[
https://issues.apache.org/jira/browse/ZOOKEEPER-4694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17724073#comment-17724073
]
赵心鑫 commented on ZOOKEEPER-4694:
--------------------------------
The problem has been resolved. The asynchronous notification of the watch has
been delayed. We thought that the node was deleted after application A created
it. Actually, the B application received the watch asynchronous notification
too late and determined that the node does not exist
> Persistent node loss
> --------------------
>
> Key: ZOOKEEPER-4694
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-4694
> Project: ZooKeeper
> Issue Type: Bug
> Affects Versions: 3.4.6
> Reporter: 赵心鑫
> Priority: Major
>
> When the client uses version 3.3.1 of the jar package, and the server uses
> version 3.4.6 of the application, when the scheduled task registration is
> implemented with Quartz, the created persistent node suddenly disappears, and
> a log is added before the delete method of the jar package, which is not
> printed, so it is speculated that it is not a client-initiated delete
> operation
> This phenomenon is an occasional situation, it may not occur for a week, or
> it may occur twice a day, we tried to upgrade the client and server versions
> to 3.4.14, and it is still the same problem, is there anything wrong with
> Quartz
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
