New to KubeDB? Please start here.

Using Prometheus (CoreOS operator) with KubeDB

This tutorial will show you how to monitor KubeDB databases using Prometheus via CoreOS Prometheus Operator.

Before You Begin

  • At first, you need to have a Kubernetes cluster, and the kubectl command-line tool must be configured to communicate with your cluster. If you do not already have a cluster, you can create one by using Minikube.

  • Now, install KubeDB cli on your workstation and KubeDB operator in your cluster following the steps here.

  • To keep things isolated, this tutorial uses a separate namespace called demo throughout this tutorial.

  $ kubectl create ns demo
  namespace "demo" created

  $ kubectl get ns
  NAME          STATUS    AGE
  demo          Active    10s

Note: The yaml files that are used in this tutorial are stored in docs/examples folder in GitHub repository kubedb/cli.

Deploy CoreOS-Prometheus Operator

Run the following command to deploy CoreOS-Prometheus operator.

$ kubectl create -f
namespace/demo created created
serviceaccount/prometheus-operator created created
deployment.extensions/prometheus-operator created

Wait for running the Deployment’s Pods.

$ kubectl get pods -n demo --watch
NAME                                   READY     STATUS              RESTARTS   AGE
prometheus-operator-857455484c-dg4qg   0/1       ContainerCreating   0          34s
prometheus-operator-857455484c-dg4qg   1/1       Running             0         45s

This CoreOS-Prometheus operator will create some supported Custom Resource Definition (CRD).

$ kubectl get crd
NAME                                          CREATED AT
...           2018-09-24T12:42:22Z            2018-09-24T12:42:22Z         2018-09-24T12:42:22Z

Once the Prometheus operator CRDs are registered, run the following command to create a Prometheus.

$ kubectl create -f created
serviceaccount/prometheus created created created
service/prometheus created

# Verify RBAC stuffs
$ kubectl get clusterroles
NAME                  AGE
prometheus            48s
prometheus-operator   1m

$ kubectl get clusterrolebindings
NAME                  AGE
prometheus            7s
prometheus-operator   25s

$ kubectl get serviceaccounts -n demo
NAME                  SECRETS   AGE
default               1         3m
prometheus            1         1m
prometheus-operator   1         3m

Prometheus Dashboard

Now to open prometheus dashboard on Browser:

$ kubectl get svc -n demo
NAME                  TYPE           CLUSTER-IP       EXTERNAL-IP   PORT(S)          AGE
prometheus            LoadBalancer   <pending>     9090:30900/TCP   1m
prometheus-operated   ClusterIP      None             <none>        9090/TCP         1m

$ minikube ip

$ minikube service prometheus -n demo --url

Now, open your browser and go to the following URL: http://{minikube-ip}:{prometheus-svc-nodeport} to visit Prometheus Dashboard. According to the above example, this URL will be

If you are not using minikube, browse prometheus dashboard using following address http://{Node's ExternalIP}:{NodePort of prometheus-service}.

Find out required label for ServiceMonitor

First, check created objects of Prometheus kind.

$ kubectl get prometheus --all-namespaces
demo        prometheus   20m

Now if we see the full spec of prometheus of Prometheus kind, we will see a field called serviceMonitorSelector. The value of matchLabels under serviceMonitorSelector part, is the required label for KubeDB monitoring spec monitor.prometheus.labels.

 $ kubectl get prometheus -n demo prometheus -o yaml
kind: Prometheus
  creationTimestamp: 2018-11-15T10:40:57Z
  generation: 1
  name: prometheus
  namespace: demo
  resourceVersion: "1661"
  selfLink: /apis/
  uid: ef59e6e6-e8c2-11e8-8e44-08002771fd7b
      memory: 400Mi
  serviceAccountName: prometheus
      app: kubedb
  version: v1.7.0

In this tutorial, the required label is app: kubedb.

Monitor MySQL with CoreOS Prometheus

KubeDB implements a MySQL CRD to define the specification of a MySQL database. Below is the MySQL object created in this tutorial.

kind: MySQL
  name: mysql-mon-coreos
  namespace: demo
  version: "8.0-v1"
    storageClassName: "standard"
    - ReadWriteOnce
        storage: 50Mi
      namespace: demo
        app: kubedb
      interval: 10s

The MySQL CRD object contains monitor field in it’s spec. It is also possible to add CoreOS-Prometheus monitor to an existing MySQL database by adding the below part in it’s spec field.

