[
https://issues.apache.org/jira/browse/ZOOKEEPER-1777?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13786016#comment-13786016
]
Flavio Junqueira commented on ZOOKEEPER-1777:
---------------------------------------------
If I understand your scenario correctly, you can't do what you want with
ZooKeeper. There is one server stopped (fault one) and right after you wipe out
the content of a second server (fault two). You have a three-node ensemble,
which tolerates only one fault. How do you expect ZooKeeper to deal with it?
Now, we can generalize this case in the following way. If there is a minority
disconnected, and you wipe out the data of one of the servers in the live
quorum, then ZooKeeper is not guaranteed to work because some committed
transactions lost majority. You're violating fundamental assumptions of the
replication protocol.
> Missing ephemeral nodes in one of the members of the ensemble
> -------------------------------------------------------------
>
> Key: ZOOKEEPER-1777
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-1777
> Project: ZooKeeper
> Issue Type: Bug
> Components: quorum
> Affects Versions: 3.4.5
> Environment: Linux, Java 1.7
> Reporter: Germán Blanco
> Assignee: Germán Blanco
> Priority: Blocker
> Fix For: 3.4.6, 3.5.0
>
> Attachments: logs_trunk.tar.gz, snaps.tar, ZOOKEEPER-1777.tar.gz
>
>
> In a 3-servers ensemble, one of the followers doesn't see part of the
> ephemeral nodes that are present in the leader and the other follower.
> The 8 missing nodes in "the follower that is not ok" were created in the end
> of epoch 1, the ensemble is running in epoch 2.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
