[
https://issues.apache.org/jira/browse/CASSANDRA-11657?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15280149#comment-15280149
]
Marcus Eriksson commented on CASSANDRA-11657:
---------------------------------------------
Tried to reproduce this today by running {{testParallelLeveledCompaction}} in a
loop, never failed, then I realized that the new {{testLeveledScanner}} breaks
the leveled manifest and we can't get compaction tasks after that (the sstable
is in the wrong level and we can't remove it from the manifest then:
https://github.com/apache/cassandra/blob/trunk/src/java/org/apache/cassandra/db/compaction/LeveledManifest.java#L489).
This gets getNextBackgroundTask stuck in a loop since we try marking a
non-existing sstable as compacting.
I have no idea why this test failed before CASSANDRA-9935 but I cant reproduce
that anymore.
Patch to run on separate tables here:
https://github.com/krummas/cassandra/commits/marcuse/11657-2
> LongLeveledCompactionStrategyTest broken since CASSANDRA-10099
> --------------------------------------------------------------
>
> Key: CASSANDRA-11657
> URL: https://issues.apache.org/jira/browse/CASSANDRA-11657
> Project: Cassandra
> Issue Type: Bug
> Reporter: Marcus Eriksson
> Assignee: Marcus Eriksson
> Fix For: 3.x
>
>
> Seems CASSANDRA-10099 broke LongLeveledCompactionStrategyTest
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
