zhijiangW commented on a change in pull request #11351: [FLINK-16404][runtime]
Solve the potential deadlock problem when reducing exclusive buffers to zero
URL: https://github.com/apache/flink/pull/11351#discussion_r397600950
##########
File path:
flink-streaming-java/src/main/java/org/apache/flink/streaming/runtime/io/CheckpointBarrierAligner.java
##########
@@ -172,10 +178,9 @@ else if (barrierId > currentCheckpointId) {
}
releaseBlocksAndResetBarriers();
- notifyCheckpoint(receivedBarrier, bufferedBytes,
latestAlignmentDurationNanos);
- return true;
+ notifyCheckpoint(receivedBarrier,
latestAlignmentDurationNanos);
+ notifyCheckpointCompletedOrCanceled(barrierId);
Review comment:
I am considering the call sequence for `notifyCheckpoint` and
`notifyCheckpointCompletedOrCanceled`.
`notifyCheckpoint` might take some time for local IO operations I guess. If
we can `notifyCheckpointCompletedOrCanceled` beforehand to unblock upstream to
send data, then it might get benefit for performance.
----------------------------------------------------------------
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]
With regards,
Apache Git Services
