[jira] [Commented] (NIFI-1678) Nodes in cluster should use ZooKeeper to store heartbeat messages instead of sending to NCM

["ASF GitHub Bot (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/NIFI-1678?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15221827#comment-15221827 ] 

ASF GitHub Bot commented on NIFI-1678:
--------------------------------------

Github user mcgilman commented on the pull request:

    https://github.com/apache/nifi/pull/317#issuecomment-204434802
  
    Ran into an issue with the PR as is. Specifically, I deployed the contributions to an existing cluster. The existing cluster had already stored state in ZK using the configuration from the state-management.xml. During start up the nodes were unable to heartbeat into ZK because the configuration in state-management.xml had used a username/password. In order for the heartbeats to be stored the ZK connection from nifi.properties would also need to be configured with the same credentials.
    
    As I mentioned in the PR for #301 we should address having duplicate connect string in nifi.properties and state-management.xml for ZK. A short term fix for addressing this issue would be to additionally duplicate the remaining ZK configuration from state-management to nifi.properties. We could take this temporary approach while we discuss a more appropriate long term strategy for ZK configuration prior the first 1.x release. Or alternatively, we could address that issue prior to re-submitting this PR.


> Nodes in cluster should use ZooKeeper to store heartbeat messages instead of sending to NCM
> -------------------------------------------------------------------------------------------
>
>                 Key: NIFI-1678
>                 URL: https://issues.apache.org/jira/browse/NIFI-1678
>             Project: Apache NiFi
>          Issue Type: Improvement
>          Components: Core Framework
>            Reporter: Mark Payne
>            Assignee: Mark Payne
>             Fix For: 1.0.0
>
>
> Currently, nodes send heartbeats to the NCM periodically in order to indicate that they are actively participating in the cluster. As we move away from using an NCM, we need these heartbeats to go somewhere else. ZooKeeper is a reasonable location to push the heartbeats to, as it provides the HA that we need



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)