Kubernetes and GitOps
Make deployments predictable. Remove drift. Control infrastructure through Git.
Overview
Kubernetes without discipline becomes complex very quickly.
Typical issues:
- configuration drift between environments
- manual changes directly in cluster
- unclear deployment state
- broken rollbacks
- too many tools layered without control
GitOps is not about adding ArgoCD.
It is about making the system observable, reproducible, and controlled.
This service focuses on stabilizing Kubernetes and introducing Git as the single source of truth.
Related use cases
Centralizing metrics and traces from remote retail sites with vmagent and OpenTelemetry
How we collected infrastructure metrics, network probes and application traces from remote POS and warehouse systems into a central Kubernetes observability stack.
Kubernetes application uses a public DNS name for an internal API: how to keep traffic inside the cluster
A Kubernetes application became intermittently slow because internal API requests were leaving the cluster through a public DNS name. How we traced the latency to external network RTT and kept the traffic inside Kubernetes.
RabbitMQ cluster recovery after a node outage: how insufficient monitoring resulted in a split-brain cluster
A RabbitMQ Streams incident where some writes succeeded while others failed with coordinator unavailable. How we identified the authoritative Raft state, recovered the cluster safely, and fixed the monitoring gap that allowed the split-brain condition to go unnoticed.