[
https://issues.apache.org/jira/browse/FLINK-3301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15126410#comment-15126410
]
ASF GitHub Bot commented on FLINK-3301:
---------------------------------------
Github user mxm commented on the pull request:
https://github.com/apache/flink/pull/1560#issuecomment-178029684
Thanks for the pull request. The issue has been identified in the meantime.
It is only a testing issue and not a synchronization issue in the RMQSource.
> Ineffective synchronization in MessageAcknowledgingSourceBase#restoreState
> --------------------------------------------------------------------------
>
> Key: FLINK-3301
> URL: https://issues.apache.org/jira/browse/FLINK-3301
> Project: Flink
> Issue Type: Bug
> Reporter: Ted Yu
>
> Here is related code:
> {code}
> public void restoreState(SerializedCheckpointData[] state) throws Exception
> {
> synchronized (pendingCheckpoints) {
> pendingCheckpoints = SerializedCheckpointData.toDeque(state,
> idSerializer);
> {code}
> When lock is held on current pendingCheckpoints field, new reference is
> assigned to the same field - making the synchronization ineffective.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
