Worker nodes over-committed and causing nodes to restart or go "NotReady"

Issue

• One or more worker nodes restart unexpectedly, become NotReady, or are replaced.

• This issue is commonly observed with one or more of the following symptoms:

  • Worker nodes restart without an obvious reason

  • Worker nodes appear as NotReady

  • Nodes are replaced automatically after becoming unhealthy

  • oc describe node shows the worker node as overcommitted

  • Pods on the affected node restart, are evicted, or become unstable

• Why are worker nodes restarting and transitioning to NotReady?

• Why does oc describe node show the node as overcommitted?

• Can node overcommitment lead to pod evictions, OOMKilled events, or node instability?

• Why are nodes automatically replaced after becoming unhealthy?