chore(kafka, kafka-operator): drop the JVM sizing workaround

The workaround passed MaxRAM and ActiveProcessorCount to every Strimzi
JVM because the JDK 17 in the 0.45 images could not see the container
limits on kernels without CONFIG_MEMCG_V1 (JDK-8347811). The Strimzi
1.2.0 images ship JDK 21.0.12, which carries the fix. On such a kernel,
with a 1Gi limit, Strimzi's heap probe in the Kafka image now
estimates a 495M maximum heap, where the 0.45.1-rc1 image estimated
13.9G.

Assisted-by: LLM
Signed-off-by: Aleksei Sviridkin <f@lex.la>
This commit is contained in:
Aleksei Sviridkin committed 2026-11-17 12:22:54 +03:00
1 parent 641faa7fde
commit a909b733ea
7 files changed
+1 -174

No files matched your search

-20
View File
@@ -1,20 +0,0 @@
{{- /*
JAVA_TOOL_OPTIONS for a Strimzi-managed container, from its resource limits.
The JDK 17 in the Strimzi images cannot see the container limits on kernels
built without CONFIG_MEMCG_V1 (JDK-8347811) and sizes itself from the host:
Strimzi's heap probe then hands every broker up to a 5Gi heap. Every java in
the container reads this variable, the probe included, so Strimzi's own
heap rules work again. jvmOptions -XX would not do: Strimzi moves those into
KAFKA_JVM_PERFORMANCE_OPTS, and kafka-run-class.sh drops its G1 tuning when
that variable is set. ActiveProcessorCount only when the chart sets a CPU
limit: a default a LimitRange adds at admission is not visible here, and
Strimzi template env takes no resourceFieldRef, so such a JVM still counts
the host CPUs.
*/ -}}
{{- define "kafka.javaToolOptions" -}}
{{- $opts := list (printf "-XX:MaxRAM=%d" (include "cozy-lib.resources.toFloat" .memory | float64 | int64)) -}}
{{- with .cpu -}}
{{- $opts = append $opts (printf "-XX:ActiveProcessorCount=%d" (include "cozy-lib.resources.toFloat" . | float64 | ceil | int64)) -}}
{{- end -}}
{{- join " " $opts -}}
{{- end -}}
+1 -10
View File
@@ -80,8 +80,7 @@ spec:
# The topic-operator and user-operator are JVM processes and OOMKill at
# that default, leaving the entity-operator in CrashLoopBackOff so
# KafkaTopics/KafkaUsers never reconcile. Set explicit resources so they
# don't inherit the namespace default. JAVA_TOOL_OPTIONS: see
# kafka.javaToolOptions.
# don't inherit the namespace default.
topicOperator:
resources:
requests:
@@ -107,11 +106,3 @@ spec:
labels:
policy.cozystack.io/allow-to-apiserver: "true"
enableServiceLinks: false
topicOperatorContainer:
env:
- name: JAVA_TOOL_OPTIONS
value: {{ include "kafka.javaToolOptions" (dict "memory" $entityOperatorMemory) | quote }}
userOperatorContainer:
env:
- name: JAVA_TOOL_OPTIONS
value: {{ include "kafka.javaToolOptions" (dict "memory" $entityOperatorMemory) | quote }}
@@ -107,12 +107,6 @@ spec:
roles:
- broker
resources: {{- include "cozy-lib.resources.defaultingSanitize" (list .Values.kafka.resourcesPreset .Values.kafka.resources $) | nindent 18 }}
