flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Stephan Ewen (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (FLINK-3094) FileStateBackend does not clean up job-id checkpoint directory
Date Thu, 03 Dec 2015 13:00:14 GMT

    [ https://issues.apache.org/jira/browse/FLINK-3094?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15037750#comment-15037750
] 

Stephan Ewen commented on FLINK-3094:
-------------------------------------

It is a bit tricky in the current design. The design works with no central coordination -
the Checkpoint Coordinator is unaware of the state backend. This is good in the sense that
it makes it easy to mix state backends in the runtime tasks.

With this comes, however, a lack of a hook that cleans up after job shutdown (proper shutdown,
where checkpoint data should be removed).

> FileStateBackend does not clean up job-id checkpoint directory
> --------------------------------------------------------------
>
>                 Key: FLINK-3094
>                 URL: https://issues.apache.org/jira/browse/FLINK-3094
>             Project: Flink
>          Issue Type: Bug
>    Affects Versions: 1.0.0
>            Reporter: Till Rohrmann
>            Priority: Minor
>
> The {{FsStateBackend}} does not properly clean up the created jobID directories for the
checkpoints after the has been successfully executed. The problem is that the {{AbstractFileState}}
only deletes their parent directory (chk-xxx) but not the job-id directory upon discarding
the state.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message