Here, spec.monitor.prometheus.labels is the serviceMonitorSelector that we found earlier.

      namespace: demo
        app: kubedb
      interval: 10s
Keys Value Description
spec.monitor.agent string Required. Indicates the monitoring agent used. Only valid value currently is coreos-prometheus-operator
spec.monitor.prometheus.namespace string Required. Indicates namespace where service monitors are created. This must be the same namespace of the Prometheus instance.
spec.monitor.prometheus.labels map Required. Indicates labels applied to service monitor.
spec.monitor.prometheus.interval string Optional. Indicates the scrape interval for database exporter endpoint (eg, ‘10s’)
spec.monitor.prometheus.port int Optional. Indicates the port for database exporter endpoint (default is 56790)

Known Limitations: If the database password is updated, exporter must be restarted to use the new credentials. This issue is tracked here.

Run the following command to deploy the above MySQL CRD object.

$ kubedb create -f created


  • spec.monitor specifies that CoreOS Prometheus operator is used to monitor this database instance. A ServiceMonitor should be created in the demo namespace with label app=kubedb. The exporter endpoint should be scrapped every 10 seconds.

KubeDB will create a separate stats service with name <mysql-crd-name>-stats for monitoring purpose. KubeDB operator will configure this monitoring service once the MySQL is successfully running.

$ kubedb get my -n demo
NAME               VERSION   STATUS     AGE
mysql-mon-coreos   8.0-v1    Creating   22s

$ kubedb describe my -n demo mysql-mon-coreos
Name:               mysql-mon-coreos
Namespace:          demo
CreationTimestamp:  Thu, 27 Sep 2018 16:29:36 +0600
Labels:             <none>
Annotations:        <none>
Replicas:           1  total
Status:             Running
  StorageType:      Durable
  StorageClass:  standard
  Capacity:      50Mi
  Access Modes:  RWO

  Name:               mysql-mon-coreos
  CreationTimestamp:  Thu, 27 Sep 2018 16:29:39 +0600
  Annotations:        <none>
  Replicas:           824640215820 desired | 1 total
  Pods Status:        1 Running / 0 Waiting / 0 Succeeded / 0 Failed

  Name:         mysql-mon-coreos
  Annotations:  <none>
  Type:         ClusterIP
  Port:         db  3306/TCP
  TargetPort:   db/TCP

  Name:         mysql-mon-coreos-stats
  Type:         ClusterIP
  Port:         prom-http  56790/TCP
  TargetPort:   prom-http/TCP

Database Secret:
  Name:         mysql-mon-coreos-auth
  Annotations:  <none>
Type:  Opaque
  password:  16 bytes
  user:      4 bytes

Monitoring System:
    Port:       56790
    Namespace:  demo
    Labels:     app=kubedb
    Interval:   10s

No Snapshots.

  Type    Reason      Age   From            Message
  ----    ------      ----  ----            -------
  Normal  Successful  1m    MySQL operator  Successfully created Service
  Normal  Successful  1m    MySQL operator  Successfully created StatefulSet
  Normal  Successful  1m    MySQL operator  Successfully created MySQL
  Normal  Successful  56s   MySQL operator  Successfully created stats service
  Normal  Successful  52s   MySQL operator  Successfully patched StatefulSet
  Normal  Successful  52s   MySQL operator  Successfully patched MySQL
  Normal  Successful  51s   MySQL operator  Successfully patched StatefulSet
  Normal  Successful  51s   MySQL operator  Successfully patched MySQL

Since spec.monitoring was configured, a ServiceMonitor object is created accordingly. You can verify it running the following commands:

$ kubectl get servicemonitor -n demo
NAME                           AGE
kubedb-demo-mysql-mon-coreos   1m

$ kubectl get servicemonitor -n demo kubedb-demo-mysql-mon-coreos -o yaml
kind: ServiceMonitor
  creationTimestamp: 2018-09-27T10:30:20Z
  generation: 1
    app: kubedb mysql-mon-coreos-stats.demo
  name: kubedb-demo-mysql-mon-coreos
  namespace: demo
  resourceVersion: "6257"
  selfLink: /apis/
  uid: 55a3ae53-c240-11e8-b2cc-080027d9f35e
  - interval: 10s
    path: /metrics
    port: prom-http
    targetPort: 0
    - demo
    matchLabels: MySQL mysql-mon-coreos

Now, if you go the Prometheus Dashboard, you should see that this database endpoint as one of the targets. prometheus-coreos

Cleaning up

To cleanup the Kubernetes resources created by this tutorial, run:

kubectl patch -n demo mysql/mysql-mon-coreos -p '{"spec":{"terminationPolicy":"WipeOut"}}' --type="merge"
kubectl delete -n demo mysql/mysql-mon-coreos

kubectl patch -n demo drmn/mysql-mon-coreos -p '{"spec":{"wipeOut":true}}' --type="merge"
kubectl delete -n demo drmn/mysql-mon-coreos

kubectl delete -f
kubectl delete -f

kubectl delete ns demo

Next Steps

Subscribe to our free technical newsletter!

Join thousands of subscribers and stay up-to-date on AppsCode.