[
https://issues.apache.org/jira/browse/NIFI-11333?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mark Payne updated NIFI-11333:
------------------------------
Fix Version/s: 2.latest
> Drop data from removed connections on cluster reconnect
> -------------------------------------------------------
>
> Key: NIFI-11333
> URL: https://issues.apache.org/jira/browse/NIFI-11333
> Project: Apache NiFi
> Issue Type: Improvement
> Components: Core Framework
> Reporter: Mark Payne
> Priority: Major
> Fix For: 2.latest
>
>
> In 1.16 we allows users to start updating flows while nodes are disconnected.
> This has been greatly helpful. However, it can lead to a problem: when a user
> removes a connection and there's data queued on a disconnected node, that
> disconnected node can no longer rejoin the cluster. Instead, it remains
> disconnected; and if the node is shutdown, it cannot be restarted without
> manually changing nifi.properties to change it from a clustered not to a
> standalone node, then restarting, and bleeding the data out, shutting down,
> manually updating properties to make it a clustered node again; and
> restarting.
> This is painful. And it is necessary because of the guardrails that have been
> put into place in order to ensure that nodes won't drop data. However, in
> this case, the user has already explicitly indicated that they want to delete
> the connection. We should instead just drop the data that lives in that
> connection, log an INFO-level notice that the data was dropped as instructed,
> and continue on.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
