Cluster Discovery

How the ClusterPersona controller discovers and reports cluster state

The ClusterPersona controller runs a discovery loop every 5 minutes to build a comprehensive picture of your cluster’s nodes, resources, add-ons, and platform type.

What gets discovered

Node discovery

The controller lists all nodes and extracts:

Field	Source
Name	node.metadata.name
Role	node-role.kubernetes.io/control-plane or master label
Ready	Ready condition in node.status.conditions
Capacity	CPU, memory, pods, ephemeral storage from node.status.capacity
Allocatable	CPU, memory, pods, ephemeral storage from node.status.allocatable
Kubelet version	node.status.nodeInfo.kubeletVersion
Container runtime	node.status.nodeInfo.containerRuntimeVersion
Taints	node.spec.taints formatted as key=value:effect
Labels	Filtered to interesting prefixes: node.kubernetes.io/, topology.kubernetes.io/, kubernetes.io/arch, kubernetes.io/os, node-role.kubernetes.io/

Resource summary

The controller aggregates resource information across all nodes:

Field	Calculation
totalCPU	Sum of all node CPU capacity
totalMemory	Sum of all node memory capacity
allocatableCPU	Sum of all node allocatable CPU
allocatableMemory	Sum of all node allocatable memory
totalPods	Sum of all node pod capacity
runningPods	Count of pods in Running phase across all namespaces
nodeCount	Total number of nodes

Add-on detection

The controller probes well-known namespaces for installed add-ons by searching for pods that contain the add-on name.

Add-on	Namespace checked	Pod name pattern	Type
ArgoCD	argocd	argocd	gitops
Prometheus	monitoring	prometheus or prometheus-server	monitoring
Grafana	monitoring	grafana	monitoring
cert-manager	cert-manager	cert-manager	cert-management
ingress-nginx	ingress-nginx	ingress-nginx-controller	ingress
external-secrets	external-secrets	external-secrets	secrets
Istio	istio-system	istiod	service-mesh
CloudNativePG	cnpg-system	cnpg-cloudnative-pg	database
OpenObserve	openobserve	openobserve	monitoring

For each detected add-on, the controller records:

• Whether it is installed
• The version (extracted from the container image tag)
• The namespace
• Whether the pod is healthy (in Running phase)

Platform detection

The controller identifies the Kubernetes platform using two methods:

Method 1: Node labels

Label contains	Platform
eks.amazonaws.com	EKS
cloud.google.com/gke	GKE
kubernetes.azure.com	AKS
node.openshift.io	OpenShift
minikube.k8s.io	Minikube
kind.x-k8s.io	Kind
k3s.io	K3s

Method 2: Provider ID (fallback)

Provider ID prefix	Platform
aws://	EKS
gce://	GKE
azure://	AKS
kind://	Kind

If neither method matches, the platform is reported as Generic.

Namespace summary

Field	Description
total	Total number of namespaces
active	Namespaces in Active phase
withPersonas	Namespaces containing at least one ApplicationPersona

Phase determination

Condition	Phase
No nodes discovered	Discovering
All nodes ready	Ready
Some nodes not ready	Degraded

When a transient API failure prevents node discovery, the controller preserves the previously discovered nodes instead of clearing them. This prevents phase regression from Ready to Discovering on temporary network issues.

Conditions

Condition	Status	When
Discovered	True	Discovery completed (with or without issues)
Discovered	False	Node discovery failed
Healthy	True	All nodes ready
Healthy	False	Some nodes not ready

Checking cluster status

dorgu cluster status

Or query the resource directly:

kubectl get clusterpersona my-cluster -o yaml

Cluster onboarding

Bootstrap your cluster with dorgu cluster init

Add-on detection

ArgoCD integration details