Kubernetes Pod Disruption Budget (PDB) Simplified

By /

Listen to this Post

Featured Image
Kubernetes Pod Disruption Budget (PDB) ensures high availability during voluntary disruptions like node maintenance or cluster upgrades. Below is a simplified breakdown of how PDB works:

Typical Kubernetes PDB Flow:

  1. Deployment Creation: A Deployment with 3 replicas and label `app=my-app` is created.
  2. Admin Action: Admin applies Deployment and PDB YAML.
  3. API Server: Receives the spec and stores it in etcd.

4. etcd: Stores Deployment and PDB objects.

  1. Controller Manager: Validates, stores, and creates a ReplicaSet.
  2. ReplicaSet: Creates 3 Pods based on the spec.

7. Scheduler: Assigns Pods to nodes.

8. Kubelet:

  • Node1 runs Pod-1 and Pod-2.
  • Node2 runs Pod-3.

9. PDB Enforcement:

– `minAvailable: 2` monitors Pods with matching labels.
– Admin drains Node1 (simulating disruption).
– Eviction request for Pod-2 is denied (violates availability).
– Eviction of Pod-1 is approved (still meets minAvailable).

📖 Full Blog: Kubernetes PDB Explained

You Should Know:

1. Creating a PDB YAML

apiVersion: policy/v1
kind: PodDisruptionBudget
metadata:
name: my-app-pdb
spec:
minAvailable: 2
selector:
matchLabels:
app: my-app

Apply using:

kubectl apply -f pdb.yaml

2. Checking PDB Status

kubectl get pdb
kubectl describe pdb my-app-pdb

3. Testing PDB Enforcement

Simulate node drain:

kubectl drain <node-name> --ignore-daemonsets

Check if Pods are evicted while maintaining availability.

4. Common Troubleshooting Commands

  • Check Pod Labels: 
    kubectl get pods --show-labels
  • Force Eviction (if stuck): 
    kubectl delete pod <pod-name> --grace-period=0 --force
  • Check Cluster Events: 
    kubectl get events --sort-by='.metadata.creationTimestamp'

5. Alternative: maxUnavailable

apiVersion: policy/v1
kind: PodDisruptionBudget
metadata:
name: my-app-pdb
spec:
maxUnavailable: 1
selector:
matchLabels:
app: my-app

What Undercode Say

PDBs are crucial for production-grade Kubernetes clusters. Misconfigurations can lead to upgrade failures or unexpected downtimes. Always:
– Test PDBs in a staging environment.
– Use `kubectl drain –dry-run` before actual node maintenance.
– Monitor with `kubectl get pdb -w` for real-time updates.

🔧 Pro Tip: Combine PDBs with `podAntiAffinity` to ensure Pods are distributed across nodes for higher resilience.

📚 Further Learning:

Expected Output:

NAME MIN AVAILABLE MAX UNAVAILABLE ALLOWED DISRUPTIONS AGE
my-app-pdb 2 N/A 1 5m

This ensures your Kubernetes workloads remain available during disruptions. 🚀

References:

Reported By: Sandip Das – Hackers Feeds
Extra Hub: Undercode MoN
Basic Verification: Pass ✅

Join Our Cyber World:

💬 Whatsapp | 💬 Telegram

Related Posts: