[
https://issues.apache.org/jira/browse/ZOOKEEPER-3936?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17198078#comment-17198078
]
maoling commented on ZOOKEEPER-3936:
------------------------------------
[~badai]
_*-->After this, zxid in the new leader (node 5) becomes the same as the old
leader (node 2).*_
What behaviour do you want about this? If any write requests in this process,
it's a expected behaviour?
Could you please give us more details?
> Zxid mismatch after new node becomes the leader
> -----------------------------------------------
>
> Key: ZOOKEEPER-3936
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-3936
> Project: ZooKeeper
> Issue Type: Bug
> Affects Versions: 3.5.8
> Reporter: Badai Aqrandista
> Priority: Minor
> Attachments: logs-zk1.tar.gz, logs-zk2.tar.gz, logs-zk3.tar.gz,
> logs-zk4.tar.gz, logs-zk5.tar.gz
>
>
> Adding 2 nodes to 3 nodes ensemble with the following steps:
> 1. Start the new nodes (node 4, 5) and let them join the ensemble
> 2. Update existing nodes' configuration (node 1, 2, 3) to add the new nodes
> addresses and ports
> 3. Rolling restart the existing nodes
> 4. Stop the current leader (node 2)
> 5. New node (node 5) becomes leader
> After this, zxid in the new leader (node 5) becomes the same as the old
> leader (node 2).
> {noformat}
> badai@192-168-1-129 3 % grep -i zxid logs-zk*/logs/before/*stat*
> logs-zk1/logs/before/zk-1-stat-before-crash.log:Zxid: 0xe0000050c
> logs-zk2/logs/before/zk-2-stat-before-crash.log:Zxid: 0xf00000000
> logs-zk3/logs/before/zk-3-stat-before-crash.log:Zxid: 0xe0000050c
> logs-zk4/logs/before/zk-4-stat-before-crash.log:Zxid: 0xe0000050c
> logs-zk5/logs/before/zk-5-stat-before-crash.log:Zxid: 0xf00000000
> badai@192-168-1-129 3 %
> {noformat}
> Please ignore the "before" wording on the logs name as they are irrelevant.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
