RFR: 8367795: HeadlessMalfunctionTest may fail due to timeout
Phil Race
prr at openjdk.org
Fri Sep 26 23:52:18 UTC 2025
On Wed, 24 Sep 2025 22:38:29 GMT, Sergey Bylokhov <serb at openjdk.org> wrote:
>> What's the command you are using to run the tests ?
>> make test .... something .. ?
>
>>Is that somehow propagated to jtreg concurrency too ?
>
> Yes, jtreg executed via make will use the same concurrency.
>
>>Why is this test special ?
>
> It is the only test in the jdk_desktop group that modifies a class file in the JDK via instrumentation. For some similar tests in other areas, the timeouts were increased by [JDK-8260555](https://bugs.openjdk.org/browse/JDK-8260555). Check the links from JDK-8260555, there are a few additional timeout issues have been reported.
>
>>What's the command you are using to run the tests ?
>
>
> bash configure --with-boot-jdk=~/jdk/jdk-24 --with-jtreg=~/jtreg && make images
> make run-test TEST="jtreg:jdk_desktop" JTREG='VERBOSE=summary;KEYWORDS=!headful&!printer;REPEAT_COUNT=1'
I think that given we need stability in tests, the concurrency of 48 on a 48 CPU system may be too much.
Some tests might themselves need multiple threads.
I think the build system should reduce it to maybe 75% of what it uses for the build.
But that's a discussion for a different forum.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/27397#discussion_r2383687871
More information about the client-libs-dev
mailing list