[
https://issues.apache.org/jira/browse/ZOOKEEPER-1777?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13786011#comment-13786011
]
Germán Blanco commented on ZOOKEEPER-1777:
------------------------------------------
Hello Flavio,
I am sure that for me is a worrying case, I can't tell if it is a valid case
for ZooKeeper in general.
The faults are the following (for a cluster of three nodes):
- One server fails, and stops for a while.
- During this while another server fails and loses all data.
If e.g. you have data in RAM, losing all data with a failure (and not losing it
with other failures) may be quite frequent.
I agree that this requires failure of 2 out of three nodes, which should not
happen, but in any case a result of permanent inconsistency is not acceptable
to me. Nodes can be taken out for maintenance and the administrator will bring
nodes up and down and never think that if he does so carelessly the system will
be broken forever when it is taken out of maintenance.
> 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)
