We have the similar requirements. We had small POC which worked very well. When we started assessing NFR, we stuck with this bottle neck.   Issue is: 1. We have a clustered environment with three nodes. 2. To check what we did, we had set of processors. And ran a flow. Then we stopped one of them from primary node. All information got reflected correctly on all nodes. 3. We scaled down primary node from where we ran the flow. 4. Earlier we were able to see replicated stuck/queued message on all non-primary nodes.  As soon as, primary node was down, other nodes does not show that queued message. 5. When we started back earlier primary node, we can see everything good.   Is there any plan for support for this HA scenario for apache ni-fi? https://cwiki.apache.org/confluence/display/NIFI/High+Availability+Processing   Please suggest. ... View more