[
https://issues.apache.org/jira/browse/NIFI-13927?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17893125#comment-17893125
]
ASF subversion and git services commented on NIFI-13927:
--------------------------------------------------------
Commit 31d92b5115003991bf7bcca66c354ad9ba447f22 in nifi's branch
refs/heads/main from Pierre Villard
[ https://gitbox.apache.org/repos/asf?p=nifi.git;h=31d92b5115 ]
NIFI-13927 Use synchronized lists in PublishGCPubSub (#9449)
Signed-off-by: David Handermann <[email protected]>
> PublishGCPPubSub processor stop working and is stucked when using Record
> Oriented mode
> --------------------------------------------------------------------------------------
>
> Key: NIFI-13927
> URL: https://issues.apache.org/jira/browse/NIFI-13927
> Project: Apache NiFi
> Issue Type: Bug
> Components: Extensions
> Affects Versions: 1.26.0, 2.0.0-M4
> Reporter: Julien G.
> Assignee: Pierre Villard
> Priority: Major
> Attachments: EXAMPLE.json, nifi.tdump
>
> Time Spent: 1.5h
> Remaining Estimate: 0h
>
> When using the PublishGCPPubSub processor with the Record Oriented mode the
> processor will stay stucked. It won't processor any FlowFile. When the
> processor is terminated the thread isn't released by the processor.
> It FlowFile Oriented mode, it works fine.
> It seems to be linked to how many records are in the FlowFile, becase when
> having a single record in the FlowFile, it seems to have no issue (may be it
> take more time to appear).
> But if you start to have more records in the FlowFile, it won't work. To make
> the processor works again you either need to restart the node or removing and
> recreating the processor. But it's temporary, because it some point it will
> be stucked again. When stucked the processor won't be unstucked even after
> days (we had a processor stucked for 3 days straight).
> You can find attached a thread dump when the processor was running and
> stucked, and with ~10 threads not released . And you can also find an example
> that will show the issue with a FlowFile of 500 records that instantly
> stucked in the publish.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
