Works with:
Fusion Server

Compatible with: 5.0, 5.1, 5.2, 5.3

Prometheus and Grafana Setup and Configuration

1. General Setup and Configuration

General setup and configuration for Prometheus and Grafana occurs in five major steps:

  1. Configure Fusion 5 microservices

  2. Install Prometheus into your namespace

  3. Install Grafana into your namespace

  4. Expose Grafana as a load balancer

  5. Configure the Prometheus datasource

Important
 Prometheus and Grafana support is available in Fusion 5.0.2+.

1.1. Configure the microservice values.yaml files

Prometheus is enabled for individual Fusion 5 microservices by configuring the service’s values.yaml file. Each microservice that supports exposing Prometheus metrics will have a pod.annotations field in its Helm template.

At a minimum, the values.yaml file must contain the following values to enable the use of Prometheus with the microservice:

pod:
  annotations:
    prometheus.io/scrape: "true"
    prometheus.io/port: "<port>"

1.2. Install Prometheus into your namespace

1.2.1. Create a prometheus-values.yaml file

At this time, there are no custom Helm charts for either Prometheus or Grafana. Their respective stable charts can be used.

Installation of Prometheus into your namespace is as simple as declaring the name, namespace, and values.yaml file you wish to use for each installation. Grafana data sources are configured within the Grafana application, so you do not need to provide data source configuration to the Helm chart.

For spring-boot services, the following Prometheus prometheus-values.yaml file should be used, replacing <size> with the appropriate disk size:

alertmanager:
  enabled: false
alertmanagerFiles:
  alertmanager.yml: ""
kubeStateMetrics:
  enabled: true
nodeExporter:
  enabled: false
pushgateway:
  enabled: true
server:
  replicaCount: 3
  statefulSet:
    enabled: true
  retention: 120h
  persistenceVolume:
    size: <size>Gi
  global:
    scrape_interval: 5s
    scrape_timeout: 3s
serverFiles:
  prometheus.yml:
    scrape_configs:
      - job_name: 'spring-services'
        metrics_path: '/actuator/prometheus'
        kubernetes_sd_configs:
          - role: pod
        relabel_configs:
          - source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_scrape]
            action: keep
            regex: true
          - source_labels: [__address__, __meta_kubernetes_pod_annotation_prometheus_io_port]
            action: replace
            regex: ([^:]+)(?::\d+)?;(\d+)
            replacement: $1:$2
            target_label: __address__
          - source_labels: [__meta_kubernetes_pod_annotation_prometheus_io_path]
            action: replace
            target_label: __metrics_path__
            regex: (.+)
          - source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_instance]
            action: replace
            target_label: instance
          - source_labels: [__meta_kubernetes_namespace]
            action: replace
            target_label: kubernetes_namespace
          - source_labels: [__meta_kubernetes_pod_name]
            action: replace
            target_label: kubernetes_pod
          - source_labels: [__meta_kubernetes_service_name]
            action: replace
            target_label: kubernetes_name
          - source_labels: [__meta_kubernetes_pod_label_app_kubernetes_io_component]
            action: replace
            target_label: service

The prometheus-values.yaml above declares that the /actuator/prometheus endpoint will be scraped for all pods that carry the prometheus.io/scrape: “true” annotation. The endpoint will be accessed via the port defined by the annotation prometheus.io/port, as established in the Fusion 5 microservices values.yaml file.

Other declared meta attributes for metrics are included as labels. This includes the namespace, pod/instance name, service name, and component name.

Additional configuration options available to the stable/prometheus Helm chart can be found on the Prometheus Helm chart GitHub repo.

1.2.2. Install Prometheus

Install Prometheus into your namespace using the following Helm command, replacing <namespace> with your namespace:

helm install <namespace>-prometheus --namespace <namespace> -f prometheus-values.yaml stable/prometheus

1.3. Install Grafana into your namespace

After Prometheus has been installed into your namespace, install Grafana into your namespace using a similar process.

1.3.1. Create a grafana-values.yaml file

To begin, create a grafana-values.yaml file:

server:
  ingress:
    enabled: true
  service:
    type: ClusterIP
persistence:
  enabled: true
  type: pvc

1.3.2. Install Grafana

After creating this file, run the following helm command to install it:

helm install --name <namespace>-grafana --namespace <namespace> -f grafana-values.yaml stable/grafana

Verify the helm command has finished rolling out Grafana by checking that the pod has a "ready" state: kubectl get pods --namespace <namespace>.

Retrieve the default admin password:

kubectl get secret --namespace <namespace> <namespace>-grafana -o jsonpath="{.data.admin-password}" | base64 --decode ; echo

This will emit a base64-encoded password.

1.4. Expose Grafana as a load balancer

Expose Grafana as a load balancer using the following command:

kubectl expose deployment <namespace>-grafana --type=LoadBalancer --name=grafana

You can use kubectl get services --namespace <namespace> to determine when the load balancer setup is complete and what its IP address is. Once complete, you are ready to begin using Grafana.

Go to your browser and enter the http://<loadbalancerPublicIP>:3000. Enter the username admin@localhost and the password that was returned in the previous step.

Tip
 After successfully logging in, create a new administrative user with a more intuitive password.

1.5. Configure the Prometheus datasource

To begin using the Prometheus datasource with Grafana, navigate to the gear icon and select Data Sources.

Click Add Data Source and Select Prometheus as the datasource type. Ensure the Default switch is toggled on. You will be directed to enter the URL for the Prometheus server. Enter http://<namespace>-prometheus-server.

Configure any additional fields as desired, then click Save and Test.

2. Import Dashboards

There are several JSON files that represent out-of-the-box dashboards that can be used with Grafana. To import them, click on the plus icon on the left side of the screen and click Import.

After importing, the dashboards are located on the home page under Dashboards.

3. Enable Solr Metrics

Create a file named solr_exporter.yaml with the following contents:

exporter:
  enabled: true
  service:
    annotations:
      prometheus.io/scrape: "true"
      prometheus.io/port: "9983"
      prometheus.io/path: "/metrics"

Regenerate the templated Kubernetes file using the following command from the Solr Helm chart folder:

helm template --name "<namespace>-solr" . --values solr_exporter.yaml > solr.yaml

When the process is complete, run the following command:

kubectl apply -f solr.yaml

Kubernetes will start a new service for the solr-exporter:

service "solr-exporter" created
deployment.apps "solr-exporter" created

You can now query Solr metrics in Prometheus.

Next steps

At this point you can begin building Dashboards that query Prometheus data being scraped from Fusion 5 services that are configured with the appropriate pod annotations as described in an earlier section of this document.

See also: