[
https://issues.apache.org/jira/browse/CASSANDRA-17525?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17711465#comment-17711465
]
Jon Meredith commented on CASSANDRA-17525:
------------------------------------------
I think this resolves the issue and makes the change I was going to apply here
too.
> Local host-id may not be set when replaying commit log
> ------------------------------------------------------
>
> Key: CASSANDRA-17525
> URL: https://issues.apache.org/jira/browse/CASSANDRA-17525
> Project: Cassandra
> Issue Type: Bug
> Components: Local/Commit Log
> Reporter: Jon Meredith
> Assignee: Jon Meredith
> Priority: Normal
> Fix For: 4.0.x, 4.1.x
>
>
> The local host id is now used when replaying commit logs to prevent data loss
> when moving SSTables between nodes (CASSANDRA-16619).
> The local host-id is assigned once on initial startup and persisted to the
> local.local table for use on subsequent startups - however it remains in the
> memtable/commitlog until the local table is eventually flushed.
> If the node dies for any reason before the flush takes place, on the next
> restart a new host id will be allocated before the commit log is replayed and
> any sstables appearing in the commit log will be skipped incorrectly.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org
