ChaosRent GitOps — Self-Healing Kubernetes Platform
Problem
Demonstrates automated deployment, configuration management, drift correction, and failure recovery instead of relying on manual production changes.
What was built
A self-healing, GitOps-driven Kubernetes platform deployed and tested in a self-managed homelab. Git acts as the single source of truth, Argo CD continuously synchronizes the cluster, and Kubernetes handles self-healing.
Key outcomes
- Automated synchronization between Git and cluster state
- Declarative deployments
- Environment-specific configuration with Kustomize
- Automated rollback and reconciliation
- Chaos testing — deleted pods automatically recovered
Architecture
Tools & Technologies
- K3s
- Kubernetes
- Argo CD
- Kustomize
- Git / GitHub
- Linux
- kubectl
- YAML
- GitOps
GitOps reconciliation
Argo CD continuously compares the declared state in Git with live cluster state and reconciles drift automatically, so the repository — not a terminal session — defines the environment.
Declarative configuration
Manifests are structured with Kustomize bases and overlays so environment-specific values stay separate from shared configuration.
Chaos testing & recovery
Pods and workloads were deliberately deleted to observe how Kubernetes controllers and Argo CD restore the desired state, and how long reconciliation takes.
Troubleshooting & lessons learned
Most of the learning came from failures: sync errors, misconfigured overlays, and resource issues on a small K3s cluster. Reading controller behaviour and events became more useful than reapplying manifests.