[
https://issues.apache.org/jira/browse/CASSANDRA-17039?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17442929#comment-17442929
]
Caleb Rackliffe commented on CASSANDRA-17039:
---------------------------------------------
As I've noted
[here|https://github.com/apache/cassandra/pull/1309#discussion_r748525029], the
starting size (according to {{jamm}}) of the {{RepairSession}} is much higher
in Java 11 than Java 8, solely due to its {{taskExecutor}} field. Its type was
recently changed to {{ThreadPoolExecutorPlus}} in CASSANDRA-16925, so this
problem doesn't manifest in 4.0, but it's not clear to me why {{jamm}}
interprets it differently on Java 11. ([~beobal] [~benedict] I can dig a bit
more into the jamm accounting if either of you is interested.)
In terms of this Jira, I went the direction of making the existing assertion in
{{RepairJobTest.testNoTreesRetainedAfterDifference}} (i.e. that expired worker
threads release tree memory) less brittle. Our expectations are now grounded to
the starting size of the session before any tasks are submitted.
> RepairJobTest.testNoTreesRetainedAfterDifference fails consistently on Java 11
> ------------------------------------------------------------------------------
>
> Key: CASSANDRA-17039
> URL: https://issues.apache.org/jira/browse/CASSANDRA-17039
> Project: Cassandra
> Issue Type: Bug
> Components: Consistency/Repair
> Reporter: Brandon Williams
> Assignee: Caleb Rackliffe
> Priority: Normal
> Time Spent: 20m
> Remaining Estimate: 0h
>
> Sometimes fails an assertion:
> {noformat}
> Expecting:
> <10000L>
> to be less than:
> <10000L>
> {noformat}
> https://app.circleci.com/pipelines/github/driftx/cassandra/269/workflows/f2b0a738-0785-4011-9ac1-071837dc9170/jobs/2049/tests#failed-test-1
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
