[
https://issues.apache.org/jira/browse/NIFI-2360?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15390531#comment-15390531
]
ASF subversion and git services commented on NIFI-2360:
-------------------------------------------------------
Commit 6932a53ec91513e7c300fdfba2a79c45614a8cea in nifi's branch
refs/heads/master from [~markap14]
[ https://git-wip-us.apache.org/repos/asf?p=nifi.git;h=6932a53 ]
NIFI-2360: Leave ZooKeeper running when a node is disconnected. Do not allow
the last node in the cluster to be disconnected. Change
ClusterProtocoLHeartbeater to use RetryNTime retry strategy instead of
RetryForever because web requests could block on this
This closes #705
Signed-off-by: jpercivall <[email protected]>
> Improve robustness of cluster when relying on embedded zookeeper
> ----------------------------------------------------------------
>
> Key: NIFI-2360
> URL: https://issues.apache.org/jira/browse/NIFI-2360
> Project: Apache NiFi
> Issue Type: Bug
> Components: Core Framework
> Affects Versions: 1.0.0
> Reporter: Mark Payne
> Assignee: Mark Payne
> Fix For: 1.0.0
>
>
> I have a 3 node cluster running an embedded zookeeper. If 2 nodes disconnect,
> I no longer have a ZooKeeper quorum. As a result, the third node is also
> disconnected. If I had more nodes, they would also all be disconnected. As a
> result, I also cannot then go to the cluster page to re-connect them, since
> there is no cluster coordinator.
> We should make this more robust, predominantly by not starting & stopping the
> embedded zookeeper server when a node connects & disconnects from cluster but
> instead just leave the embedded zookeeper server running, even when the node
> is disconnected from the cluster.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