# JAVA_TOOL_OPTIONS: see kafka.javaToolOptions.
template:
kafkaContainer:
env:
- name: JAVA_TOOL_OPTIONS
value: {{ include "kafka.javaToolOptions" (include "cozy-lib.resources.defaultingSanitize" (list .Values.kafka.resourcesPreset .Values.kafka.resources $) | fromYaml).limits | quote }}
storage:
type: jbod
volumes:
@@ -152,11 +146,6 @@ spec:
roles:
- controller
resources: {{- include "cozy-lib.resources.defaultingSanitize" (list .Values.kafka.resourcesPreset .Values.kafka.resources $) | nindent 18 }}
template:
kafkaContainer:
env:
- name: JAVA_TOOL_OPTIONS
value: {{ include "kafka.javaToolOptions" (include "cozy-lib.resources.defaultingSanitize" (list .Values.kafka.resourcesPreset .Values.kafka.resources $) | fromYaml).limits | quote }}
storage:
type: jbod
volumes:
@@ -1,83 +0,0 @@
suite: Kafka JVMs are sized from their container limits
templates:
- templates/migration-hook.yaml
- templates/kafka.yaml
release:
name: test-kafka
namespace: tenant-test
tests:
- it: passes the default preset limits to every node pool JVM
template: templates/migration-hook.yaml
documentIndex: 3
asserts:
- matchRegex:
path: spec.template.spec.containers[0].args[0]
pattern: "release-namespace: \\$\\{ns\\}\\s+spec:\\s+replicas: \\d+\\s+roles:\\s+- broker\\s+resources:\\s+limits:(\\s+\\S+: \\S+)+\\s+requests:(\\s+\\S+: \\S+)+(\\s+#[^\\n]*)*\\s+template:\\s+kafkaContainer:\\s+env:\\s+- name: JAVA_TOOL_OPTIONS\\s+value: \"-XX:MaxRAM=1073741824 -XX:ActiveProcessorCount=1\"\\s+storage:"
- matchRegex:
path: spec.template.spec.containers[0].args[0]
pattern: "roles:\\s+- controller\\s+resources:\\s+limits:(\\s+\\S+: \\S+)+\\s+requests:(\\s+\\S+: \\S+)+(\\s+#[^\\n]*)*\\s+template:\\s+kafkaContainer:\\s+env:\\s+- name: JAVA_TOOL_OPTIONS\\s+value: \"-XX:MaxRAM=1073741824 -XX:ActiveProcessorCount=1\"\\s+storage:"
# Strimzi moves jvmOptions -XX into KAFKA_JVM_PERFORMANCE_OPTS, and
# kafka-run-class.sh then drops its G1 settings.
- it: leaves the node pool JVM options to Strimzi and Kafka
template: templates/migration-hook.yaml
documentSelector:
path: kind
value: Job
asserts:
- notMatchRegex:
path: spec.template.spec.containers[0].args[0]
pattern: "jvmOptions"
- it: rounds a fractional CPU limit up
template: templates/migration-hook.yaml
documentIndex: 3
set:
kafka.resources:
cpu: 2500m
memory: 3Gi
asserts:
- matchRegex:
path: spec.template.spec.containers[0].args[0]
pattern: "roles:\\s+- controller\\s+resources:\\s+limits:(\\s+\\S+: \\S+)+\\s+requests:(\\s+\\S+: \\S+)+(\\s+#[^\\n]*)*\\s+template:\\s+kafkaContainer:\\s+env:\\s+- name: JAVA_TOOL_OPTIONS\\s+value: \"-XX:MaxRAM=3221225472 -XX:ActiveProcessorCount=3\"\\s+storage:"
- it: takes the limits, not the requests, when the allocation ratios shrink the requests
template: templates/migration-hook.yaml
documentIndex: 3
set:
_cluster:
cpu-allocation-ratio: "4"
memory-allocation-ratio: "2"
kafka.resources:
cpu: "2"
memory: 2Gi
asserts:
- matchRegex:
path: spec.template.spec.containers[0].args[0]
