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

Posted to issues@flink.apache.org by "Chesnay Schepler (Jira)" <ji...@apache.org> on 2021/10/15 07:26:00 UTC

[jira] [Updated] (FLINK-24484) Support manual cleanup of async operation results

     [ https://issues.apache.org/jira/browse/FLINK-24484?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Chesnay Schepler updated FLINK-24484:
-------------------------------------
    Description: 
The REST API state for asynchronous operations results is currently kept for 5 minutes.

This behavior is fine for human interactions.
For automated interactions it is not ideal because if the external system fails to persist the result then it may not be able to retrieve the result again.

To solve this issue I propose:
a) make the caching timeout configurable (currently hard-coded to 5 minutes)
b) add a DELETE endpoint for operation results.

  was:
The REST API state for asynchronous operations results is currently kept for 5 minutes or until it was consumed by the user.

This behavior is fine for human interactions.
For automated interactions it is not ideal because if the external system fails to persist the result then it cannot retrieve the result again. In other words it only has exactly one attempt at consuming the result.

To solve this issue I propose:
a) make the caching timeout configurable (currently hard-coded to 5 minutes)
b) add a DELETE endpoint for operation results.


> Support manual cleanup of async operation results
> -------------------------------------------------
>
>                 Key: FLINK-24484
>                 URL: https://issues.apache.org/jira/browse/FLINK-24484
>             Project: Flink
>          Issue Type: Improvement
>          Components: Runtime / REST
>            Reporter: Chesnay Schepler
>            Assignee: Chesnay Schepler
>            Priority: Major
>             Fix For: 1.15.0
>
>
> The REST API state for asynchronous operations results is currently kept for 5 minutes.
> This behavior is fine for human interactions.
> For automated interactions it is not ideal because if the external system fails to persist the result then it may not be able to retrieve the result again.
> To solve this issue I propose:
> a) make the caching timeout configurable (currently hard-coded to 5 minutes)
> b) add a DELETE endpoint for operation results.



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