Skip to content

Prometheus node-exporter and operator are on a CrashLoopBackOff due to OOM

We are using Kubernetes v1.22.9 for the swan service, in the kube-system namespace we see pods below crashing always

  • cern-magnum-kube-prometheu-operator has a memory limit of 64M and is OOMKilled
  • cern-magnum-prometheus-node-exporter has a memory limit of only 20M and are OOMKilled periodically

(This doesn't seem to impact metrics though.)