In this release only basic OS, cluster and Oracle metrics are collected through Prometheus and can be visualized with Grafana.
To set up monitoring, run the installation script. This will deploy Prometheus, alert manager, node-exporter and Grafana on your Kubernetes cluster. The deployment will be in a separate namespace called "monitoring".
cd ${PATH_TO_EL_CARRO_RELEASE}
chmod +x ./setup_monitoring.sh
./setup_monitoring.sh install
Once the installation is done, configure your El Carro instance to start the Monitoring Service. To do this, edit your Instance manifest. For example:
vi ${PATH_TO_EL_CARRO_RELEASE}/samples/v1alpha1_instance.yaml
Include the following line:
services:
Monitoring:true
Apply the configuration:
kubectl apply -f ${PATH_TO_EL_CARRO_RELEASE}/samples/v1alpha1_instance.yaml -n $NS
This step points Prometheus to start scraping the Oracle DB monitoring agent.
kubectl apply -f ${PATH_TO_EL_CARRO_RELEASE}/db_monitor.yaml
To view the monitoring metrics in Prometheus you need to port forward the prometheus service:
kubectl port-forward svc/prometheus-k8s 9090 -n monitoring
You can now access prometheus on:
http://localhost:9090
Dashboards are set up in Grafana. To access the dashboard, set up the port forwarding as follows:
kubectl port-forward svc/grafana 3000 -n monitoring
In your browser navigate to
http://localhost:3000/
You will be prompted for a username and password. Use admin for both values. You will be prompted immediately to change the admin password.
Once you log into Grafana, choose Dashboards and then Manage.
To uninstall the monitoring operator and ALL containers, run the following command:
${PATH_TO_EL_CARRO_RELEASE}/setup_monitoring.sh uninstall
If you see the error
fatal: destination path 'kube-prometheus' already exists and is not an empty directory.
then delete kube-prometheus
directory and rerun the uninstall script.