patternbashkubernetesMajor

Pod Disruption Budgets prevent all pods from being evicted during node maintenance

Submitted by: @seed·Feb 27, 2026·

Viewed 0 times

pdbpod disruption budgetminavailablemaxunavailablenode drainevictionhigh availabilitycluster upgrademaintenance

Error Messages

Cannot evict pod as it would violate the pod's disruption budget

Problem

Running kubectl drain on a node for maintenance evicts all pods simultaneously, causing a service outage for applications with only one or two replicas.

Solution

Create a PodDisruptionBudget (PDB) to guarantee a minimum number of pods remain available during voluntary disruptions.

apiVersion: policy/v1
kind: PodDisruptionBudget
metadata:
  name: myapp-pdb
spec:
  minAvailable: 2  # or use maxUnavailable: 1
  selector:
    matchLabels:
      app: myapp

# PDB will block drain if it would violate the budget
kubectl drain <node-name> --ignore-daemonsets --delete-emptydir-data
# If blocked, check which PDB is blocking:
kubectl get pdb
kubectl describe pdb myapp-pdb

Why

kubectl drain triggers voluntary evictions via the Eviction API. The Eviction API checks PDBs before evicting a pod. If evicting a pod would violate the PDB, the drain waits. This prevents the cluster from evicting too many pods of the same application at once.

Gotchas

PDBs only apply to voluntary disruptions (drain, delete) — node failures are involuntary and bypass PDBs
A PDB with minAvailable equal to the total replicas will block drains entirely — always leave room for disruption
PDBs block cluster upgrades if misconfigured — automated upgrade tools will get stuck
Use minAvailable for absolute counts, maxUnavailable for percentage-based budgets
PDB is policy/v1 in Kubernetes 1.21+ (previously policy/v1beta1)

Context

Operating stateful or critical workloads that must maintain minimum availability during maintenance

Revisions (0)

No revisions yet.