You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@flink.apache.org by "aitozi (JIRA)" <ji...@apache.org> on 2018/12/05 10:56:00 UTC

[jira] [Updated] (FLINK-11077) Make subtask aware of the timeout of checkpoint and abort the current ongoing asynccheckpoint

     [ https://issues.apache.org/jira/browse/FLINK-11077?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

aitozi updated FLINK-11077:
---------------------------
    Description: Now checkpoint coordinator will cancel the checkpoint when checkpoint timeout. But the subtask will continue to execute the async part / sync part of the timeout checkpoint. When checkpoint state is large, it is a waste of bandwidth. So I think the AsyncCheckpointRunnable should be aware of the timeout checkpoint and interrupt them. What do you think [~StephanEwen]  [~srichter]  (was: Now checkpoint coordinator will cancel the checkpoint when checkpoint timeout. But the subtask will continue to execute the async part / sync part of the timeout checkpoint. When checkpoint state is large, it is a waste of bandwidth. So I think the AsyncCheckpointRunnable should be aware of the timeout checkpoint and interrupt them. What do you think [~StephanEwen] [~stefanrichter83@gmail.com])

> Make subtask aware of the timeout of checkpoint and abort the current ongoing asynccheckpoint
> ---------------------------------------------------------------------------------------------
>
>                 Key: FLINK-11077
>                 URL: https://issues.apache.org/jira/browse/FLINK-11077
>             Project: Flink
>          Issue Type: Improvement
>          Components: State Backends, Checkpointing
>    Affects Versions: 1.8.0
>            Reporter: aitozi
>            Assignee: aitozi
>            Priority: Major
>
> Now checkpoint coordinator will cancel the checkpoint when checkpoint timeout. But the subtask will continue to execute the async part / sync part of the timeout checkpoint. When checkpoint state is large, it is a waste of bandwidth. So I think the AsyncCheckpointRunnable should be aware of the timeout checkpoint and interrupt them. What do you think [~StephanEwen]  [~srichter]



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)