[
https://issues.apache.org/jira/browse/FLINK-22326?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17364750#comment-17364750
]
Dawid Wysakowicz edited comment on FLINK-22326 at 6/17/21, 7:39 AM:
--------------------------------------------------------------------
I think updating the documentation is a good idea. If you like to help with
that I can review the change.
{{tryYield}} is not a good solution as well, as it basically leads to busy
looping. We'd have to change the queue to one that gives us a Future, which
tells if a record is available. That way we could suspend the default action of
the mailbox processor.
{{Yield}} in the {{AsyncWaitOperator}} is correct imo, as the only way to put
new records into the queue is from processing records via the
{{mailboxProcessor}}. In other words there must be a new mail in the mailbox to
unblock the queue.
was (Author: dawidwys):
I think updating the documentation is a good idea. If you like to help with
that I can review the change.
{{tryYield}} is not a good solution as well, as it basically leads to busy
looping. We'd have to change the queue to one that gives us a Future, which
tells if a record is available. That way we could suspend the default action of
the mailbox processor.
> Job contains Iterate Operator always fails on Checkpoint
> ---------------------------------------------------------
>
> Key: FLINK-22326
> URL: https://issues.apache.org/jira/browse/FLINK-22326
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Checkpointing
> Affects Versions: 1.11.1
> Reporter: Lu Niu
> Assignee: Lu Niu
> Priority: Major
> Labels: pull-request-available
> Attachments: Screen Shot 2021-04-16 at 12.40.34 PM.png, Screen Shot
> 2021-04-16 at 12.43.38 PM.png
>
>
> Job contains Iterate Operator will always fail on checkpoint.
> How to reproduce:
> [https://gist.github.com/qqibrow/f297babadb0bb662ee398b9088870785]
> this is based on
> [https://github.com/apache/flink/blob/master/flink-examples/flink-examples-streaming/src/main/java/org/apache/flink/streaming/examples/iteration/IterateExample.java,]
> but a few line difference:
> 1. Make maxWaitTime large enough when create IterativeStream
> 2. No output back to Itertive Source
> Result:
> The same code is able to checkpoint in 1.9.1
> !Screen Shot 2021-04-16 at 12.43.38 PM.png!
>
> but always fail on checkpoint in 1.11
> !Screen Shot 2021-04-16 at 12.40.34 PM.png!
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
