feat: support size-based retention for MonitoringStack

bundle/manifests/monitoring.rhobs_monitoringstacks.yaml

@@ -1715,10 +1715,23 @@ spec:
               retention:
                 default: 120h
                 description: |-
-                  Time duration to retain data for. Default is '120h',
-                  and must match the regular expression `[0-9]+(ms|s|m|h|d|w|y)` (milliseconds seconds minutes hours days weeks years).
+                  Time duration to retain data for. Default is '120h', and the value must
+                  match the regular expression `[0-9]+(ms|s|m|h|d|w|y)` (milliseconds
+                  seconds minutes hours days weeks years).
+
+                  When both retention and retentionSize are defined, whichever triggers
+                  first will be applied.
                 pattern: ^(0|(([0-9]+)y)?(([0-9]+)w)?(([0-9]+)d)?(([0-9]+)h)?(([0-9]+)m)?(([0-9]+)s)?(([0-9]+)ms)?)$
                 type: string
+              retentionSize:
+                description: |-
+                  retentionSize defines the maximum number of bytes used by the Prometheus
+                  data. By default the size is unlimited.
+
+                  When both retention and retentionSize are defined, whichever triggers
+                  first will be applied.
+                pattern: (^0|([0-9]*[.])?[0-9]+((K|M|G|T|E|P)i?)?B)$
+                type: string
               tolerations:
                 description: Define tolerations for Monitoring Stack Pods.
                 items:

bundle/manifests/observability-operator.clusterserviceversion.yaml

@@ -42,7 +42,7 @@ metadata:
     categories: Monitoring
     certified: "false"
     containerImage: observability-operator:1.3.0
-    createdAt: "2025-11-27T18:47:09Z"
+    createdAt: "2025-11-28T13:17:19Z"
     description: A Go based Kubernetes operator to setup and manage highly available
       Monitoring Stack using Prometheus, Alertmanager and Thanos Querier.
     operatorframework.io/cluster-monitoring: "true"

deploy/crds/common/monitoring.rhobs_monitoringstacks.yaml

@@ -1715,10 +1715,23 @@ spec:
               retention:
                 default: 120h
                 description: |-
-                  Time duration to retain data for. Default is '120h',
-                  and must match the regular expression `[0-9]+(ms|s|m|h|d|w|y)` (milliseconds seconds minutes hours days weeks years).
+                  Time duration to retain data for. Default is '120h', and the value must
+                  match the regular expression `[0-9]+(ms|s|m|h|d|w|y)` (milliseconds
+                  seconds minutes hours days weeks years).
+
+                  When both retention and retentionSize are defined, whichever triggers
+                  first will be applied.
                 pattern: ^(0|(([0-9]+)y)?(([0-9]+)w)?(([0-9]+)d)?(([0-9]+)h)?(([0-9]+)m)?(([0-9]+)s)?(([0-9]+)ms)?)$
                 type: string
+              retentionSize:
+                description: |-
+                  retentionSize defines the maximum number of bytes used by the Prometheus
+                  data. By default the size is unlimited.
+
+                  When both retention and retentionSize are defined, whichever triggers
+                  first will be applied.
+                pattern: (^0|([0-9]*[.])?[0-9]+((K|M|G|T|E|P)i?)?B)$
+                type: string
               tolerations:
                 description: Define tolerations for Monitoring Stack Pods.
                 items:

docs/api.md

@@ -173,12 +173,27 @@ To disable service discovery, set to null. E.g. resourceSelector:.<br/>
         <td><b>retention</b></td>
         <td>string</td>
         <td>
-          Time duration to retain data for. Default is '120h',
-and must match the regular expression `[0-9]+(ms|s|m|h|d|w|y)` (milliseconds seconds minutes hours days weeks years).<br/>
+          Time duration to retain data for. Default is '120h', and the value must
+match the regular expression `[0-9]+(ms|s|m|h|d|w|y)` (milliseconds
+seconds minutes hours days weeks years).
+
+When both retention and retentionSize are defined, whichever triggers
+first will be applied.<br/>
           <br/>
             <i>Default</i>: 120h<br/>
         </td>
         <td>false</td>
+      </tr><tr>
+        <td><b>retentionSize</b></td>
+        <td>string</td>
+        <td>
+          retentionSize defines the maximum number of bytes used by the Prometheus
+data. By default the size is unlimited.
+
+When both retention and retentionSize are defined, whichever triggers
+first will be applied.<br/>
+        </td>
+        <td>false</td>
       </tr><tr>
         <td><b><a href="#monitoringstackspectolerationsindex">tolerations</a></b></td>
         <td>[]object</td>

pkg/apis/monitoring/v1alpha1/types.go

@@ -103,11 +103,25 @@ type MonitoringStackSpec struct {
 	// +optional
 	CreateClusterRoleBindings ClusterRoleBindingPolicy `json:"createClusterRoleBindings,omitempty"`
 
