[
https://issues.apache.org/jira/browse/CASSANDRA-15581?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17213926#comment-17213926
]
Marcus Eriksson commented on CASSANDRA-15581:
---------------------------------------------
I think this task should focus mostly on the mechanics of picking sstables for
compaction, not the actual merging of sstables (though, that will of course
also be tested by anything we do here). What [~paulo] defined above would be a
good start
Major compaction changes were CASSANDRA-6696 and CASSANDRA-7019
* Run all tests with different amounts of data directories (1/5/20)
* Run all tests with different compaction strategies (LCS/STCS/TWCS)
* Run LCS tests with {{single_sstable_uplevel}} on/off - CASSANDRA-12526
* Bootstrap/decom/replace, make sure disk usage is balanced on new + old nodes
* Heavy compaction load + range movements
* Heavy compaction load + ALTER .. WITH compaction = ..
* Heavy compaction load + incremental repair / anticompaction
* Test large node upgrades with several data directories (3.0 -> 4.0 probably
most interesting here)
* Test `nodetool garbagecollect` with large datasets and many tombstones.
> 4.0 quality testing: Compaction
> -------------------------------
>
> Key: CASSANDRA-15581
> URL: https://issues.apache.org/jira/browse/CASSANDRA-15581
> Project: Cassandra
> Issue Type: Task
> Components: Test/dtest/python
> Reporter: Josh McKenzie
> Priority: Normal
> Fix For: 4.0-beta
>
>
> Reference [doc from
> NGCC|https://docs.google.com/document/d/1uhUOp7wpE9ZXNDgxoCZHejHt5SO4Qw1dArZqqsJccyQ/edit#]
> for context.
> *Shepherd: Marcus Eriksson*
> Alongside the local and distributed read/write paths, we'll also want to
> validate compaction. CASSANDRA-6696 introduced substantial
> changes/improvements that require testing (esp. JBOD).
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
