[
https://issues.apache.org/jira/browse/CASSANDRA-21799?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18130304#comment-18130304
]
Dmitry Konstantinov commented on CASSANDRA-21799:
-------------------------------------------------
This kind of issue is also known as GC nepotism:
https://psy-lob-saw.blogspot.com/2016/03/gc-nepotism-and-linked-queues.html
> OpOrder groups do not clear their next pointer
> ----------------------------------------------
>
> Key: CASSANDRA-21799
> URL: https://issues.apache.org/jira/browse/CASSANDRA-21799
> Project: Apache Cassandra
> Issue Type: Improvement
> Components: Local/Other
> Reporter: Branimir Lambov
> Assignee: Branimir Lambov
> Priority: Normal
>
> While this is not required for correctness, when the forward links are left
> in place and a group/barrier manages to get promoted to the heap's old
> generation, all barriers after the promoted one also become eligible for
> promotion and cannot be cleared until an old-gen GC cycle.
> We have seen this cause significant GC pauses for off-heap trie memtables.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
