[jdk11u-dev] RFR: 8347129: cpuset cgroups controller is required for no good reason

[Voznia Anton]

On Tue, 11 Nov 2025 11:52:31 GMT, Severin Gehwolf <sgehwolf at openjdk.org> wrote:

>> Hi all,
>> 
>> Clean backport of JDK-8347129
>> 
>> Ran test locally on Fedora 43 arm, under `test/hotspot/jtreg/containers/`
>> 
>> Without the fix:
>> 
>> containers/cgroup/CgroupSubsystemFactory.java  Passed. Execution successful
>> containers/cgroup/PlainRead.java               Passed. Execution successful
>> containers/docker/DockerBasicTest.java         Passed. Execution successful
>> containers/docker/ShareTmpDir.java             Passed. Execution successful
>> containers/docker/TestCPUAwareness.java        Failed. Execution failed: `main' threw exception: java.lang.RuntimeException: 'active_processor_count.*1' missing from stdout/stderr
>> containers/docker/TestCPUSets.java             Failed. Execution failed: `main' threw exception: java.lang.RuntimeException: assertTrue: expected true, was false
>> containers/docker/TestJFREvents.java           Failed. Execution failed: `main' threw exception: java.lang.RuntimeException: 'totalSize = 209715200' missing from stdout/stderr
>> containers/docker/TestJFRNetworkEvents.java    Passed. Execution successful
>> containers/docker/TestJcmdWithSideCar.java     Passed. Execution successful
>> containers/docker/TestMemoryAwareness.java     Failed. Execution failed: `main' threw exception: java.lang.RuntimeException: 'Memory Limit is:.*104857600' missing from stdout/stderr
>> containers/docker/TestMemoryWithCgroupV1.java  Passed. Execution successful
>> containers/docker/TestMisc.java                Failed. Execution failed: `main' threw exception: java.lang.RuntimeException: 'CheckContainerized: Running inside a container' missing from stdout/stderr
>> containers/docker/TestPids.java                Failed. Execution failed: `main' threw exception: java.lang.RuntimeException: assertTrue: expected true, was false
>> 
>> 
>> With the fix:
>> 
>> containers/cgroup/CgroupSubsystemFactory.java  Passed. Execution successful
>> containers/cgroup/PlainRead.java               Passed. Execution successful
>> containers/docker/DockerBasicTest.java         Passed. Execution successful
>> containers/docker/ShareTmpDir.java             Passed. Execution successful
>> containers/docker/TestCPUAwareness.java        Failed. Execution failed: `main' threw exception: java.lang.RuntimeException: Expected to get exit value of [0]
>> containers/docker/TestCPUSets.java             Passed. Execution successful
>> containers/docker/TestJFREvents.java           Passed. Execution successful
>> containers/docker/TestJFRNetworkEvents.java    Passed. Execution successfu...
>
>> Test `containers/docker/TestCPUAwareness.java` fails because of JDK-8282684.
> 
> I'm pretty sure the test fails because of [JDK-8370492](https://bugs.openjdk.org/browse/JDK-8370492) on F43. JDK-8282684 is the reason why it doesn't show up for JDK 17+ (i.e. isn't tested there, but the issue exists).

@jerboaa thank you for the note.

In my case the test failed `TestCPUAwareness.java` because of
"java.lang.RuntimeException: 'active_processor_count.*1' missing from stdout/stderr "

The test expects one active processor count, but the actual value is two:
[0.002s][trace][os,container] CgroupSubsystem::active_processor_count (cached): 2

Does the cpushare affect the active processor count number?

-------------

PR Comment: https://git.openjdk.org/jdk11u-dev/pull/3109#issuecomment-3527449123