[
https://issues.apache.org/jira/browse/CASSANDRA-19863?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17889642#comment-17889642
]
Berenguer Blasi commented on CASSANDRA-19863:
---------------------------------------------
In the end it wasn't necessary to do any big re-work, just adjust the junit:
- CASSANDRA-19863 Take 2 5.0 [PR|https://github.com/apache/cassandra/pull/3615]
- CASSANDRA-19863 Take 2 trunk
[PR|https://github.com/apache/cassandra/pull/3616]
[~blambov] this should be a super quick review for you.
> Add extra compaction junit
> --------------------------
>
> Key: CASSANDRA-19863
> URL: https://issues.apache.org/jira/browse/CASSANDRA-19863
> Project: Cassandra
> Issue Type: Improvement
> Components: Local/Compaction
> Reporter: Berenguer Blasi
> Assignee: Berenguer Blasi
> Priority: Normal
> Fix For: 4.0.14, 4.1.7, 5.0.1, 5.1
>
>
> Working on a custom compaction strategy we have found there is no junit
> making sure compactions strategies account for this specific scenario:
> A- IR sessions starts on N sstables
> B- Just before the IR session completes and the N sstables get moved out of
> pending repair a compaction kicks in
> C- As the IR session is finalizing the compaction task removes the N sstables
> and creates a new one that will obviously be pending repair.
> D- The IR session finalizes, the original N sstables are gone, the compaction
> finishes and a new sstable pending repair is created
> E- That new sstable is orphaned as A. No active IR session knows about it B.
> No compaction is going to ever touch it as long as it is pending repair. So
> it it’s just left behind forever.
> This ticket will add such junit.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
