Configuring NooBaa Prometheus Alert Rules - OpenShift Data Foundation 4.22 (Developer Preview)

Updated 9 Jul 2026

Overview

NooBaa ships a PrometheusRule custom resource that contains recording rules and alert definitions. This CR is automatically created by the NooBaa operator during system setup.

The CR is named <noobaa-name>-prometheus-rules in the NooBaa namespace (e.g. noobaa-prometheus-rules for a default installation).

Important: ODF Does Not Reconcile Prometheus Rules

Neither the NooBaa operator nor ODF reconciles the PrometheusRule spec after its initial creation. The operator only ensures the CR exists — it does not enforce or overwrite its contents on subsequent reconciliation cycles.

This means:

You can directly edit the PrometheusRule CR to customize alerts and recording rules.
Your changes will persist across operator reconciliations and pod restarts.
If the CR is deleted, the operator will recreate it with the default rules on the next reconciliation cycle.

List the PrometheusRule CR.

Verify the `PrometheusRule` exists in your namespace:


```
$ oc -n  get prometheusrules
```


Expected output:

NAME                        AGE
noobaa-prometheus-rules     1d

View the Current Rules.

Inspect the full CR to understand the existing rules before making changes:


```
$ oc -n  get prometheusrule -prometheus-rules -o yaml
```


The `PrometheusRule` CR contains several rule groups organized by function. Knowing these group names can help you target specific sections when editing the YAML:

Rule Group	Type	Description
`noobaa-telemeter.rules`	Recording	Aggregated metrics for unhealthy buckets, bucket counts, object counts, accounts, and usage
`noobaa-odf.rules`	Recording	ODF integration metrics — health status, capacity, IOPS, throughput
`noobaa-replication.rules`	Recording	Replication write sizes, counts, errors, and cycle totals
`replication-alert.rules`	Alert	Fires when a replication target bucket is unreachable for more than 5 minutes
`bucket-state-alert.rules`	Alert	Fires on bucket error states, size/quantity quota warnings, and backing store low capacity
`resource-state-alert.rules`	Alert	Fires when a backing store or namespace resource enters error state
`system-capacity-alert.rules`	Alert	Fires at 85%, 95%, and 100% system capacity thresholds
`noobaa-db-alert.rules`	Alert	Fires when the NooBaa database PVC reaches 80% or exceeds 90% capacity

Edit the PrometheusRule CR.

Use `oc edit` to modify the rules directly:


```
$ oc -n  edit prometheusrule -prometheus-rules
```


*For example*: 

Change an alert threshold — raise the system capacity warning from 85% to 90% and fire the bucket error alert only after 15 minutes instead of 5:

  # Before
  - alert: NooBaaSystemCapacityWarning85
    expr: |
      NooBaa_system_capacity > 85
    for: 5m

  # After
  - alert: NooBaaSystemCapacityWarning90  # Optionally rename alert for clarity
    expr: |
      NooBaa_system_capacity > 90
    for: 15m

Verify the Changes

Confirm the CR was updated correctly:


```
$ oc -n  get prometheusrule -prometheus-rules -o yaml
```

Restoring Default Rules

If you want to reset the rules back to defaults:

Delete the PrometheusRule CR:

 ```
 $ oc -n <namespace> delete prometheusrule <noobaa-name>-prometheus-rules
 ```

The NooBaa operator will automatically recreate it with the default rules on the next reconciliation cycle (typically within seconds).

Key points

Edits are safe — The NooBaa operator and ODF will not overwrite your changes.
Deletion triggers recreation — Deleting the CR restores defaults automatically.
Labels matter — The CR uses labels prometheus: k8s and role: alert-rules. Your Prometheus instance must have matching ruleSelector labels to discover these rules.

Article Type

General