- an OCP Cluster installation with the shipped Prometheus Operator (for the kubelet metrics used by the heketi dashboard)
- an Independent OCS installation
- the node_exporter installed on the OCS nodes
clone this project
git clone https://github.com/spagno/ocs-monitoring.git && cd ocs-monitoring
create the project
oc new-project rhgs-metrics
edit the namespace adding openshift.io/node-selector: "" to let the statefulsets run in to the infra servers
install prometheus from the template
oc new-app -f templates/prometheus.yaml -p NAMESPACE=rhgs-metrics -p HEKETI_APP_NAMESPACE=rhgs-app -p HEKETI_INFRA_NAMESPACE=rhgs-infra -p PROMETHEUS_STORAGE_CLASS=glusterfs-registry-block
or
oc new-app -f templates/prometheus.yaml -p NAMESPACE=rhgs-metrics -p HEKETI_APP_NAMESPACE=rhgs
if you didn't install the glusterfs-registry
install grafana from the template
oc new-app -f templates/grafana.yaml -p NAMESPACE=rhgs-metrics -p GRAFANA_STORAGE_CLASS=glusterfs-registry
or
oc new-app -f templates/grafana.yaml -p NAMESPACE=rhgs-metrics
if you didn't install the glusterfs-registry
launch the playbook using the ocp inventory to create the svc and the ep
ansible-playbook -i <ocp_inventory> ocs_monitoring.yaml
if you use multitenant network plugin join the rhgs-metrics namespace to the default one
oc adm pod-network join-projects rhgs-metrics --to default
import the dashboards from the directory dashboard to your grafana installation and create the "prometheus" datasource using "http://prometheus.rhgs-metrics:9090" as server
Add the k8s-app label to the services:
oc -n rhgs-app label svc heketi-storage k8s-app=heketi-storage
oc -n rhgs-infra label svc heketi-registry k8s-app=heketi-registry
or
oc -n rhgs label svc heketi-storage k8s-app=heketi-storage
if you didn't install the glusterfs-registry