This proof-of-concept (POC) deploys Observatorium + Prometheus + Grafana. See also architecture diagrams.
Components used:
You will need TLS certificates for Observatorium API, Prometheus and Grafana. See Creating TLS certificates for more details on how to generate them. Alternatively, Observatorium also supports OIDC (OpenID Connect) for authentication.
Create observatorium namespace. Everything else will be deployed into this namespace:
$ oc apply --kustomize observatorium-namespace/base
Observatorium requires an S3-compatible storage. If you don't have an existing S3 storage, you can leverage MinIO. Deploy MinIO S3 compatible object storage:
$ oc apply --kustomize minio/base
Install Observatorium operator:
$ oc apply --kustomize observatorium-operator/base
Deploy Observatorium. Search for REPLACE_ME string and customize manifests before deploying:
$ oc apply --kustomize observatorium-instance/base
Deploy Prometheus operator:
$ oc apply --kustomize prometheus-operator/base
Deploy Prometheus instance. Search for REPLACE_ME string and customize the manifests before deploying.
$ kustomize build prometheus-instance/base | oc apply --filename -
Deploy Grafana operator:
$ oc apply --kustomize grafana-operator/base
Deploy Grafana instance:
$ oc apply --kustomize grafana-instance/base
For testing purposes, you can expose some of the endpoints. Note that the commands below create OpenShift routes which expose unprotected services to the outside of the cluster. If this is an issue, use oc port-forward instead.
Create route for Prometheus UI:
$ oc expose svc prometheus-operated
Create route for Thanos Query UI:
$ oc expose svc observatorium-thanos-query --port http
Create route for Thanos Alerts Web UI (inactive, pending, firing alerts):
$ oc expose svc observatorium-thanos-rule --port http
