[jira] [Comment Edited] (FLINK-18748) Savepoint would be queued unexpected if pendingCheckpoints less than maxConcurrentCheckpoints

["tao wang (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-18748?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17169597#comment-17169597 ] 

tao wang edited comment on FLINK-18748 at 8/3/20, 3:10 AM:
-----------------------------------------------------------

Hi, [~klion26] [~roman_khachatryan]: I have create a pull_request, would you like to review this: [https://github.com/apache/flink/pull/13045]?

thanks.


was (Author: wayland):
Hi, [~klion26] [~roman_khachatryan]: I have create a pull_request, would you like to review?

thanks.

> Savepoint would be queued unexpected if pendingCheckpoints less than maxConcurrentCheckpoints
> ---------------------------------------------------------------------------------------------
>
>                 Key: FLINK-18748
>                 URL: https://issues.apache.org/jira/browse/FLINK-18748
>             Project: Flink
>          Issue Type: Bug
>          Components: Runtime / Checkpointing
>    Affects Versions: 1.11.0, 1.11.1
>            Reporter: Congxian Qiu(klion26)
>            Assignee: tao wang
>            Priority: Major
>              Labels: pull-request-available
>
> Inspired by a [user-zh email|http://apache-flink.147419.n8.nabble.com/flink-1-11-rest-api-saveppoint-td5497.html]
> After FLINK-17342, when triggering a checkpoint/savepoint, we'll check whether the request can be triggered in {{CheckpointRequestDecider#chooseRequestToExecute}}, the logic is as follow:
> {code:java}
> Preconditions.checkState(Thread.holdsLock(lock));
> // 1. 
> if (isTriggering || queuedRequests.isEmpty()) {
>    return Optional.empty();
> }
> // 2 too many ongoing checkpoitn/savepoint
> if (pendingCheckpointsSizeSupplier.get() >= maxConcurrentCheckpointAttempts) {
>    return Optional.of(queuedRequests.first())
>       .filter(CheckpointTriggerRequest::isForce)
>       .map(unused -> queuedRequests.pollFirst());
> }
> // 3 check the timestamp of last complete checkpoint
> long nextTriggerDelayMillis = nextTriggerDelayMillis(lastCompletionMs);
> if (nextTriggerDelayMillis > 0) {
>    return onTooEarly(nextTriggerDelayMillis);
> }
> return Optional.of(queuedRequests.pollFirst());
> {code}
> But if currently {{pendingCheckpointsSizeSupplier.get()}} < {{maxConcurrentCheckpointAttempts}}, and the request is a savepoint, the savepoint will still wait some time in step 3. 
> I think we should trigger the savepoint immediately if {{pendingCheckpointSizeSupplier.get()}} < {{maxConcurrentCheckpointAttempts}}.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)