[
https://issues.apache.org/jira/browse/CASSANDRA-16146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17225128#comment-17225128
]
Yifan Cai commented on CASSANDRA-16146:
---------------------------------------
Thanks David!
It looks like the backports in CASSANDRA-16127 is merged after I posted the [CI
link|https://issues.apache.org/jira/browse/CASSANDRA-16146?focusedCommentId=17203554&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-17203554].
Therefore, CI did not show any jvm dtest failures.
For the failures in ClientNetworkStopStartTest, the cause is that the cluster
without GOSSIP enabled does not set the instance's operation mode correctly.
(So it stuck at Starting and fail).
> Node state incorrectly set to NORMAL after nodetool disablegossip and
> enablegossip during bootstrap
> ---------------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-16146
> URL: https://issues.apache.org/jira/browse/CASSANDRA-16146
> Project: Cassandra
> Issue Type: Bug
> Components: Cluster/Gossip
> Reporter: Yifan Cai
> Assignee: Yifan Cai
> Priority: Normal
> Fix For: 3.0.x, 3.11.x, 4.0-beta3
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> At high level, {{StorageService#setGossipTokens}} set the gossip state to
> {{NORMAL}} blindly. Therefore, re-enabling gossip (stop and start gossip)
> overrides the actual gossip state.
>
> It could happen in the below scenario.
> # Bootstrap failed. The gossip state remains in {{BOOT}} / {{JOINING}} and
> code execution exits StorageService#initServer.
> # Operator runs nodetool to stop and re-start gossip. The gossip state gets
> flipped to {{NORMAL}}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
