[jira] [Commented] (KAFKA-5393) in-memory state store memory issue

["Guozhang Wang (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/KAFKA-5393?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16499769#comment-16499769 ] 

Guozhang Wang commented on KAFKA-5393:
--------------------------------------

Is there an easy way to reproduce this issue for further investigate?

> in-memory state store memory issue 
> -----------------------------------
>
>                 Key: KAFKA-5393
>                 URL: https://issues.apache.org/jira/browse/KAFKA-5393
>             Project: Kafka
>          Issue Type: Bug
>          Components: streams
>    Affects Versions: 0.10.2.1
>            Reporter: Kevin Chen
>            Priority: Major
>         Attachments: Screen Shot 2017-06-06 at 9.45.42 AM.png
>
>
> We are running 2 kafka stream instance that use low level Processor API, and we are using in-memory state store.
> When we upgrade  instance A, which will move all the tasks to instance B, and re-balance between the 2 after A is back up and running. 
> But the problem is that, even after the re-balance, the memory in instance A did not drop to the previous level(the load is about the same).
> see the screen shot below.



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