Disruptions and PodDisruptionBudgets

Limit voluntary disruptions

Nodes are drained for upgrades, scaling down and maintenance (voluntary disruptions). A PodDisruptionBudget (PDB) limits how many pods of an app may be taken down at once (minAvailable or maxUnavailable), so drains wait rather than removing too many replicas. Combine with multiple replicas, topology spread constraints or anti-affinity across nodes and zones, and graceful shutdown (handle SIGTERM, a preStop hook, a sensible terminationGracePeriodSeconds).

A PDB and zone spreading

Not applied to a live cluster in this course; check field names against the API reference for your version.

apiVersion: policy/v1
kind: PodDisruptionBudget
metadata: {name: web}
spec:
  maxUnavailable: 1
  selector: {matchLabels: {app: web}}
---
# in the Deployment pod template
topologySpreadConstraints:
- maxSkew: 1
  topologyKey: topology.kubernetes.io/zone
  whenUnsatisfiable: ScheduleAnyway
  labelSelector: {matchLabels: {app: web}}

Handle SIGTERM

Stop accepting new requests, finish in-flight work and exit within the grace period; otherwise every rollout drops requests.

त्वरित जाँच: What does a PodDisruptionBudget protect against?

• High CPU usage
• Node hardware failures
• Bad code
• Too many pods of an app being evicted at once during voluntary disruptions