Kubernetes & Postgres Replication Setup

I’m architecting a stack that combines PostgreSQL streaming replication, Dockerised services, and a full CI/CD pipeline, but where I need the sharpest guidance is on the Kubernetes layer. The goal is to move from a proof-of-concept into a production-grade cluster that cleanly orchestrates our containers, handles fail-over for the replicated databases, and slots into an automated build-and-deploy workflow.

Here’s the current landscape
• PostgreSQL primary/replica images are already containerised and tested locally.
• CI/CD runs in GitLab; it builds images, pushes to a private registry, and triggers deployments.
• A bare-bones Kubernetes cluster (3 worker nodes) is online but only running sample apps.

What I need from you
• Design or refine the Helm charts (or Kustomize if you prefer) that will deploy our replicated Postgres set, application pods, and supporting services.
• Implement readiness/liveness probes, PodDisruptionBudgets, and sensible resource limits so the cluster behaves well under failure or scaling events.
• Wire the GitLab pipeline into Kubernetes with zero-downtime rollouts and automatic rollback on failed health checks.
• Document the solution clearly enough that another engineer can reproduce it on a new environment.

Acceptance criteria
1. kubectl get pods shows healthy, version-tagged workloads for primary and replica databases, plus app pods.
2. Simulated node loss triggers automatic fail-over, confirmed by test queries.
3. git push to main triggers a pipeline that finishes with a successful, rolling deployment.
4. A short run-book (Markdown) covers cluster bootstrapping, secrets handling, and recovery steps.

If you’ve tackled similar end-to-end Kubernetes builds, especially ones featuring Postgres replication, I’d like to see a concise example or repo link when you reply. Let’s make this cluster bulletproof.