You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Yun Tang (Jira)" <ji...@apache.org> on 2020/08/27 09:17:00 UTC

[jira] [Commented] (FLINK-19060) Checkpoint not triggered when use broadcast stream

    [ https://issues.apache.org/jira/browse/FLINK-19060?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17185734#comment-17185734 ] 

Yun Tang commented on FLINK-19060:
----------------------------------

Did you check the logs of jobmanager to see why no checkpoints?

I guess your broadcast stream operators just switched to FINISHED status as you create the stream via {{env.fromElements}} while other tasks still in RUNNING not FINISHED status. However, current checkpoint mechanism needs all tasks in RUNNING status, if not, no checkpoints would be triggered.

 

If my guess is right, this is not a bug but a limit for Flink for a long time.

> Checkpoint not triggered when use broadcast stream
> --------------------------------------------------
>
>                 Key: FLINK-19060
>                 URL: https://issues.apache.org/jira/browse/FLINK-19060
>             Project: Flink
>          Issue Type: Bug
>          Components: API / DataStream
>    Affects Versions: 1.11.1
>            Reporter: henvealf
>            Priority: Major
>         Attachments: image-2020-08-27-16-41-23-699.png, image-2020-08-27-16-44-37-442.png, image-2020-08-27-16-45-28-134.png, image-2020-08-27-16-51-10-512.png
>
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> Code:
> !image-2020-08-27-16-51-10-512.png!
> KafkaSourceConfig:
> consumer.setStartFromGroupOffsets()
> Web UI:
>     !image-2020-08-27-16-45-28-134.png!
> Checkpoint always doesn't happen. Did I write something wrong?
> Thanks!



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