Kubernetes in Production: Best Practices and Common Pitfalls to Avoid

Kubernetes has become the de facto standard for container orchestration, but running it successfully in production requires careful planning and adherence to best practices.

Resource Management

Right-Sizing Your Containers

One of the most common mistakes is not properly setting resource requests and limits:

resources:
  requests:
    memory: "256Mi"
    cpu: "250m"
  limits:
    memory: "512Mi"
    cpu: "500m"

Horizontal Pod Autoscaling

Implement HPA to automatically scale based on CPU/memory or custom metrics:

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: myapp-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: myapp
  minReplicas: 3
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 70

Security Hardening

Network Policies

Implement network policies to control traffic between pods:

• Default deny all traffic
• Explicitly allow only necessary connections
• Separate namespaces for different security zones

Pod Security Standards

Use Pod Security Standards (PSS) to enforce security policies:

• **Privileged**: Unrestricted (avoid in production)
• **Baseline**: Minimally restrictive
• **Restricted**: Heavily restricted (recommended)

RBAC Best Practices

• Use least privilege access
• Create service accounts for each application
• Regularly audit RBAC permissions
• Avoid using cluster-admin role

High Availability

Multi-Zone Deployments

Distribute workloads across availability zones:

affinity:
  podAntiAffinity:
    preferredDuringSchedulingIgnoredDuringExecution:
    - weight: 100
      podAffinityTerm:
        topologyKey: topology.kubernetes.io/zone

Health Checks

Always implement proper liveness and readiness probes:

livenessProbe:
  httpGet:
    path: /healthz
    port: 8080
  initialDelaySeconds: 30
  periodSeconds: 10

readinessProbe:
  httpGet:
    path: /ready
    Port: 8080
  initialDelaySeconds: 5
  periodSeconds: 5

Monitoring and Observability

Essential monitoring stack:

• **Prometheus**: Metrics collection
• **Grafana**: Visualization
• **Jaeger/Tempo**: Distributed tracing
• **Loki**: Log aggregation

Common Pitfalls to Avoid

1. **Not Setting Resource Limits**: Can lead to resource exhaustion
2. **Ignoring Security**: Running containers as root
3. **Poor Logging**: Not collecting logs centrally
4. **No Backup Strategy**: Always backup etcd and persistent volumes
5. **Inadequate Testing**: Test upgrades in staging first

Conclusion

Running Kubernetes in production is a journey, not a destination. Regular updates, monitoring, and continuous improvement are essential for success.

Ready to optimize your Kubernetes infrastructure? Contact our DevOps team for a consultation.