Increasing the Strimzi Cluster Operator's memory
If you want a single installation of the Strimzi Cluster Operator to watch and manage more than 20 Kafka clusters, you must increase the memory and heap allocated for the Strimzi Cluster Operator. Otherwise, you will encounter out of memory and heap related errors. To do this, you configure memory-related properties.
-XX:MinRAMPercentage
and
-XX:MaxRAMPercentage
Java parameters. Additionally, configure
resources.limits.memory
and
resources.requests.memory
Helm properties.The following example contains memory settings recommended by Cloudera for a deployment with more than 20 Kafka clusters. You can fine-tune your setting as needed.
helm upgrade [***RELEASE***] [***CHART***] \
--namespace [***NAMESPACE***] \
--set 'extraEnvs[0].name=JAVA_OPTS' \
--set 'extraEnvs[0].value=-XX:MinRAMPercentage=25 -XX:MaxRAMPercentage=70' \
--set 'resources.limits.memory=600Mi' \
--set 'resources.requests.memory=600Mi' \
--reuse-values
The default values for the properties configured in
the example are as follows.
-XX:MinRAMPercentage=15
-XX:MaxRAMPercentage=20
resources.limits.memory=384Mi
resources.requests.memory=384Mi