You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@phoenix.apache.org by "Andrew Kyle Purtell (Jira)" <ji...@apache.org> on 2026/05/01 23:22:00 UTC
[jira] [Created] (PHOENIX-7825) ConcurrentMutationsIT make MyClock.shouldAdvance volatile
Andrew Kyle Purtell created PHOENIX-7825:
--------------------------------------------
Summary: ConcurrentMutationsIT make MyClock.shouldAdvance volatile
Key: PHOENIX-7825
URL: https://issues.apache.org/jira/browse/PHOENIX-7825
Project: Phoenix
Issue Type: Sub-task
Components: test
Reporter: Andrew Kyle Purtell
Assignee: Andrew Kyle Purtell
Fix For: 5.4.0, 5.3.1
{{ConcurrentMutationsIT}} installs a custom {{{}EnvironmentEdge{}}}, and the test thread calls {{setAdvance(false)}} to freeze time so two mutations share a timestamp. However, {{MyClock.shouldAdvance}} is not {{{}volatile{}}}, so the JMM does not require regionserver handler threads in the same JVM to observe the freeze, and in practice they keep advancing time, causing mutations expected to share a timestamp to diverge and producing non-deterministic index-scrutiny results, with the shared {{ParallelStatsDisabled}} mini-cluster amplifying the race.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)