[jira] [Commented] (CASSANDRA-16072) Reduce thread contention in CommitLogSegment and HintsBuffer by rewriting CAS loops to atomic adds

["Benjamin Lerer (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CASSANDRA-16072?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17188321#comment-17188321 ] 

Benjamin Lerer commented on CASSANDRA-16072:
--------------------------------------------

If we do some rough calculation on the limits of things for {{HintBuffer}}:
If the slab size is around {{1GB}}, the maximum hint size will be around {{500MB}}.
Once we exceed the slab capacity the {{position}} will be set to {{-2GB}}. It will only take 5 hints at a size close from the maximum hint size to be back on the positive side.
It seems to me that we are on a risky side. What do you think?

Regarding {{CommitLogSegment}}, it will be good to have a comment explaining the negative value logic. People might not get it straight away as it is an obvious way to handle overflow.
 
  

> Reduce thread contention in CommitLogSegment and HintsBuffer by rewriting CAS loops to atomic adds
> --------------------------------------------------------------------------------------------------
>
>                 Key: CASSANDRA-16072
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-16072
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Consistency/Hints, Local/Commit Log
>            Reporter: Michael Semb Wever
>            Assignee: Michael Semb Wever
>            Priority: Normal
>             Fix For: 3.11.x, 4.0-beta
>
>
> Follow up to CASSANDRA-15922
> Both CommitLogSegment and HintsBuffer use AtomicIntegers for the current offset when allocating. Like in CASSANDRA\-15922 the loops on {{.compareAndSet(..)}} can be replaced with atomic adds using the {{. getAndAdd(..)}} method.
> In highly contended environments the CAS failures can be high, starving writes in a running Cassandra node. On the same cluster CASSANDRA\-15922 was found, after CASSANDRA\-15922's fix was deployed, there was still problems around commit log flushing and hints. No flamegraph was collected that demonstrated the thread contention as clearly as was found in CASSANDRA\-15922, but the performance fix proposed here hopefully is obvious enough.



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

---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@cassandra.apache.org
For additional commands, e-mail: commits-help@cassandra.apache.org