-	// Time duration to retain data for. Default is '120h',
-	// and must match the regular expression `[0-9]+(ms|s|m|h|d|w|y)` (milliseconds seconds minutes hours days weeks years).
+	// Time duration to retain data for. Default is '120h', and the value must
+	// match the regular expression `[0-9]+(ms|s|m|h|d|w|y)` (milliseconds
+	// seconds minutes hours days weeks years).
+	//
+	// When both retention and retentionSize are defined, whichever triggers
+	// first will be applied.
+	//
 	// +kubebuilder:default="120h"
 	Retention monv1.Duration `json:"retention,omitempty"`
 
+	// retentionSize defines the maximum number of bytes used by the Prometheus
+	// data. By default the size is unlimited.
+	//
+	// When both retention and retentionSize are defined, whichever triggers
+	// first will be applied.
+	//
+	// +optional
+	RetentionSize monv1.ByteSize `json:"retentionSize,omitempty"`
+
 	// Define resources requests and limits for Monitoring Stack Pods.
 	// +optional
 	// +kubebuilder:default={requests:{cpu: "100m", memory: "256Mi"}, limits:{memory: "512Mi", cpu: "500m"}}

pkg/controllers/monitoring/monitoring-stack/components.go

@@ -210,6 +210,7 @@ func newPrometheus(
 				}(),
 			},
 			Retention:             ms.Spec.Retention,
+			RetentionSize:         ms.Spec.RetentionSize,
 			RuleSelector:          prometheusSelector,
 			RuleNamespaceSelector: ms.Spec.NamespaceSelector,
 			Thanos: &monv1.ThanosSpec{

test/e2e/monitoring_stack_controller_test.go

@@ -293,7 +293,6 @@ func reconcileRevertsManualChanges(t *testing.T) {
 	f.GetResourceWithRetry(t, ms.Name, ms.Namespace, &generated)
 
 	// update the prometheus created by monitoring-stack controller
-
 	modified := generated.DeepCopy()
 	modified.Spec.ServiceMonitorSelector = &metav1.LabelSelector{
 		MatchLabels: map[string]string{
@@ -348,28 +347,52 @@ func validateStackLogLevel(t *testing.T) {
 }
 
 func validateStackRetention(t *testing.T) {
-	invalidRetention := []monv1.Duration{
-		"100days",
-		"100ducks",
-		"100 days",
-		"100 hours",
-		"100 h",
-		"100 s",
-		"100d   ",
-	}
-
-	ms := newMonitoringStack(t, "invalid-retention")
-	for _, v := range invalidRetention {
-		ms.Spec.Retention = v
-		err := f.K8sClient.Create(context.Background(), ms)
-		assert.ErrorContains(t, err, `spec.retention: Invalid value`)
-	}
+	t.Run("time-based", func(t *testing.T) {
+		invalidRetention := []monv1.Duration{
+			"100days",
+			"100ducks",
+			"100 days",
+			"100 hours",
+			"100 h",
+			"100 s",
+			"100d   ",
+		}
 
-	validMS := newMonitoringStack(t, "valid-retention")
-	validMS.Spec.Retention = "100h"
+		ms := newMonitoringStack(t, "invalid-retention")
+		for _, v := range invalidRetention {
+			ms.Spec.Retention = v
+			err := f.K8sClient.Create(context.Background(), ms)
+			assert.ErrorContains(t, err, `spec.retention: Invalid value`)
+		}
 
-	err := f.K8sClient.Create(context.Background(), validMS)
-	assert.NilError(t, err, `100h is a valid retention period`)
+		validMS := newMonitoringStack(t, "valid-retention")
+		validMS.Spec.Retention = "100h"
+
+		err := f.K8sClient.Create(context.Background(), validMS)
+		assert.NilError(t, err, `100h is a valid retention period`)
+	})
+
+	t.Run("size-based", func(t *testing.T) {
+		invalidRetention := []monv1.ByteSize{
+			"1gb",
+			"1foo",
+			"1 GB",
+			"1GB   ",
+		}
+
+		ms := newMonitoringStack(t, "invalid-retention-size")
+		for _, v := range invalidRetention {
+			ms.Spec.RetentionSize = v
+			err := f.K8sClient.Create(context.Background(), ms)
+			assert.ErrorContains(t, err, `spec.retentionSize: Invalid value`)
+		}
+
+		validMS := newMonitoringStack(t, "valid-retention-size")
+		validMS.Spec.RetentionSize = "1GB"
+
+		err := f.K8sClient.Create(context.Background(), validMS)
+		assert.NilError(t, err)
+	})
 }
 
 func validatePrometheusConfig(t *testing.T) {