Merge pull request fluxcd#1630 from bacherfl/poc/keptn-provider

feat: implement a Keptn metrics provider

artifacts/flagger/crd.yaml

@@ -1298,6 +1298,7 @@ spec:
                         - newrelic
                         - graphite
                         - dynatrace
+                        - keptn
                     address:
                       description: API address of this provider
                       type: string

charts/flagger/crds/crd.yaml

@@ -1298,6 +1298,7 @@ spec:
                         - newrelic
                         - graphite
                         - dynatrace
+                        - keptn
                     address:
                       description: API address of this provider
                       type: string

charts/flagger/templates/rbac.yaml

@@ -259,6 +259,19 @@ rules:
       - update
       - patch
       - delete
+  - apiGroups:
+      - metrics.keptn.sh
+    resources:
+      - keptnmetrics
+      - analyses
+    verbs:
+      - get
+      - list
+      - watch
+      - create
+      - update
+      - patch
+      - delete
   - nonResourceURLs:
       - /version
     verbs:

cmd/flagger/main.go

@@ -253,6 +253,7 @@ func main() {
 		fromEnv("EVENT_WEBHOOK_URL", eventWebhook),
 		clusterName,
 		noCrossNamespaceRefs,
+		cfg,
 	)
 
 	// leader election context

docs/gitbook/usage/metrics.md

@@ -668,3 +668,65 @@ Reference the template in the canary analysis:
           max: 1000
         interval: 1m
 ```
+
+## Keptn
+
+You can create custom metric checks using the Keptn provider.
+This Provider allows to verify either the value of a single [KeptnMetric](https://keptn.sh/stable/docs/reference/crd-reference/metric/),
+representing the value of a single metric,
+or of a [Keptn Analysis](https://keptn.sh/stable/docs/reference/crd-reference/analysis/),
+which provides a flexible grading logic for analysing and prioritising a number of different
+metric values coming from different data sources.
+
+This provider requires [Keptn](https://keptn.sh/stable/docs/installation/) to be installed in the cluster.
+
+Example for a Keptn metric template:
+
+```yaml
+apiVersion: flagger.app/v1beta1
+kind: MetricTemplate
+metadata:
+  name: response-time
+  namespace: istio-system
+spec:
+  provider:
+    type: keptn
+  query: keptnmetric/my-namespace/response-time/2m/reporter=destination
+```
+
+This will reference the `KeptnMetric` with the name `response-time` in
+the namespace `my-namespace`, which could look like the following:
+
+```yaml
+apiVersion: metrics.keptn.sh/v1beta1
+kind: KeptnMetric
+metadata:
+  name: response-time
+  namespace: my-namespace
+spec:
+  fetchIntervalSeconds: 10
+  provider:
+    name: my-prometheus-keptn-provider
+  query: histogram_quantile(0.8, sum by(le) (rate(http_server_request_latency_seconds_bucket{status_code='200',
+    job='simple-go-backend'}[5m[])))
+```
+
+The `query` contains the following components, which are divided by `/` characters:
+
+```
+<type>/<namespace>/<resource-name>/<timeframe>/<arguments>
+```
+
+* **type (required)**: Must be either `keptnmetric` or `analysis`.
+* **namespace (required)**: The namespace of the referenced `KeptnMetric`/`AnalysisDefinition`.
+* **resource-name (required):** The name of the referenced `KeptnMetric`/`AnalysisDefinition`.
+* **timeframe (optional)**: The timeframe used for the Analysis.
+This will usually be set to the same value as the analysis interval of a `Canary`.
+Only relevant if the `type` is set to `analysis`.
+* **arguments (optional)**: Arguments to be passed to an `Analysis`.
+Arguments are passed as a list of key value pairs, separated by `;` characters,
+e.g. `foo=bar;bar=foo`. 
+Only relevant if the `type` is set to `analysis`.
+
+For the type `analysis`, the value returned by the provider is either `0`
+(if the analysis failed), or `1` (analysis passed).

go.mod

@@ -9,12 +9,14 @@ require (
 	github.com/davecgh/go-spew v1.1.1
 	github.com/go-logr/zapr v1.3.0
 	github.com/google/go-cmp v0.6.0
+	github.com/google/uuid v1.6.0
 	github.com/googleapis/gax-go/v2 v2.12.4
 	github.com/hashicorp/go-retryablehttp v0.7.7
 	github.com/influxdata/influxdb-client-go/v2 v2.13.0
 	github.com/prometheus/client_golang v1.19.1
 	github.com/stretchr/testify v1.9.0
 	go.uber.org/zap v1.27.0
+	golang.org/x/sync v0.7.0
 	google.golang.org/api v0.182.0
 	google.golang.org/genproto v0.0.0-20240528184218-531527333157
 	google.golang.org/grpc v1.64.0
