[jira] [Commented] (NIFI-16153) Avoid FlowController read lock when reading clustered state

["ASF subversion and git services (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/NIFI-16153?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18100156#comment-18100156 ] 

ASF subversion and git services commented on NIFI-16153:
--------------------------------------------------------

Commit 89828cf75cbfb3fb0ea2e40f1e2afd7472f2e306 in nifi's branch refs/heads/main from Pierre Villard
[ https://gitbox.apache.org/repos/asf?p=nifi.git;h=89828cf75cb ]

NIFI-16153 Avoid FlowController read lock when reading clustered status (#11484)

Signed-off-by: David Handermann <ex...@apache.org>

> Avoid FlowController read lock when reading clustered state
> -----------------------------------------------------------
>
>                 Key: NIFI-16153
>                 URL: https://issues.apache.org/jira/browse/NIFI-16153
>             Project: Apache NiFi
>          Issue Type: Improvement
>            Reporter: Pierre Villard
>            Assignee: Pierre Villard
>            Priority: Major
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> FlowController.isClustered() acquires the read lock to return a single boolean, and every other cluster-state accessor reaches that lock through it, so a component that reads cluster state while being enabled during flow synchronization blocks until the synchronizing thread releases the write lock. Making the clustered field volatile and returning it directly removes the contention, with no loss of atomicity since the lock only ever guarded a single boolean read.



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