[
https://issues.apache.org/jira/browse/FLINK-18641?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17162667#comment-17162667
]
Piotr Nowojski commented on FLINK-18641:
----------------------------------------
So the problem is that checkpoint is finalised after JM receives all of the
ACKS, without waiting for async master hook to complete? And the solution would
be to include waiting for async hook to complete, before completing/finalising
the checkpoint?
{quote}
In Flink 1.10, the checkpoint thread blocks on the future returned by the
master hook. This becomes asynchronous in Flink 1.11.
{quote}
[~becket_qin], as [~SleePy] mentioned that's not correct. Those changes were
done already in Flink 1.10.
> "Failure to finalize checkpoint" error in MasterTriggerRestoreHook
> ------------------------------------------------------------------
>
> Key: FLINK-18641
> URL: https://issues.apache.org/jira/browse/FLINK-18641
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Checkpointing
> Affects Versions: 1.11.0
> Reporter: Brian Zhou
> Priority: Major
>
> https://github.com/pravega/flink-connectors is a Pravega connector for Flink.
> The ReaderCheckpointHook[1] class uses the Flink `MasterTriggerRestoreHook`
> interface to trigger the Pravega checkpoint during Flink checkpoints to make
> sure the data recovery. The checkpoint recovery tests are running fine in
> Flink 1.10, but it has below issues in Flink 1.11 causing the tests time out.
> Suspect it is related to the checkpoint coordinator thread model changes in
> Flink 1.11
> Error stacktrace:
> {code}
> 2020-07-09 15:39:39,999 30945 [jobmanager-future-thread-5] WARN
> o.a.f.runtime.jobmaster.JobMaster - Error while processing checkpoint
> acknowledgement message
> org.apache.flink.runtime.checkpoint.CheckpointException: Could not finalize
> the pending checkpoint 3. Failure reason: Failure to finalize checkpoint.
> at
> org.apache.flink.runtime.checkpoint.CheckpointCoordinator.completePendingCheckpoint(CheckpointCoordinator.java:1033)
> at
> org.apache.flink.runtime.checkpoint.CheckpointCoordinator.receiveAcknowledgeMessage(CheckpointCoordinator.java:948)
> at
> org.apache.flink.runtime.scheduler.SchedulerBase.lambda$acknowledgeCheckpoint$4(SchedulerBase.java:802)
> at
> java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
> at java.util.concurrent.FutureTask.run(FutureTask.java:266)
> at
> java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:180)
> at
> java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> at java.lang.Thread.run(Thread.java:748)
> Caused by: org.apache.flink.util.SerializedThrowable: Pending checkpoint has
> not been fully acknowledged yet
> at
> org.apache.flink.util.Preconditions.checkState(Preconditions.java:195)
> at
> org.apache.flink.runtime.checkpoint.PendingCheckpoint.finalizeCheckpoint(PendingCheckpoint.java:298)
> at
> org.apache.flink.runtime.checkpoint.CheckpointCoordinator.completePendingCheckpoint(CheckpointCoordinator.java:1021)
> ... 9 common frames omitted
> {code}
> More detail in this mailing thread:
> http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/Pravega-connector-cannot-recover-from-the-checkpoint-due-to-quot-Failure-to-finalize-checkpoint-quot-td36652.html
> Also in https://github.com/pravega/flink-connectors/issues/387
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
