Grafana Operator

not meant for production use

This Operator is based on the grafana-watcher sidecar from Prometheus Operator and the Rolebinding Operator example.

Currently it simply watches for new ConfigMaps and if they define the annotation grafana.net/dashboards as "true" it will POST each dashboard from the ConfigMap to Grafana.

Additionally, if the ConfigMaps define the annotation grafana.net/datasource as "true" it will POST each daatsource from the ConfigMap to Grafana. This requires Grafana 5.x.

Usage

--run-outside-cluster # Uses ~/.kube/config rather than in cluster configuration
--grafana-url # Sets the URL and authentication to use to access the Grafana API

Development

Build from source

1. make install_deps
2. make build
3. ./bin/grafana-operator --run-outside-cluster 1 --grafana-url <GRAFANA URL>

Easiest way to install just Grafana to Kubernetes for playing with helm: helm install stable/grafana then add the dashboards, kubectl apply -f examples/grafana-dashboards.yaml

Needed

List of some of the capabilities needed to make this operator functional.

• Grafana CRD. The Operator should handle launching configured Grafana deployments.
• Support for datasources in configmaps being created in the deployed Grafanas.
• Post all existing dashboards to new Grafana deployments.
• Label based association of Grafana instances and its dashboards/datasources?
• Cross namespace support?