[
https://issues.apache.org/jira/browse/CASSANDRA-3071?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13089588#comment-13089588
]
Jonathan Ellis commented on CASSANDRA-3071:
-------------------------------------------
Is it possible for this to remove an endpoint that it shouldn't?
E.g., X has token T
X moves to token U but node N was down
N comes back up, thinks T and U are both owned by X
node Y takes token T
we remove X from gossip
> Gossip state is not removed after a new IP takes over a token
> -------------------------------------------------------------
>
> Key: CASSANDRA-3071
> URL: https://issues.apache.org/jira/browse/CASSANDRA-3071
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Reporter: Brandon Williams
> Assignee: Brandon Williams
> Priority: Minor
> Fix For: 0.7.9, 0.8.5
>
> Attachments: 3071.txt
>
>
> When a new node takes over a token, the endpoint state in the gossiper is
> never removed for the old node.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
