[
https://issues.apache.org/jira/browse/CASSANDRA-10842?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15106750#comment-15106750
]
Sylvain Lebresne commented on CASSANDRA-10842:
----------------------------------------------
Looking at the last few failures, it seems the throughput we get is higher but
pretty close to the expected one (for instance {{5.092345}} instead of
{{5.0}}). I don't think this is really a problem, especially since, as far as I
can tell, the throughput during compaction is done on the input files, while
the value computed at the end of compaction (that the test use) is based on the
output files, and so I don't think we can expect things to be too exact. I've
pushed a trivial pull request at
https://github.com/riptano/cassandra-dtest/pull/752 to round the avg throughput
computed post-compaction. It's probably not perfect but as far as I can tell
from the failure history, that should be enough to make the test pass reliably.
> compaction_throughput_tests are failing
> ---------------------------------------
>
> Key: CASSANDRA-10842
> URL: https://issues.apache.org/jira/browse/CASSANDRA-10842
> Project: Cassandra
> Issue Type: Sub-task
> Components: Compaction, Testing
> Reporter: Philip Thompson
> Fix For: 2.1.x, 2.2.x, 3.0.x, 3.x
>
>
> {{compaction_test.TestCompaction_with_DateTieredCompactionStrategy.compaction_throughput_test}}
> and
> {{compaction_test.TestCompaction_with_LeveledCompactionStrategy.compaction_throughput_test}}
> are failing on 3.0-head, 2.1-head, and 2.2-head for the last two builds.
> See: http://cassci.datastax.com/job/cassandra-3.0_dtest/429/testReport/
> The test sets compaction throughput to 5, via nodetool, but is finding an
> average throughput greater than 5 in the logs. I cannot reproduce this
> locally on OSX.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
