kubernetes-manifest

Define a Deployment with resource requests and limits, liveness and readiness probes, and a RollingUpdate strategy with maxUnavailable set to 0 for zero downtime. Add a ClusterIP Service, Ingress, and HPA to complete the production setup.

Liveness probes detect deadlocked or broken processes and trigger pod restarts, while readiness probes determine whether a pod can receive traffic and remove it from the service endpoints when failing. Slow-starting apps should also use a startup probe.

Use a StatefulSet for stateful workloads like databases that need stable network identities and persistent storage via volumeClaimTemplates. Deployments are for stateless apps, since pods lose data on restart without persistent volumes.

Create a HorizontalPodAutoscaler targeting your Deployment with minReplicas, maxReplicas, and CPU or memory utilization targets such as 70%. Configure scale-down stabilization windows and scaling policies to prevent flapping.

Requests let the scheduler place pods on nodes with sufficient capacity, while limits cap usage to prevent one pod from starving others. Set limits at 1.5-2x requests to avoid OOM kills under normal load spikes.

Plain Secrets should never be committed to Git because values are only base64-encoded, not encrypted. Use sealed-secrets, external-secrets, or Vault to manage sensitive values safely in version control.