pattern: "release-namespace: \\$\\{ns\\}\\s+spec:\\s+replicas: \\d+\\s+roles:\\s+- broker\\s+resources:\\s+limits:(\\s+\\S+: \\S+)+\\s+requests:(\\s+\\S+: \\S+)+(\\s+#[^\\n]*)*\\s+template:\\s+kafkaContainer:\\s+env:\\s+- name: JAVA_TOOL_OPTIONS\\s+value: \"-XX:MaxRAM=2147483648 -XX:ActiveProcessorCount=2\"\\s+storage:"
- matchRegex:
path: spec.template.spec.containers[0].args[0]
pattern: "roles:\\s+- controller\\s+resources:\\s+limits:(\\s+\\S+: \\S+)+\\s+requests:(\\s+\\S+: \\S+)+(\\s+#[^\\n]*)*\\s+template:\\s+kafkaContainer:\\s+env:\\s+- name: JAVA_TOOL_OPTIONS\\s+value: \"-XX:MaxRAM=2147483648 -XX:ActiveProcessorCount=2\"\\s+storage:"
- it: passes the entity operator memory limit to its JVMs
template: templates/kafka.yaml
asserts:
- equal:
path: spec.entityOperator.topicOperator.resources.limits.memory
value: 512Mi
- equal:
path: spec.entityOperator.template.topicOperatorContainer.env
value:
- name: JAVA_TOOL_OPTIONS
value: -XX:MaxRAM=536870912
- equal:
path: spec.entityOperator.userOperator.resources.limits.memory
value: 512Mi
- equal:
path: spec.entityOperator.template.userOperatorContainer.env
value:
- name: JAVA_TOOL_OPTIONS
value: -XX:MaxRAM=536870912
-3
View File
@@ -3,9 +3,6 @@ export NAMESPACE=cozy-$(NAME)
include ../../../hack/package.mk
test:
helm unittest .
update:
rm -rf charts
helm pull oci://quay.io/strimzi-helm/strimzi-kafka-operator --untar --untardir charts --version 1.2.0
@@ -1,31 +0,0 @@
suite: cluster operator JVM is sized from the container limits
release:
name: kafka-operator
namespace: cozy-kafka-operator
templates:
- charts/strimzi-kafka-operator/templates/060-Deployment-strimzi-cluster-operator.yaml
tests:
- it: passes the container limits to the JVM, which cannot read them on kernels without cgroup v1 controllers
asserts:
# $(VAR) expands only from entries earlier in the list.
- equal:
path: spec.template.spec.containers[0].env[19]
value:
name: CONTAINER_MEMORY_LIMIT
valueFrom:
resourceFieldRef:
resource: limits.memory
- equal:
path: spec.template.spec.containers[0].env[20]
value:
name: CONTAINER_CPU_LIMIT
valueFrom:
resourceFieldRef:
resource: limits.cpu
- equal:
path: spec.template.spec.containers[0].env[21]
value:
name: JAVA_OPTS
value: -XX:MaxRAM=$(CONTAINER_MEMORY_LIMIT) -XX:ActiveProcessorCount=$(CONTAINER_CPU_LIMIT)
- exists:
path: spec.template.spec.containers[0].resources.limits.cpu
@@ -5,19 +5,3 @@ strimzi-kafka-operator:
resources:
limits:
memory: 512Mi
# The JDK 17 in the operator image decides whether to read the container
# limits by looking at /proc/cgroups, which loses its memory line on kernels
# built without CONFIG_MEMCG_V1 (6.12+). The JVM then sizes its heap from
# host RAM and the container is OOMKilled. Drop this once the image ships a
# JDK with JDK-8347811 (17.0.20, 21.0.10).
extraEnvs:
- name: CONTAINER_MEMORY_LIMIT
valueFrom:
resourceFieldRef:
resource: limits.memory
- name: CONTAINER_CPU_LIMIT
valueFrom:
resourceFieldRef:
resource: limits.cpu
- name: JAVA_OPTS
value: -XX:MaxRAM=$(CONTAINER_MEMORY_LIMIT) -XX:ActiveProcessorCount=$(CONTAINER_CPU_LIMIT)