Mark Payne created NIFI-5111:
--------------------------------
Summary: When cluster coordinator is changed, it can kick nodes
out of the cluster
Key: NIFI-5111
URL: https://issues.apache.org/jira/browse/NIFI-5111
Project: Apache NiFi
Issue Type: Bug
Reporter: Mark Payne
Assignee: Mark Payne
Fix For: 1.7.0
If Node A is elected cluster coordinator, all of the nodes in the cluster will
begin to send it heartbeats. It will store the latest heartbeat from each node
into a map along with a timestamp. If Cluster Coordinator role is then
delegated to another node and later Node A is elected again, it has potentially
very old, stale heartbeats. If it begins to monitor these heartbeats before
receiving a new heartbeat from a node, it will determine that said node has not
sent a heartbeat since before Node A lost its role as Cluster Coordinator. We
need to ensure that we are clearing out any previously stored heartbeats when a
node loses its role as Cluster Coordinator.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
