When a user configures Prometheus with persistent storage and CMO knows about the storage size, it would be useful that CMO automatically configures the size-based retention.

To account for WAL data and compaction, the value should be less than the storage capacity (10% is probably a good value but some research would be useful).

Based on the following config, CMO should set the retention size parameter.

apiVersion: v1
kind: ConfigMap
metadata:
  name: cluster-monitoring-config
  namespace: openshift-monitoring
data:
  config.yaml: |
    prometheusK8s:

      retention: 1y

      volumeClaimTemplate:

        spec:

          resources:

            requests:

              storage: 10Gi

Based on the following config, CMO shouldn't set/override the retention size parameter.

apiVersion: v1
kind: ConfigMap
metadata:
  name: cluster-monitoring-config
  namespace: openshift-monitoring
data:
  config.yaml: |
    prometheusK8s:

      retention: 1y

      retentionSize: "" # or 0 or "0"

      volumeClaimTemplate:

        spec:

          resources:

            requests:

              storage: 10Gi

DoD

• If a user configures prometheus with a PersistentVolumeClaim of 100Gi and no retentionSize, CMO should have automatically configured a retention size of 90Gi.
• If retentionSize is defined by the user, CMO shouldn't change it.