@@ -47,7 +49,6 @@ require (
 	github.com/google/gnostic-models v0.6.8 // indirect
 	github.com/google/gofuzz v1.2.0 // indirect
 	github.com/google/s2a-go v0.1.7 // indirect
-	github.com/google/uuid v1.6.0 // indirect
 	github.com/googleapis/enterprise-certificate-proxy v0.3.2 // indirect
 	github.com/h2non/parth v0.0.0-20190131123155-b4df798d6542 // indirect
 	github.com/hashicorp/go-cleanhttp v0.5.2 // indirect
@@ -77,7 +78,6 @@ require (
 	golang.org/x/mod v0.15.0 // indirect
 	golang.org/x/net v0.25.0 // indirect
 	golang.org/x/oauth2 v0.20.0 // indirect
-	golang.org/x/sync v0.7.0 // indirect
 	golang.org/x/sys v0.20.0 // indirect
 	golang.org/x/term v0.20.0 // indirect
 	golang.org/x/text v0.15.0 // indirect

kustomize/base/flagger/crd.yaml

@@ -1298,6 +1298,7 @@ spec:
                         - newrelic
                         - graphite
                         - dynatrace
+                        - keptn
                     address:
                       description: API address of this provider
                       type: string

kustomize/base/flagger/rbac.yaml

@@ -241,6 +241,19 @@ rules:
       - update
       - patch
       - delete
+  - apiGroups:
+      - metrics.keptn.sh
+    resources:
+      - keptnmetrics
+      - analyses
+    verbs:
+      - get
+      - list
+      - watch
+      - create
+      - update
+      - patch
+      - delete
   - nonResourceURLs:
       - /version
     verbs:

pkg/controller/controller.go

@@ -21,6 +21,8 @@ import (
 	"sync"
 	"time"
 
+	"k8s.io/client-go/rest"
+
 	"github.com/google/go-cmp/cmp"
 	"go.uber.org/zap"
 	corev1 "k8s.io/api/core/v1"
@@ -49,6 +51,7 @@ const controllerAgentName = "flagger"
 
 // Controller is managing the canary objects and schedules canary deployments
 type Controller struct {
+	kubeConfig           *rest.Config
 	kubeClient           kubernetes.Interface
 	flaggerClient        clientset.Interface
 	flaggerInformers     Informers
@@ -91,6 +94,7 @@ func NewController(
 	eventWebhook string,
 	clusterName string,
 	noCrossNamespaceRefs bool,
+	kubeConfig *rest.Config,
 ) *Controller {
 	logger.Debug("Creating event broadcaster")
 	flaggerscheme.AddToScheme(scheme.Scheme)
@@ -105,6 +109,7 @@ func NewController(
 	recorder.SetInfo(version, meshProvider)
 
 	ctrl := &Controller{
+		kubeConfig:           kubeConfig,
 		kubeClient:           kubeClient,
 		flaggerClient:        flaggerClient,
 		flaggerInformers:     flaggerInformers,

pkg/controller/scheduler_metrics.go

@@ -74,7 +74,7 @@ func (c *Controller) checkMetricProviderAvailability(canary *flaggerv1.Canary) e
 			}
 
 			factory := providers.Factory{}
-			provider, err := factory.Provider(metric.Interval, template.Spec.Provider, credentials)
+			provider, err := factory.Provider(metric.Interval, template.Spec.Provider, credentials, c.kubeConfig)
 			if err != nil {
 				return fmt.Errorf("metric template %s.%s provider %s error: %v",
 					metric.TemplateRef.Name, namespace, template.Spec.Provider.Type, err)
@@ -260,7 +260,7 @@ func (c *Controller) runMetricChecks(canary *flaggerv1.Canary) bool {
 			}
 
 			factory := providers.Factory{}
-			provider, err := factory.Provider(metric.Interval, template.Spec.Provider, credentials)
+			provider, err := factory.Provider(metric.Interval, template.Spec.Provider, credentials, c.kubeConfig)
 			if err != nil {
 				c.recordEventErrorf(canary, "Metric template %s.%s provider %s error: %v",
 					metric.TemplateRef.Name, namespace, template.Spec.Provider.Type, err)

pkg/metrics/providers/factory.go

@@ -18,15 +18,12 @@ package providers
 
 import (
 	flaggerv1 "github.com/fluxcd/flagger/pkg/apis/flagger/v1beta1"
+	rest "k8s.io/client-go/rest"
 )
 
 type Factory struct{}
 
-func (factory Factory) Provider(
-	metricInterval string,
-	provider flaggerv1.MetricTemplateProvider,
-	credentials map[string][]byte,
-) (Interface, error) {
+func (factory Factory) Provider(metricInterval string, provider flaggerv1.MetricTemplateProvider, credentials map[string][]byte, config *rest.Config) (Interface, error) {
 	switch provider.Type {
 	case "prometheus":
 		return NewPrometheusProvider(provider, credentials)
@@ -44,6 +41,8 @@ func (factory Factory) Provider(
 		return NewInfluxdbProvider(provider, credentials)
 	case "dynatrace":
 		return NewDynatraceProvider(metricInterval, provider, credentials)
+	case "keptn":
+		return NewKeptnProvider(config)
 	default:
 		return NewPrometheusProvider(provider, credentials)
 	}