[
https://issues.apache.org/jira/browse/FLINK-10966?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16695609#comment-16695609
]
vinoyang commented on FLINK-10966:
----------------------------------
Hi [~zjwang] ,
answer your question:
1). For 1.5+, maybe you are right, but when we found this problem it was at
Flink 1.3.2. My intention is that we should not just store data. In the case of
a large amount of data, both memory and disk are limited by capacity. My
suggestion is to release the blocking as soon as possible so that the data can
participate in the calculation.
2). In the scenario I said, if the upstream blocking an input channel and the
other input channels are marked as blocked, then id2 has no chance to determine
the CheckpointBarrier event, so the blocking will not be released (It will go
into the if branch of the code I provided).
3). The trigger mechanism of the current checkpoint is periodic. I think we
should not speculate on the cause of the checkpoint timeout (may be caused by
various reasons, such as back pressure or the logic of the source itself, yes,
this is what we encountered) , the trigger logic that should not change the
checkpoint should not change. I just suggested that after a checkpoint timeout,
the CheckpointCoordinator will trigger the task to cancel the last round of
blocking.
> Optimize the release blocking logic in BarrierBuffer
> ----------------------------------------------------
>
> Key: FLINK-10966
> URL: https://issues.apache.org/jira/browse/FLINK-10966
> Project: Flink
> Issue Type: Improvement
> Components: State Backends, Checkpointing
> Reporter: vinoyang
> Assignee: vinoyang
> Priority: Major
>
> Issue:
> Currently, mixing CancelCheckpointMarker control events with data flow to
> drive task to release blocking logic in BarrierBuffer may result in blocking
> logic not being released in time, further leading to a large amount of data
> being spilled to disk.
> The source code for this problem is as follows:
> {code:java}
> BufferOrEvent bufferOrEvent = next.get();
> if (isBlocked(bufferOrEvent.getChannelIndex())) { //issue line
> // if the channel is blocked we, we just store the BufferOrEvent
> bufferBlocker.add(bufferOrEvent);
> checkSizeLimit();
> }
> else if (bufferOrEvent.isBuffer()) {
> return bufferOrEvent;
> }
> else if (bufferOrEvent.getEvent().getClass() == CheckpointBarrier.class) {
> if (!endOfStream) {
> // process barriers only if there is a chance of the checkpoint
> completing
> processBarrier((CheckpointBarrier) bufferOrEvent.getEvent(),
> bufferOrEvent.getChannelIndex());
> }
> }
> else if (bufferOrEvent.getEvent().getClass() == CancelCheckpointMarker.class)
> {
> processCancellationBarrier((CancelCheckpointMarker)
> bufferOrEvent.getEvent());
> }
> else {
> if (bufferOrEvent.getEvent().getClass() == EndOfPartitionEvent.class) {
> processEndOfPartition();
> }
> return bufferOrEvent;
> }
> {code}
> Scenarios:
> Considering a simple DAG:source->map (network shuffle), the degree of
> parallelism is 10. The checkpoint semantics is exactly once.
> The first checkpoint: barriers of 9 source subtask are received by all map
> subtask. One of the source subtasks is blocked, resulting in the failure to
> send barrier. Eventually, the checkpoint will fail due to timeout. At this
> point, 9 corresponding input channel are blocked because they have received
> barrier.
> Second checkpoint: At this point, the special source subtask is still blocked
> and cannot send any events to downstream, while the nine input channels are
> still blocked. From the current implementation, the data or events it
> receives will not be processed, but will be stored directly. Therefore, the
> barrier of the downstream task will not be released. The only hope is that
> the cached data reaches the maximum limit.
> I think the main problem here is that we should not store data which comes
> from blocked input channels directly. Especially when one input channel is
> blocked by upstream and nine input channels are marked as blocked, we may not
> always be able to release the blocking.
> A better mechanism might be that we send notifyCheckpointFailed callback via
> CheckpointCoordinator, allowing each task to unblock itself. This mechanism
> can make the release of the old checkpoint align independent of the trigger
> of the new checkpoint. If the interval of the checkpoints are very long but
> the timeout is very short, then the effect of the optimization will be more
> obvious.
> Ultimately, we want to reduce unnecessary blocking and data spill to disk.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
