Update JVM_IsUseContainerSupport

[babsingh]

Currently, in comparison to the RI, OpenJ9 shows a different output for
-XshowSettings:system outside a container because our
JVM_IsUseContainerSupport implementation differs from the RI.

This PR matches the latest RI behaviour for JVM_IsUseContainerSupport.

The RI returns TRUE from JVM_IsUseContainerSupport IFF
-XX:+UseContainerSupport is specified. This option is enabled by
default.

Currently, we return TRUE from JVM_IsUseContainerSupport if
-XX:+UseContainerSupport is specified && we are inside a container.

The return value of JVM_IsUseContainerSupport determines the output
of -XshowSettings:system.

If JVM_IsUseContainerSupport returns FALSE, -XshowSettings:system has
the below output:

Operating System Metrics:
    No metrics available for this platform

If JVM_IsUseContainerSupport returns TRUE, -XshowSettings:system has
the below output:

Operating System Metrics:
    Provider: cgroupv2
    Effective CPU Count: 8
    CPU Period: -1
    CPU Quota: -1
    CPU Shares: -1
    List of Processors: N/A
    List of Effective Processors: N/A
    List of Memory Nodes: N/A
    List of Available Memory Nodes: N/A
    Memory Limit: Unlimited
    Memory Soft Limit: 0.00K
    Memory & Swap Limit: Unlimited
    Maximum Processes Limit: Unlimited

Signed-off-by: Babneet Singh sbabneet@ca.ibm.com

[babsingh]

@pshipton @tajila I would also like to backport these changes to the Oct release. I feel that the risk is low. Please let me know if you agree on backporting these changes.

fyi @mstoodle

[mstoodle]

i suppose the main risk is that these changes could enable downstream code to execute outside of containers whereas previously it would have only executed in a container context. Presumably that environment is also where most of our testing runs, but I think we should run tests cross-platform before we consider to merge it especially given where we are in the release cycle. But all that's irrelevant for the backport question unless @pshipton and @tajila agree it's worthwhile to consider at this point.

[tajila]

It looks like the usage of JVM_IsUseContainerSupport is limited to jdk.internal.platform.CgroupMetrics which as far as I can tell is only used for printing System metrics in the launcher code for -XshowSettings:system

[babsingh]

The cgroups Java API associated with -XshowSettings:system and JVM_IsUseContainerSupport: https://github.com/ibmruntimes/openj9-openjdk-jdk8/tree/openj9/jdk/src/linux/classes/jdk/internal/platform. Internally, OpenJ9 doesn't use this Java API; it relies upon the cgroups API from the OMR port library. From a JVM perspective, there should be no change to OpenJ9 behaviour with this PR. The impact will be seen on the class library side (extensions repo), which includes common/shared code between OpenJ9 and the RI. With this PR, we will just inherit the risk that the RI has already accepted with their implementation of JVM_IsUseContainerSupport. My guess ... any problems that show up in OpenJ9 SDKs will also exist in the RI SDKs.

[pshipton]

jenkins test sanity.openjdk,extended.openjdk amac jdk11

[pshipton]

jenkins test sanity.openjdk,extended.openjdk xlinux jdk11

[babsingh]

Old PR builds:

• JDK11 amac: https://openj9-jenkins.osuosl.org/view/Pull%20Requests/job/PullRequest-OpenJ9/4448 (aborted)
• JDK11 xlinux: https://openj9-jenkins.osuosl.org/view/Pull%20Requests/job/PullRequest-OpenJ9/4449 (still running)

[pshipton]

Pls go ahead with the backport.