// Deployment-Referenz
Running a spoke-agent with plain manifests
An agent monitors one location and runs wherever that location is — a homelab, a Proxmox host, another Kubernetes cluster — usually not in the hub cluster. It reaches the hub over a single URL (HTTPS bootstrap, then the NATS WebSocket), so it needs only the hub URL and a one-time registration token.
Steps
- In the hub UI, Locations → Onboard to get a registration token.
- Create the token Secret (see
fleeger-agent-token.example.yaml). Paste the token at the prompt: it is not echoed and stays off the command line:printf 'Registration token: '; stty -echo; IFS= read -r FLEEGER_TOKEN; stty echo; echo printf '%s' "$FLEEGER_TOKEN" | kubectl create secret generic fleeger-agent-token \ --from-file=registration-token=/dev/stdin unset FLEEGER_TOKEN - Edit
fleeger-agent.yaml: setFLEEGER_HUB_URLto your hub (e.g.https://fleeger.example.com) and, if pulling from a registry, the image. - Apply:
kubectl apply -f fleeger-agent.yaml
The agent bootstraps on first start, persists its NATS creds to the PVC, and reconnects across restarts without re-onboarding. To onboard it again (after offboarding, or a credential rotation), see re-onboarding.
Monitoring the cluster the agent runs in
If this agent has a k8s_cluster resource with an empty credential, it
collects its own cluster via its ServiceAccount. Grant it read-only access:
# edit the ClusterRoleBinding namespace in agent-rbac.yaml to match where you
# deployed the agent, then:
kubectl apply -f agent-rbac.yamlProxmox / iDRAC / SSH-only agents don't need agent-rbac.yaml.
Monitoring the hub's own cluster? Don't run a separate agent — use the bundled core-agent in Monitoring the hub's own cluster (bundled core-agent).