[OCP 3.11] Add Inventory file options to support default prometheus deployment

[netzzer]

For OCP 3.11.16 by default prometheus+grafana are installed and launched from the System Console. The inventory file options below will change the default storage for the operator deployment of prometheus from emptyDir to persistent storage. Currently (OCP 3.11.16) storage will be claimed from the default storageclass. There is a PR - openshift/openshift-ansible#10387 - to use gluster storage by specifying the storageclass. I made the changes to the two files as detailed in this PR, tested using adhoc openshift-monitoring/config.yml playbook and it worked without default storageclass. Will be solution when released in OCP patch release to NOT have to use default storageclass.

OCP documentation - https://docs.openshift.com/container-platform/3.11/install_config/prometheus_cluster_monitoring.html#persistent-storage

Here are the inventory file options and the result with the glustefs-storage-block as the default storageclass.

Added to my OCP 3.11 inventory file for operator deployment

# prometheous operator
openshift_cluster_monitoring_operator_install=true
openshift_cluster_monitoring_operator_node_selector={"node-role.kubernetes.io/infra":"true"}
openshift_cluster_monitoring_operator_prometheus_storage_enabled=true
openshift_cluster_monitoring_operator_alertmanager_storage_enabled=true
openshift_cluster_monitoring_operator_prometheus_storage_capacity=30Gi
openshift_cluster_monitoring_operator_alertmanager_storage_capacity=5Gi

When PR - openshift/openshift-ansible#10387 is merged and OCP GA rpms have PR changes.

openshift_cluster_monitoring_operator_prometheus_storage_class_name='glusterfs-storage-block'
openshift_cluster_monitoring_operator_alertmanager_storage_class_name='glusterfs-storage-block'

[netzzer]

If two cluster solution then storageclass name would be 'gluster-registry-block'.

[netzzer]

Currently by default there are two prometheus pods and three alertmanager installed. Each pod is using the storage as RWO therefore you 2xPVCs for prometheus and 3xPVCs for alertmanager. For example above this would be 2x30GB and 3x5GB that would use the block_host_volume as glusterblock volumes.

[netzzer]

Good resource - https://docs.openshift.com/container-platform/3.11/install_config/prometheus_cluster_monitoring.html#configuring-openshift-cluster-monitoring.

[netzzer]

OCP 3.11 Release notes. Prometheus fully supported, Metrics deprecated.

OCP 3.11 Prometheus and Logging support - https://docs.openshift.com/container-platform/3.11/release_notes/ocp_3_11_release_notes.html#ocp-311-metrics-and-logging

Also, in same OCP 3.11 Release notes
Hawkular Now Deprecated

Hawkular is now deprecated and will be removed in a future release.