Support for collector readinessProbe (#2944)

* enable readiness Probe for otel operator Signed-off-by: Janario Oliveira <[email protected]> * generate CRD and controller changes Signed-off-by: Janario Oliveira <[email protected]> * Adjusted code to be similar to Liveness logic Signed-off-by: Janario Oliveira <[email protected]> * Generated manifests Signed-off-by: Janario Oliveira <[email protected]> * Add changelog Signed-off-by: Janario Oliveira <[email protected]> * Fix lint Signed-off-by: Janario Oliveira <[email protected]> * Removed readinessProbe from alpha CRD Signed-off-by: Janario Oliveira <[email protected]> * Generated manifests Signed-off-by: Janario Oliveira <[email protected]> * Fix lint Signed-off-by: Janario Oliveira <[email protected]> * Centralized probe validation Signed-off-by: Janario Oliveira <[email protected]> --------- Signed-off-by: Janario Oliveira <[email protected]> Co-authored-by: hesam.hamdarsi <[email protected]>
open-telemetry · May 13, 2024 · 3169efd · 3169efd
1 parent c55a97a
commit 3169efd
Show file tree

Hide file tree

Showing 10 changed files with 308 additions and 23 deletions.
diff --git a/.chloggen/collector-readiness-support.yaml b/.chloggen/collector-readiness-support.yaml
@@ -0,0 +1,17 @@
+# One of 'breaking', 'deprecation', 'new_component', 'enhancement', 'bug_fix'
+change_type: enhancement
+
+# The name of the component, or a single word describing the area of concern, (e.g. collector, target allocator, auto-instrumentation, opamp, github action)
+component: collector
+
+# A brief description of the change. Surround your text with quotes ("") if it needs to start with a backtick (`).
+note: Add support for readinessProbe on OpenTelemetryCollector CRD.
+
+# One or more tracking issues related to the change
+issues: [2943]
+
+# (Optional) One or more lines of additional information to render under the primary note.
+# These lines will be padded with 2 spaces and then inserted directly into the document.
+# Use pipe (|) for multiline entries.
+subtext: |
+  Add support for readinessProbe on `OpenTelemetryCollector` and its default similar to the already supported livenessProbe.
diff --git a/apis/v1beta1/collector_webhook.go b/apis/v1beta1/collector_webhook.go
@@ -289,25 +289,14 @@ func (c CollectorWebhook) validate(ctx context.Context, r *OpenTelemetryCollecto
 		return warnings, fmt.Errorf("a valid Ingress hostname has to be defined for subdomain ruleType")
 	}
 
-	if r.Spec.LivenessProbe != nil {
-		if r.Spec.LivenessProbe.InitialDelaySeconds != nil && *r.Spec.LivenessProbe.InitialDelaySeconds < 0 {
-			return warnings, fmt.Errorf("the OpenTelemetry Spec LivenessProbe InitialDelaySeconds configuration is incorrect. InitialDelaySeconds should be greater than or equal to 0")
-		}
-		if r.Spec.LivenessProbe.PeriodSeconds != nil && *r.Spec.LivenessProbe.PeriodSeconds < 1 {
-			return warnings, fmt.Errorf("the OpenTelemetry Spec LivenessProbe PeriodSeconds configuration is incorrect. PeriodSeconds should be greater than or equal to 1")
-		}
-		if r.Spec.LivenessProbe.TimeoutSeconds != nil && *r.Spec.LivenessProbe.TimeoutSeconds < 1 {
-			return warnings, fmt.Errorf("the OpenTelemetry Spec LivenessProbe TimeoutSeconds configuration is incorrect. TimeoutSeconds should be greater than or equal to 1")
-		}
-		if r.Spec.LivenessProbe.SuccessThreshold != nil && *r.Spec.LivenessProbe.SuccessThreshold < 1 {
-			return warnings, fmt.Errorf("the OpenTelemetry Spec LivenessProbe SuccessThreshold configuration is incorrect. SuccessThreshold should be greater than or equal to 1")
-		}
-		if r.Spec.LivenessProbe.FailureThreshold != nil && *r.Spec.LivenessProbe.FailureThreshold < 1 {
-			return warnings, fmt.Errorf("the OpenTelemetry Spec LivenessProbe FailureThreshold configuration is incorrect. FailureThreshold should be greater than or equal to 1")
-		}
-		if r.Spec.LivenessProbe.TerminationGracePeriodSeconds != nil && *r.Spec.LivenessProbe.TerminationGracePeriodSeconds < 1 {
-			return warnings, fmt.Errorf("the OpenTelemetry Spec LivenessProbe TerminationGracePeriodSeconds configuration is incorrect. TerminationGracePeriodSeconds should be greater than or equal to 1")
-		}
+	// validate probes Liveness/Readiness
+	err := validateProbe("LivenessProbe", r.Spec.LivenessProbe)
+	if err != nil {
+		return warnings, err
+	}
+	err = validateProbe("ReadinessProbe", r.Spec.ReadinessProbe)
+	if err != nil {
+		return warnings, err
 	}
 
 	// validate updateStrategy for DaemonSet
@@ -365,6 +354,30 @@ func (c CollectorWebhook) validateTargetAllocatorConfig(ctx context.Context, r *
 	return nil, nil
 }
 
+func validateProbe(probeName string, probe *Probe) error {
+	if probe != nil {
+		if probe.InitialDelaySeconds != nil && *probe.InitialDelaySeconds < 0 {
+			return fmt.Errorf("the OpenTelemetry Spec %s InitialDelaySeconds configuration is incorrect. InitialDelaySeconds should be greater than or equal to 0", probeName)
+		}
+		if probe.PeriodSeconds != nil && *probe.PeriodSeconds < 1 {
+			return fmt.Errorf("the OpenTelemetry Spec %s PeriodSeconds configuration is incorrect. PeriodSeconds should be greater than or equal to 1", probeName)
+		}
+		if probe.TimeoutSeconds != nil && *probe.TimeoutSeconds < 1 {
+			return fmt.Errorf("the OpenTelemetry Spec %s TimeoutSeconds configuration is incorrect. TimeoutSeconds should be greater than or equal to 1", probeName)
+		}
+		if probe.SuccessThreshold != nil && *probe.SuccessThreshold < 1 {
+			return fmt.Errorf("the OpenTelemetry Spec %s SuccessThreshold configuration is incorrect. SuccessThreshold should be greater than or equal to 1", probeName)
+		}
+		if probe.FailureThreshold != nil && *probe.FailureThreshold < 1 {
+			return fmt.Errorf("the OpenTelemetry Spec %s FailureThreshold configuration is incorrect. FailureThreshold should be greater than or equal to 1", probeName)
+		}
+		if probe.TerminationGracePeriodSeconds != nil && *probe.TerminationGracePeriodSeconds < 1 {
+			return fmt.Errorf("the OpenTelemetry Spec %s TerminationGracePeriodSeconds configuration is incorrect. TerminationGracePeriodSeconds should be greater than or equal to 1", probeName)
+		}
+	}
+	return nil
+}
+
 func checkAutoscalerSpec(autoscaler *AutoscalerSpec) error {
 	if autoscaler.Behavior != nil {
 		if autoscaler.Behavior.ScaleDown != nil && autoscaler.Behavior.ScaleDown.StabilizationWindowSeconds != nil &&

diff --git a/apis/v1beta1/collector_webhook_test.go b/apis/v1beta1/collector_webhook_test.go
@@ -1026,6 +1026,17 @@ func TestOTELColValidatingWebhook(t *testing.T) {
 			},
 			expectedErr: "the OpenTelemetry Spec LivenessProbe InitialDelaySeconds configuration is incorrect",
 		},
+		{
+			name: "invalid InitialDelaySeconds readiness",
+			otelcol: OpenTelemetryCollector{
+				Spec: OpenTelemetryCollectorSpec{
+					ReadinessProbe: &Probe{
+						InitialDelaySeconds: &minusOne,
+					},
+				},
+			},
+			expectedErr: "the OpenTelemetry Spec ReadinessProbe InitialDelaySeconds configuration is incorrect",
+		},
 		{
 			name: "invalid PeriodSeconds",
 			otelcol: OpenTelemetryCollector{
@@ -1037,6 +1048,17 @@ func TestOTELColValidatingWebhook(t *testing.T) {
 			},
 			expectedErr: "the OpenTelemetry Spec LivenessProbe PeriodSeconds configuration is incorrect",
 		},
+		{
+			name: "invalid PeriodSeconds readiness",
+			otelcol: OpenTelemetryCollector{
+				Spec: OpenTelemetryCollectorSpec{
+					ReadinessProbe: &Probe{
+						PeriodSeconds: &zero,
+					},
+				},
+			},
+			expectedErr: "the OpenTelemetry Spec ReadinessProbe PeriodSeconds configuration is incorrect",
+		},
 		{
 			name: "invalid TimeoutSeconds",
 			otelcol: OpenTelemetryCollector{
@@ -1048,6 +1070,17 @@ func TestOTELColValidatingWebhook(t *testing.T) {
 			},
 			expectedErr: "the OpenTelemetry Spec LivenessProbe TimeoutSeconds configuration is incorrect",
 		},
+		{
+			name: "invalid TimeoutSeconds readiness",
+			otelcol: OpenTelemetryCollector{
+				Spec: OpenTelemetryCollectorSpec{
+					ReadinessProbe: &Probe{
+						TimeoutSeconds: &zero,
+					},
+				},
+			},
+			expectedErr: "the OpenTelemetry Spec ReadinessProbe TimeoutSeconds configuration is incorrect",
+		},
 		{
 			name: "invalid SuccessThreshold",
 			otelcol: OpenTelemetryCollector{
@@ -1059,6 +1092,17 @@ func TestOTELColValidatingWebhook(t *testing.T) {
 			},
 			expectedErr: "the OpenTelemetry Spec LivenessProbe SuccessThreshold configuration is incorrect",
 		},
+		{
+			name: "invalid SuccessThreshold readiness",
+			otelcol: OpenTelemetryCollector{
+				Spec: OpenTelemetryCollectorSpec{
+					ReadinessProbe: &Probe{
+						SuccessThreshold: &zero,
+					},
+				},
+			},
+			expectedErr: "the OpenTelemetry Spec ReadinessProbe SuccessThreshold configuration is incorrect",
+		},
 		{
 			name: "invalid FailureThreshold",
 			otelcol: OpenTelemetryCollector{
@@ -1070,6 +1114,17 @@ func TestOTELColValidatingWebhook(t *testing.T) {
 			},
 			expectedErr: "the OpenTelemetry Spec LivenessProbe FailureThreshold configuration is incorrect",
 		},
+		{
+			name: "invalid FailureThreshold readiness",
+			otelcol: OpenTelemetryCollector{
+				Spec: OpenTelemetryCollectorSpec{
+					ReadinessProbe: &Probe{
+						FailureThreshold: &zero,
+					},
+				},
+			},
+			expectedErr: "the OpenTelemetry Spec ReadinessProbe FailureThreshold configuration is incorrect",
+		},
 		{
 			name: "invalid TerminationGracePeriodSeconds",
 			otelcol: OpenTelemetryCollector{
@@ -1081,6 +1136,17 @@ func TestOTELColValidatingWebhook(t *testing.T) {
 			},
 			expectedErr: "the OpenTelemetry Spec LivenessProbe TerminationGracePeriodSeconds configuration is incorrect",
 		},
+		{
+			name: "invalid TerminationGracePeriodSeconds readiness",
+			otelcol: OpenTelemetryCollector{
+				Spec: OpenTelemetryCollectorSpec{
+					ReadinessProbe: &Probe{
+						TerminationGracePeriodSeconds: &zero64,
+					},
+				},
+			},
+			expectedErr: "the OpenTelemetry Spec ReadinessProbe TerminationGracePeriodSeconds configuration is incorrect",
+		},
 		{
 			name: "invalid AdditionalContainers",
 			otelcol: OpenTelemetryCollector{

