[
https://issues.apache.org/jira/browse/KUDU-3323?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
vmaster.cc updated KUDU-3323:
-----------------------------
Attachment: Kudu.html
> Suspected memory leak?
> ----------------------
>
> Key: KUDU-3323
> URL: https://issues.apache.org/jira/browse/KUDU-3323
> Project: Kudu
> Issue Type: Bug
> Components: tablet, tserver
> Affects Versions: 1.8.0
> Environment: 1.8.0+cdh6.1.0
> Reporter: vmaster.cc
> Priority: Blocker
> Attachments: Kudu.html, image-2021-09-16-11-36-52-911.png,
> image-2021-09-16-11-40-20-431.png, image-2021-09-16-11-44-23-483.png
>
>
> Problem 1: Total consumption is much bigger than the other metric of Current
> Consumption, Where is the memory go ?
> !image-2021-09-16-11-36-52-911.png|width=590,height=225!
>
> Problem 2: This two tablets are the top memory consumption, is it normal of
> DeltaMemStores? This two tablets are own the same table,one is Leader,another
> one is follower, and this will be many update scenarios on this table.
> !image-2021-09-16-11-44-23-483.png|width=585,height=121!
>
> Note：There are many similar phenomena on our side: the memory cannot be
> reclaimed, and the only way is to restart the TabletServer. How should we do
> when this happen, or is the some way to reclaim memory manually ?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
