[
https://issues.apache.org/jira/browse/CASSANDRA-14801?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17054911#comment-17054911
]
Aleksandr Sorokoumov commented on CASSANDRA-14801:
--------------------------------------------------
Thank you for a comprehensive response, [~benedict]! I am quite new to C* 4.0
code, so it will take me some time to ramp up. If anyone has planned to work on
this issue in the next 1-2 weeks, it probably makes sense for me to work on
something else. Otherwise, I'd be happy to contribute.
In the latter case, in the next couple of days, I plan to read on how pending
ranges are calculated and what changes since 3.11 introduced the bug. Then I'll
write a test case that reproduces the issue.
> calculatePendingRanges no longer safe for multiple adjacent range movements
> ---------------------------------------------------------------------------
>
> Key: CASSANDRA-14801
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14801
> Project: Cassandra
> Issue Type: Bug
> Components: Legacy/Coordination, Legacy/Distributed Metadata
> Reporter: Benedict Elliott Smith
> Priority: Normal
> Fix For: 4.0, 4.0-beta
>
>
> Correctness depended upon the narrowing to a {{Set<InetAddressAndport>}},
> which we no longer do - we maintain a collection of all {{Replica}}. Our
> {{RangesAtEndpoint}} collection built by {{getPendingRanges}} can as a result
> contain the same endpoint multiple times; and our {{EndpointsForToken}}
> obtained by {{TokenMetadata.pendingEndpointsFor}} may fail to be constructed,
> resulting in cluster-wide failures for writes to the affected token ranges
> for the duration of the range movement.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
