RFR: 8359135: New test TestCPUTimeSampleThrottling fails intermittently [v2]
Erik Gahlin
egahlin at openjdk.org
Wed Jun 11 14:13:29 UTC 2025
On Wed, 11 Jun 2025 11:11:50 GMT, Johannes Bechberger <jbechberger at openjdk.org> wrote:
>> I increased the tolerated rate deviation to make the test less flaky on systems like PowerPC. The deviation comes from the thread not being in the Java and native states on some platforms compared to others.
>
> Johannes Bechberger has updated the pull request incrementally with two additional commits since the last revision:
>
> - Remove debug code
> - Make upper-bound narrower
My experience with tests that make a recording and check some time-related bounds is that they fail sooner or later. The hardware is slow, a GC happens etc.
Maybe you could use an event stream and let it run until you hit, let's say, 50-150 samples / second?
-------------
PR Comment: https://git.openjdk.org/jdk/pull/25745#issuecomment-2962997568
More information about the hotspot-jfr-dev
mailing list