Installation fails on Ubuntu 24.04 + kernel 6.14 - JVM cgroup v2 compatibility issue

Individua le immagini Docker e i punti di ingresso dell’installazione utilizzati per i pod Kafka e Controller, quindi esamina le versioni JVM incluse. Riproduci il fallimento su Ubuntu 24.04 con cgroup v2 e verifica che le modifiche alle immagini scelte consentano di completare l’installazione con controller-0 e gli altri pod in buone condizioni.

Ⅰ. Issue Description

OpenServerless installation fails on Ubuntu 24.04 with kernel 6.14 due to JVM cgroup v2 compatibility issue. Kafka and Controller pods crash with NullPointerException in CgroupV2Subsystem.getInstance.

Ⅱ. Describe what happened

Kafka pod crashes immediately on startup with the following exception:

Exception in thread "main" java.lang.reflect.InvocationTargetException
    at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
    at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(Unknown Source)
    at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(Unknown Source)
    at java.base/java.lang.reflect.Method.invoke(Unknown Source)
    at java.instrument/sun.instrument.InstrumentationImpl.loadClassAndStartAgent(Unknown Source)
    at java.instrument/sun.instrument.InstrumentationImpl.loadClassAndCallPremain(Unknown Source)
Caused by: java.lang.NullPointerException
    at java.base/jdk.internal.platform.cgroupv2.CgroupV2Subsystem.getInstance(Unknown Source)
    at java.base/jdk.internal.platform.CgroupSubsystemFactory.create(Unknown Source)
    at java.base/jdk.internal.platform.CgroupMetrics.getInstance(Unknown Source)
    at java.base/jdk.internal.platform.SystemMetrics.instance(Unknown Source)
    at java.base/jdk.internal.platform.Metrics.systemMetrics(Unknown Source)
    at java.base/jdk.internal.platform.Container.metrics(Unknown Source)
    at jdk.management/com.sun.management.internal.OperatingSystemImpl.<init>(Unknown Source)
    at jdk.management/com.sun.management.internal.PlatformMBeanProviderImpl.getOperatingSystemMXBean(Unknown Source)
    at jdk.management/com.sun.management.internal.PlatformMBeanProviderImpl$3.nameToMBeanMap(Unknown Source)
    at java.management/sun.management.spi.PlatformMBeanProvider$PlatformComponent.getMBeans(Unknown Source)
    at java.management/java.lang.management.ManagementFactory.getPlatformMXBean(Unknown Source)
    at java.management/java.lang.management.ManagementFactory.getOperatingSystemMXBean(Unknown Source)
    at io.prometheus.jmx.shaded.io.prometheus.client.hotspot.StandardExports.<init>(StandardExports.java:43)
    at io.prometheus.jmx.shaded.io.prometheus.client.hotspot.DefaultExports.register(DefaultExports.java:37)
    at io.prometheus.jmx.shaded.io.prometheus.client.hotspot.DefaultExports.initialize(DefaultExports.java:28)
    at io.prometheus.jmx.JavaAgent.premain(JavaAgent.java:30)
    ... 6 more
*** java.lang.instrument ASSERTION FAILED ***: "result" with message agent load/premain call failed at src/java.instrument/share/native/libinstrument/JPLISAgent.c line: 422
FATAL ERROR in native method: processing of -javaagent failed, processJavaStart failed

The Controller pod never starts - installation hangs indefinitely waiting for pod/controller-0 to become available.

Ⅲ. Describe what you expected to happen

Installation should complete successfully with all pods running and healthy, allowing user creation and login.

Ⅳ. How to reproduce it (as minimally and precisely as possible)

1. Install Ubuntu 24.04 on hardware with NVIDIA kernel 6.14
2. Configure ops with minimal settings
3. Run installation
4. Installation fails with timeout waiting for controller-0

Ⅴ. Anything else we need to know?

Root Cause Analysis:

This is a known issue with JVM versions prior to JDK 21 running on Linux kernel 6.12+ with cgroup v2. The problem occurs because:

1. Ubuntu 24.04 uses kernel 6.x + systemd 256+ which enforces cgroup v2
2. The memory cgroup controller is missing from /proc/cgroups:

   $ cat /proc/cgroups
   #subsys_name    hierarchy    num_cgroups    enabled
cpu    0    860    1
cpuacct    0    860    1
blkio    0    860    1
devices    0    860    1
freezer    0    860    1
net_cls    0    860    1
perf_event    0    860    1
net_prio    0    860    1
hugetlb    0    860    1
pids    0    860    1
rdma    0    860    1
misc    0    860    1
dmem    0    860    1

Note: No "memory" controller present
3. JVM's CgroupV2Subsystem.getInstance() expects the memory controller and crashes with NPE when it's missing

Temporary Workaround:

Force cgroup v1 by adding kernel boot parameter:

# In /etc/default/grub:
GRUB_CMDLINE_LINUX="systemd.unified_cgroup_hierarchy=0"
# Then: sudo update-grub && sudo reboot

Permanent Solution:

Update Docker images to use JDK 21+ or apply JVM patches for cgroup v2 compatibility. Similar issues have been fixed in:

• Elasticsearch 7.17.26+
• OpenJDK bug tracker: https://bugs.openjdk.org/browse/JDK-8287107

Ⅵ. Environment:

• K8S Runtime and version: k3s (installed by ops)
• OPS CLI version: 0.1.0-2409121919.dev
• OS: Ubuntu 24.04 LTS
• Kernel: 6.14.0-1015-nvidia
• Hardware: Dell Pro Max GB10 (NVIDIA GB10 Grace CPU + Blackwell GPU)
• Java version on host: OpenJDK 1.8.0_472
• Cgroup version: v2 (enforced by kernel 6.14)