diff --git a/apis/v1beta1/opentelemetrycollector_types.go b/apis/v1beta1/opentelemetrycollector_types.go
@@ -103,6 +103,10 @@ type OpenTelemetryCollectorSpec struct {
 	// It is only effective when healthcheckextension is configured in the OpenTelemetry Collector pipeline.
 	// +optional
 	LivenessProbe *Probe `json:"livenessProbe,omitempty"`
+	// Readiness config for the OpenTelemetry Collector except the probe handler which is auto generated from the health extension of the collector.
+	// It is only effective when healthcheckextension is configured in the OpenTelemetry Collector pipeline.
+	// +optional
+	ReadinessProbe *Probe `json:"readinessProbe,omitempty"`
 
 	// ObservabilitySpec defines how telemetry data gets handled.
 	//
@@ -206,7 +210,7 @@ type TargetAllocatorEmbedded struct {
 	PodDisruptionBudget *PodDisruptionBudgetSpec `json:"podDisruptionBudget,omitempty"`
 }
 
-// Probe defines the OpenTelemetry's pod probe config. Only Liveness probe is supported currently.
+// Probe defines the OpenTelemetry's pod probe config.
 type Probe struct {
 	// Number of seconds after the container has started before liveness probes are initiated.
 	// Defaults to 0 seconds. Minimum value is 0.

diff --git a/apis/v1beta1/zz_generated.deepcopy.go b/apis/v1beta1/zz_generated.deepcopy.go
diff --git a/bundle/manifests/opentelemetry.io_opentelemetrycollectors.yaml b/bundle/manifests/opentelemetry.io_opentelemetrycollectors.yaml
@@ -6654,6 +6654,27 @@ spec:
                 x-kubernetes-list-type: atomic
               priorityClassName:
                 type: string
+              readinessProbe:
+                properties:
+                  failureThreshold:
+                    format: int32
+                    type: integer
+                  initialDelaySeconds:
+                    format: int32
+                    type: integer
+                  periodSeconds:
+                    format: int32
+                    type: integer
+                  successThreshold:
+                    format: int32
+                    type: integer
+                  terminationGracePeriodSeconds:
+                    format: int64
+                    type: integer
+                  timeoutSeconds:
+                    format: int32
+                    type: integer
+                type: object
               replicas:
                 format: int32
                 type: integer

diff --git a/config/crd/bases/opentelemetry.io_opentelemetrycollectors.yaml b/config/crd/bases/opentelemetry.io_opentelemetrycollectors.yaml
@@ -6640,6 +6640,27 @@ spec:
                 x-kubernetes-list-type: atomic
               priorityClassName:
                 type: string
+              readinessProbe:
+                properties:
+                  failureThreshold:
+                    format: int32
+                    type: integer
+                  initialDelaySeconds:
+                    format: int32
+                    type: integer
+                  periodSeconds:
+                    format: int32
+                    type: integer
+                  successThreshold:
+                    format: int32
+                    type: integer
+                  terminationGracePeriodSeconds:
+                    format: int64
+                    type: integer
+                  timeoutSeconds:
+                    format: int32
+                    type: integer
+                type: object
               replicas:
                 format: int32
                 type: integer

diff --git a/docs/api.md b/docs/api.md
@@ -30050,6 +30050,14 @@ If not specified, the pod priority will be default or zero if there is no
 default.<br/>
         </td>
         <td>false</td>
+      </tr><tr>
+        <td><b><a href="#opentelemetrycollectorspecreadinessprobe">readinessProbe</a></b></td>
+        <td>object</td>
+        <td>
+          Readiness config for the OpenTelemetry Collector except the probe handler which is auto generated from the health extension of the collector.
+It is only effective when healthcheckextension is configured in the OpenTelemetry Collector pipeline.<br/>
+        </td>
+        <td>false</td>
       </tr><tr>
         <td><b>replicas</b></td>
         <td>integer</td>
@@ -40043,6 +40051,97 @@ More info: https://kubernetes.io/docs/concepts/services-networking/service/#defi
 </table>
 
