Skip to content

Symptoms

  • Running kubectl drain node-5.internal fails immediately with an error referencing DaemonSet-managed pods.
  • The error message lists pods from kube-system namespace: fluentd-ds-xr4t2, node-exporter-7hk9p.
  • The engineer expected the drain to gracefully move all workloads off the node.
  • The node is being decommissioned as part of a cluster scale-down.
  • No --ignore-daemonsets flag was used in the drain command.
  • The cluster runs Kubernetes 1.27 with Calico CNI and a logging stack using Fluentd as a DaemonSet.