[ 
https://issues.apache.org/jira/browse/FLINK-38433?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Zakelly Lan resolved FLINK-38433.
---------------------------------
    Fix Version/s: 2.2.0
       Resolution: Fixed

> Avoid delete ForSt's directory when there happened to be an existing one
> ------------------------------------------------------------------------
>
>                 Key: FLINK-38433
>                 URL: https://issues.apache.org/jira/browse/FLINK-38433
>             Project: Flink
>          Issue Type: Bug
>    Affects Versions: 2.0.0, 2.1.0
>            Reporter: Zakelly Lan
>            Assignee: Han Yin
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 2.2.0
>
>
> Currently, ForSt State backend will create the working directory, if there 
> exists one with same name the directory will be cleared. The name contains 
> jobid, vertexid, task_index, parallelism and attempt. In JM failover 
> scenario, this would be a collision as the attempt number restart from 0, 
> resulting in checkpoint broken.
>  
> We should remove the clear existing directory logic.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to