upvote
You can get into difficulty with kubernetes here, as your jvm will detect all cores on the node but you may have set a resources limit on the pod/whatever, so it’ll assume it can spend more time doing stuff than it actually can, so often times it’s quite necessary to tune some things to prevent excessive switching etc.
reply
Modern JVMs will detect orchestrator-set cgroup limits and size themselves accordingly. If you, for example, set a cpu limit for a pod to “1”, the JVM will size itself as if it was running on a single core machine.
reply