// the find
argoproj-labs/argocd-operator
A Kubernetes operator for managing Argo CD clusters.
A Kubernetes operator that manages the full lifecycle of Argo CD installations via a custom `ArgoCD` CRD — installs, configures, and keeps Argo CD reconciled without you hand-managing deployments and ConfigMaps. Primarily aimed at OpenShift/OLM environments where the Red Hat GitOps Operator ships it, but it works on vanilla Kubernetes too. If you're running Argo CD at scale across multiple clusters, this is the right abstraction layer.
The CRD surface covers nearly everything you'd otherwise manage by hand: SSO (Dex), Redis HA with Sentinel, ingress/route configuration, notifications, image updater, and RBAC — all as typed spec fields with validation webhooks. The v1beta1 API is a real improvement over v1alpha1, with conversion webhooks so you're not stuck on the old schema. Test coverage is thorough — almost every controller file has a corresponding `_test.go` that exercises reconcile paths. The new `argocdagent` controller subtree is interesting: it's building toward a principal/agent topology for managing remote clusters, which is the right answer for large multi-cluster setups.
The README is nearly empty — a Travis CI badge pointing at a dead URL and a link to readthedocs. The actual documentation is external and incomplete in places; figuring out upgrade paths between operator versions requires archaeology through the OLM catalog directories. The `ArgoCD` CRD has accumulated so many fields that the `argocd_types.go` file is a beast to read, and the split between v1alpha1 and v1beta1 is not fully cleaned up — conversion logic adds surface area for subtle bugs. The `argocdagent` controller is clearly experimental and undocumented beyond a single quickstart markdown, so don't reach for it in production yet.