Running Metricbeat on Kubernetes

Metricbeat Docker images can be used on Kubernetes to retrieve cluster metrics.

Metricbeat is deployed in two different ways at the same time:

By deploying Metricbeat as a DaemonSet we ensure we get a running instance on each node of the cluster. It will be used to retrieve most metrics from the host, like system metrics, Docker stats and metrics from all the services running on top of Kubernetes.

A single Metricbeat instance is also created using a Deployment. It will retrieve metrics that are unique for the whole cluster, like Kubernetes events or kube-state-metrics.

Everything is deployed under kube-system namespace, you can change that by updating the YAML file.

To get the manifests just run:

curl -L -O https://raw.githubusercontent.com/elastic/beats/6.0/deploy/kubernetes/metricbeat-kubernetes.yaml

Some parameters are exposed in the manifest to configure logs destination, by default they will use an existing Elasticsearch deploy if it’s present, but you may want to change that behavior, so just edit the YAML file and modify them:

- name: ELASTICSEARCH_HOST
  value: elasticsearch
- name: ELASTICSEARCH_PORT
  value: "9200"
- name: ELASTICSEARCH_USERNAME
  value: elastic
- name: ELASTICSEARCH_PASSWORD
  value: changeme

Metricbeat gets some metrics from kube-state-metrics, you will need to deploy it if it’s not already running.

To deploy Metricbeat to Kubernetes just run:

kubectl create -f metricbeat-kubernetes.yaml

Then you should be able to check the status by running:

$ kubectl --namespace=kube-system  get ds/metricbeat

NAME       DESIRED   CURRENT   READY     UP-TO-DATE   AVAILABLE   NODE-SELECTOR   AGE
metricbeat   32        32        0         32           0           <none>          1m

$ kubectl --namespace=kube-system  get deploy/metricbeat

NAME                    DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
metricbeat                1         1         1            1           1m

Metrics should start flowing to Elasticsearch.