 
+### OpenTelemetryCollector.spec.readinessProbe
+<sup><sup>[↩ Parent](#opentelemetrycollectorspec-1)</sup></sup>
+
+
+
+Readiness config for the OpenTelemetry Collector except the probe handler which is auto generated from the health extension of the collector.
+It is only effective when healthcheckextension is configured in the OpenTelemetry Collector pipeline.
+
+<table>
+    <thead>
+        <tr>
+            <th>Name</th>
+            <th>Type</th>
+            <th>Description</th>
+            <th>Required</th>
+        </tr>
+    </thead>
+    <tbody><tr>
+        <td><b>failureThreshold</b></td>
+        <td>integer</td>
+        <td>
+          Minimum consecutive failures for the probe to be considered failed after having succeeded.
+Defaults to 3. Minimum value is 1.<br/>
+          <br/>
+            <i>Format</i>: int32<br/>
+        </td>
+        <td>false</td>
+      </tr><tr>
+        <td><b>initialDelaySeconds</b></td>
+        <td>integer</td>
+        <td>
+          Number of seconds after the container has started before liveness probes are initiated.
+Defaults to 0 seconds. Minimum value is 0.
+More info: https://kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle#container-probes<br/>
+          <br/>
+            <i>Format</i>: int32<br/>
+        </td>
+        <td>false</td>
+      </tr><tr>
+        <td><b>periodSeconds</b></td>
+        <td>integer</td>
+        <td>
+          How often (in seconds) to perform the probe.
+Default to 10 seconds. Minimum value is 1.<br/>
+          <br/>
+            <i>Format</i>: int32<br/>
+        </td>
+        <td>false</td>
+      </tr><tr>
+        <td><b>successThreshold</b></td>
+        <td>integer</td>
+        <td>
+          Minimum consecutive successes for the probe to be considered successful after having failed.
+Defaults to 1. Must be 1 for liveness and startup. Minimum value is 1.<br/>
+          <br/>
+            <i>Format</i>: int32<br/>
+        </td>
+        <td>false</td>
+      </tr><tr>
+        <td><b>terminationGracePeriodSeconds</b></td>
+        <td>integer</td>
+        <td>
+          Optional duration in seconds the pod needs to terminate gracefully upon probe failure.
+The grace period is the duration in seconds after the processes running in the pod are sent
+a termination signal and the time when the processes are forcibly halted with a kill signal.
+Set this value longer than the expected cleanup time for your process.
+If this value is nil, the pod's terminationGracePeriodSeconds will be used. Otherwise, this
+value overrides the value provided by the pod spec.
+Value must be non-negative integer. The value zero indicates stop immediately via
+the kill signal (no opportunity to shut down).
+This is a beta field and requires enabling ProbeTerminationGracePeriod feature gate.
+Minimum value is 1. spec.terminationGracePeriodSeconds is used if unset.<br/>
+          <br/>
+            <i>Format</i>: int64<br/>
+        </td>
+        <td>false</td>
+      </tr><tr>
+        <td><b>timeoutSeconds</b></td>
+        <td>integer</td>
+        <td>
+          Number of seconds after which the probe times out.
+Defaults to 1 second. Minimum value is 1.
+More info: https://kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle#container-probes<br/>
+          <br/>
+            <i>Format</i>: int32<br/>
+        </td>
+        <td>false</td>
+      </tr></tbody>
+</table>
+
+
 ### OpenTelemetryCollector.spec.resources
 <sup><sup>[↩ Parent](#opentelemetrycollectorspec-1)</sup></sup>