PostgreSQL Kubernetes Operator

🚀 Overview

A custom Kubernetes Controller written in Go that automates the lifecycle management of PostgreSQL database clusters.

Unlike standard deployments, this Operator acts as a "Site Reliability Engineer in a box." It continuously watches the cluster state and enforces the desired configuration, offering Self-Healing capabilities that automatically recover deleted pods in under 5 seconds.

⚡ Key Features

• Custom Resource Definition (CRD): Extends the Kubernetes API with a new SimpleDB kind, allowing developers to request databases using native YAML.
• Active Reconciliation: Implements a control loop that constantly compares the "Desired State" (YAML) vs "Actual State" (Cluster).
• Self-Healing Infrastructure: Automatically detects if a database Pod is deleted or crashes and recreates it instantly.
• Elastic Scaling: Supports real-time horizontal scaling via kubectl patch or YAML updates.
• Status Reporting: Provides real-time feedback on the database health status directly in the terminal.

🏗 Architecture

The Operator follows the standard Kubernetes Controller pattern:

1. User applies a SimpleDB YAML.
2. API Server stores the request.
3. Controller (My Code) detects the new event.
4. Reconciler Loop creates and manages a Kubernetes Deployment for Postgres.
5. Observer watches the Pods; if one dies, the Reconciler spins up a replacement.

🏃‍♂️ How to Run

Prerequisites

• Go 1.22+
• Docker Desktop & Kind (Kubernetes in Docker)
• kubectl

1. Installation

Clone the repo and install the Custom Resource Definitions (CRDs) into your cluster:

make install

2. Start the Controller

Run the operator locally (it will connect to your current K8s context):

make run

3. Deploy a Database

Open a new terminal and apply the sample database configuration:

kubectl apply -f config/samples/database_v1_simpledb.yaml

4. Verify & Test Self-Healing

Check the status of your database:

kubectl get simpledbs
# Output: NAME              STATUS   AGE
#         simpledb-sample   Ready    2m

🔥 The "Chaos Monkey" Test: Delete the pod manually to test resilience:

kubectl delete pod -l app=simpledb
kubectl get pods -w
# Result: You will see a new pod spin up instantly!

5. Test Scaling

Scale your database from 1 replica to 3 replicas on the fly:

kubectl patch simpledb simpledb-sample --type='merge' -p '{"spec":{"replicas":3}}'
kubectl get pods
# Result: 3 Pods running

🛠 Tech Stack

• Language: Go (Golang)
• Framework: Kubebuilder, Controller-Runtime
• Orchestration: Kubernetes (Kind)

---

Built by Hemraj