Support Questions
Find answers, ask questions, and share your expertise
Announcements
Alert: Welcome to the Unified Cloudera Community. Former HCC members be sure to read and learn how to activate your account here.

Nifi Stuck Offloading Node

Highlighted

Nifi Stuck Offloading Node

One of the nodes in our nifi cluster experienced some issues and had to be shutdown and deleted. In Cluster, under the hamburger menu, the node is showing as offloading for almost a week even though the node is deleted. Is there any way to forcefully remove the node from the cluster so that I can make changes to the flow?

3 REPLIES 3

Re: Nifi Stuck Offloading Node

Master Guru

@Derek Calderon

Since your NiFi cluster is waiting on a response from the offloading node, it will not give you an option to delete node from list thus reducing your cluster node count by one.

-

If you stop all your remaining nodes and then start them, he cluster will come back up with only those nodes which join during startup. The other node will no longer be present.

-

Thank you,

Matt

-

If you found this answer addressed your question, please take a moment to login in and click the "ACCEPT" link.

Re: Nifi Stuck Offloading Node

@Matt Clarke, do you mean stopping the nifi process on all nodes, and can a rolling restart of the process also work?

Re: Nifi Stuck Offloading Node

Master Guru

@Derek Calderon

Unfortunately, a rolling restart will not work here. When you shutdown only one node another node in the cluster would get elected as cluster coordinator and still retain list of nodes known to be part of cluster. By shutting down all nodes you essential wipe the slate clean of what nodes are in cluster. On start nodes will come up, components (processors, controller services, reporting tasks. etc..) will return to last known state (running, stopped, etc..), FlowFiles will be loaded back in to last reported connection queue, and processing will continue.
-

One of the nodes will get elected as the cluster coordinator by Zookeeper and then other nodes will learn from ZK who was elected and start sending heartbeats directly to that elected cluster coordinator to join the cluster. As node join they will be added to list of connected nodes.
-
The dead node that never checks in will no longer be in the list.

-

No need to stop components before stopping NiFi. They will return to last known state on start.

-

Thank you,
Matt

Don't have an account?
Coming from Hortonworks? Activate your account here