[
https://issues.apache.org/jira/browse/CASSANDRA-15773?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17100342#comment-17100342
]
David Capwell commented on CASSANDRA-15773:
-------------------------------------------
*
https://github.com/apache/cassandra/compare/trunk...spmallette:CASSANDRA-15773-trunk#diff-790a43fcd6ed69f4122cd0fe205c847bR72.
leaking, should cleanup after the test is over; same for the other test
*
https://github.com/apache/cassandra/compare/trunk...spmallette:CASSANDRA-15773-trunk#diff-790a43fcd6ed69f4122cd0fe205c847bR75
can this be >= totalBytesRequestedFromPool? Since we don't free, and the
metric is how many chunks are leant out, >= totalBytesRequestedFromPool should
hold true and would be more accurate than > 0
* the test makes an assumption at buffer pool size, which may not be true in
CI, see [1] pool size calculation. For this reason the tests can be flaky
under environments with less resources. One way to deal with this is to define
file_cache_size_in_mb in test/conf/cassandra.yaml, that way memory is constant
for all environments, and the loops can be adjusted to it.
Overall the patch LGTM, mostly small stuff.
[1] -
{code}
if (conf.file_cache_size_in_mb == null)
conf.file_cache_size_in_mb = Math.min(512, (int)
(Runtime.getRuntime().maxMemory() / (4 * 1048576)));
{code}
> Add test to cover metrics related to the BufferPool
> ---------------------------------------------------
>
> Key: CASSANDRA-15773
> URL: https://issues.apache.org/jira/browse/CASSANDRA-15773
> Project: Cassandra
> Issue Type: Improvement
> Components: Test/unit
> Reporter: Stephen Mallette
> Assignee: Stephen Mallette
> Priority: Normal
>
> At this time there do not appear to be unit tests to validate
> {{BufferPoolMetrics}}.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
