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

Posted to issues@flink.apache.org by "Lijie Wang (Jira)" <ji...@apache.org> on 2023/07/23 08:21:00 UTC

[jira] [Commented] (FLINK-32386) Add ShuffleDescriptorsCache in TaskExecutor to cache ShuffleDescriptors

    [ https://issues.apache.org/jira/browse/FLINK-32386?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17746039#comment-17746039 ] 

Lijie Wang commented on FLINK-32386:
------------------------------------

Done via master(1.18): 4d642635c9409883724bd0127f2a00ef14993bf0

> Add ShuffleDescriptorsCache in TaskExecutor to cache ShuffleDescriptors
> -----------------------------------------------------------------------
>
>                 Key: FLINK-32386
>                 URL: https://issues.apache.org/jira/browse/FLINK-32386
>             Project: Flink
>          Issue Type: Sub-task
>          Components: Runtime / Coordination
>            Reporter: Weihua Hu
>            Assignee: Weihua Hu
>            Priority: Major
>              Labels: pull-request-available
>
> Introduce a new struct named ShuffleDescriptorsCache to cache ShuffleDescriptorAndIndex which are offloaded by the blob server. 
> The cache should have the following capabilities:
> 1. Expired after exceeding the TTL.
> 2. Limit the size of the cache. Remove the oldest element from the cache when its maximum size has been exceeded.
> 3. Clear elements belong to a job when it disconnects from TaskExecutor.



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