Skip to content

Insufficient replicas: single point of failure

Quick summary

Your service has only one replica — if it fails, the entire feature or application becomes unavailable.

Why this happens

  • The deployment is configured with replicas: 1 for cost savings

  • The service was deployed for development and not scaled for production

  • Auto-scaling is not configured and the initial replica count is 1

Minimal example

✗ Broken code
apiVersion: apps/v1
kind: Deployment
spec:
  replicas: 1 # single point of failure
  template:
    spec:
      containers:
        - name: app
          image: myapp:v1
✓ Fixed code
apiVersion: apps/v1
kind: Deployment
spec:
  replicas: 3 # multiple replicas for HA
  template:
    spec:
      containers:
        - name: app
          image: myapp:v1
---
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
spec:
  scaleTargetRef:
    kind: Deployment
    name: app
  minReplicas: 3
  maxReplicas: 10

Running 3 replicas and configuring auto-scaling ensures the service remains available even if one pod fails and can handle load spikes.

How to diagnose

  • Check the deployment replica count with kubectl get deployment or in the manifest

  • Review the application architecture for single points of failure

  • Verify that auto-scaling is configured if the load varies

How to fix

  • Increase the replica count to at least 2 for high availability

  • Configure horizontal pod auto-scaling based on CPU or custom metrics

  • Distribute replicas across multiple nodes or availability zones for fault tolerance

How to prevent

  • Always run at least 2 replicas in production for stateless services

  • Configure auto-scaling and spread replicas across availability zones for resilience

Related Resources

Related Glossary

Related Lessons

← Back to language