You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@nifi.apache.org by "Mark Payne (JIRA)" <ji...@apache.org> on 2019/04/17 14:52:00 UTC

[jira] [Created] (NIFI-6222) Improve efficiency of node heartbeats

Mark Payne created NIFI-6222:
--------------------------------

             Summary: Improve efficiency of node heartbeats
                 Key: NIFI-6222
                 URL: https://issues.apache.org/jira/browse/NIFI-6222
             Project: Apache NiFi
          Issue Type: Improvement
          Components: Core Framework
            Reporter: Mark Payne
            Assignee: Mark Payne


Currently, when a node sends a heartbeat to the cluster coordinator, it includes the node's understanding of the cluster topology - all nodes, their current status, the 'update id' of the latest update, etc. This is okay for a small cluster. But when the size of the cluster grows to hundreds of nodes and beyond, it becomes quite expensive.

This information is included so that if the Cluster Coordinator determines that the node's understanding of the topology is incorrect, it can send the correct topology so that all nodes stay in-sync. However, this could be done much more efficiently if, instead of sending all of this info, we sent only a mapping of Node Identifier to Update ID. Then, if anything is incorrect, the Cluster Coordinator can respond with the most accurate information.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)