New to Kubed? Please start here.

Monitoring Kubed

Kubed operator exposes Prometheus ready metrics via the following endpoints on port :8443:

• /metrics: Scrape this to monitor operator.

Follow the steps below to view the metrics:

1. Give system:anonymous user access to /metrics url. This is not safe to do on a production cluster.

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: appscode:system:metrics-collector
rules:
- nonResourceURLs: ["/metrics"]
  verbs: ["get"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: appscode:system:metrics-collector
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: appscode:system:metrics-collector
subjects:
- apiGroup: rbac.authorization.k8s.io
  kind: User
  name: system:anonymous

$ kubectl auth reconcile -f docs/examples/monitoring/metrics-collector.yaml
clusterrole.rbac.authorization.k8s.io "appscode:system:metrics-collector" reconciled
clusterrolebinding.rbac.authorization.k8s.io "appscode:system:metrics-collector" reconciled

2. Now, forward the port :8443 to your workstation.

$ kubectl get pods -n kube-system | grep voyager
voyager-operator-f89dcccdb-plvmt        1/1       Running   0          27m

$ kubectl port-forward -n kube-system voyager-operator-f89dcccdb-plvmt 8443
Forwarding from 127.0.0.1:8443 -> 8443
Forwarding from [::1]:8443 -> 8443

3. Now, visit the url: https://127.0.0.1:8443/metrics

4. Once you are done, remove access to system:anonymous user.

$ kubectl delete -f docs/examples/monitoring/metrics-collector.yaml
clusterrole.rbac.authorization.k8s.io "appscode:system:metrics-collector" deleted
clusterrolebinding.rbac.authorization.k8s.io "appscode:system:metrics-collector" deleted

Next Steps

• Learn how to use Kubed to protect your Kubernetes cluster from disasters here.
• Need to keep configmaps/secrets synchronized across namespaces or clusters? Try Kubed config syncer.
• Want to keep an eye on your cluster with automated notifications? Setup Kubed event forwarder.
• Out of disk space because of too much logs in Elasticsearch or metrics in InfluxDB? Configure janitors to delete old data.
• Wondering what features are coming next? Please visit here.
• Want to hack on Kubed? Check our contribution guidelines.