelane).

Head commit for run:
e44287bedc002e0b88379beba44fcac7afbdf4ab / Dale Lane <[email protected]>
fix: wake the fetcher when enqueuing commit task

The new tests introduced in this branch were taking longer than
is reasonable, introducing a 13 sec delay for each checkpoint's
async offset commit to complete on idle splits.

The cause appears to be in
KafkaSourceFetcherManager.enqueueOffsetsCommitTask which was
enqueuing the commit task, but never waking the fetcher. The
outcome was a poll.timeout.ms delay introduced for every test
recreating an idle partition.

This commit adds a call to wake up the consumer, copying the
pattern used in other places that tasks are enqueued onto live
fetchers, such as the addSplits, removeSplits, and
pauseOrResumeSplits methods in SplitFetcher from
flink-connector-base

The commitOffsets path looks like it was an exception, going
through the public enqueueTask but skipping the wakeup.

Signed-off-by: Dale Lane <[email protected]>

Report URL: 
https://github.com/apache/flink-connector-kafka/actions/runs/35501213535

With regards,
GitHub Actions via GitBox

Reply via email to