[
https://issues.apache.org/jira/browse/CASSANDRA-19026?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17789236#comment-17789236
]
Jacek Lewandowski commented on CASSANDRA-19026:
-----------------------------------------------
Some conclusions - I've noticed that disabling flushing has different effect on
my laptop with disk encryption and on a bare metal box with Ryzen without disk
encryption. The diff is significant, while I could observe almost 2x
improvement in testing speed on my laptop, the diff on Ryzen was between
patched / not-patched version was smaller ~1.25x. It is still significant
speedup though.
Although it may look as the disk encryption plays the main role here, I cannot
confirm whether it is a real cause or just an accidental correlation.
In a separate ticket, I'll consider using {{CQLTester.InMemory}} for the tests
which can work with it, and maybe refactor the others. I wish I knew about that
great application of JimFS in our testing infrastructure earlier.
> Do not drop CQLTester.KEYSPACE content in @After asynchronously
> ---------------------------------------------------------------
>
> Key: CASSANDRA-19026
> URL: https://issues.apache.org/jira/browse/CASSANDRA-19026
> Project: Cassandra
> Issue Type: Task
> Components: Test/unit
> Reporter: Jacek Lewandowski
> Assignee: Jacek Lewandowski
> Priority: Normal
> Fix For: 4.0.x, 4.1.x, 5.0.x, 5.1
>
>
> CASSANDRA-18710 showed that removal of tables, views, functions, etc. in
> {{@After}} method asynchronously may span the next test case and modify that
> test behaviour in some unexpected way
> Concretely, {{@After}} method includes dropping {{KEYSPACE_PER_TEST}}
> synchronously and the whole content of {{KEYSPACE}} asynchronously. I don't
> see a reason for dropping the content of {{KEYSPACE}} after each test because
> that keyspace is not expected to be clean for each test case. It is the case
> for {{KEYSPACE_PER_TEST}}. Therefore I propose removal of that async cleanup
> of {{KEYSPACE}} after each test case.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
