Release v0.14.0 changelog

Flagger docs SMI

.circleci/config.yml

@@ -9,6 +9,15 @@ jobs:
       - run: test/e2e-build.sh
       - run: test/e2e-tests.sh
 
+  e2e-smi-istio-testing:
+    machine: true
+    steps:
+      - checkout
+      - run: test/e2e-kind.sh
+      - run: test/e2e-istio.sh
+      - run: test/e2e-smi-istio-build.sh
+      - run: test/e2e-tests.sh canary
+
   e2e-supergloo-testing:
     machine: true
     steps:
@@ -18,6 +27,15 @@ jobs:
       - run: test/e2e-build.sh supergloo:test.supergloo-system
       - run: test/e2e-tests.sh canary
 
+  e2e-gloo-testing:
+    machine: true
+    steps:
+      - checkout
+      - run: test/e2e-kind.sh
+      - run: test/e2e-gloo.sh
+      - run: test/e2e-gloo-build.sh
+      - run: test/e2e-gloo-tests.sh
+
   e2e-nginx-testing:
     machine: true
     steps:
@@ -38,6 +56,13 @@ workflows:
                 - /gh-pages.*/
                 - /docs-.*/
                 - /release-.*/
+      - e2e-smi-istio-testing:
+          filters:
+            branches:
+              ignore:
+                - /gh-pages.*/
+                - /docs-.*/
+                - /release-.*/
       - e2e-supergloo-testing:
           filters:
             branches:
@@ -46,6 +71,13 @@ workflows:
                 - /docs-.*/
                 - /release-.*/
       - e2e-nginx-testing:
+          filters:
+            branches:
+              ignore:
+                - /gh-pages.*/
+                - /docs-.*/
+                - /release-.*/
+      - e2e-gloo-testing:
           filters:
             branches:
               ignore:

CHANGELOG.md

@@ -2,6 +2,15 @@
 
 All notable changes to this project are documented in this file.
 
+## 0.14.0 (2019-05-21) 
+
+Adds support for Service Mesh Interface and [Gloo](https://docs.flagger.app/usage/gloo-progressive-delivery) ingress controller
+
+#### Features
+
+- Add support for SMI (Istio weighted traffic) [#180](https://github.com/weaveworks/flagger/pull/180)
+- Add support for Gloo ingress controller (weighted traffic) [#179](https://github.com/weaveworks/flagger/pull/179)
+
 ## 0.13.2 (2019-04-11) 
 
 Fixes for Jenkins X deployments (prevent the jx GC from removing the primary instance)

Makefile

@@ -24,6 +24,18 @@ run-nginx:
 	-slack-url=https://hooks.slack.com/services/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 \
 	-slack-channel="devops-alerts"
 
+run-smi:
+	go run cmd/flagger/* -kubeconfig=$$HOME/.kube/config -log-level=info -mesh-provider=smi:istio -namespace=smi \
+	-metrics-server=https://prometheus.istio.weavedx.com \
+	-slack-url=https://hooks.slack.com/services/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 \
+	-slack-channel="devops-alerts"
+
+run-gloo:
+	go run cmd/flagger/* -kubeconfig=$$HOME/.kube/config -log-level=info -mesh-provider=gloo -namespace=gloo \
+	-metrics-server=https://prometheus.istio.weavedx.com \
+	-slack-url=https://hooks.slack.com/services/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 \
+	-slack-channel="devops-alerts"
+
 build:
 	docker build -t weaveworks/flagger:$(TAG) . -f Dockerfile
 

README.md

@@ -40,6 +40,7 @@ Flagger documentation can be found at [docs.flagger.app](https://docs.flagger.ap
   * [Istio A/B testing](https://docs.flagger.app/usage/ab-testing)
   * [App Mesh canary deployments](https://docs.flagger.app/usage/appmesh-progressive-delivery)
   * [NGINX ingress controller canary deployments](https://docs.flagger.app/usage/nginx-progressive-delivery)
+  * [Gloo Canary Deployments](https://docs.flagger.app/usage/gloo-progressive-delivery.md)
   * [Monitoring](https://docs.flagger.app/usage/monitoring)
   * [Alerting](https://docs.flagger.app/usage/alerting)
 * Tutorials
@@ -82,7 +83,6 @@ spec:
     # Istio gateways (optional)
     gateways:
     - public-gateway.istio-system.svc.cluster.local
-    - mesh
     # Istio virtual service host names (optional)
     hosts:
     - podinfo.example.com
@@ -93,17 +93,12 @@ spec:
     # HTTP rewrite (optional)
     rewrite:
       uri: /
-    # Envoy timeout and retry policy (optional)
-    headers:
-      request:
-        add:
-          x-envoy-upstream-rq-timeout-ms: "15000"
-          x-envoy-max-retries: "10"
-          x-envoy-retry-on: "gateway-error,connect-failure,refused-stream"
     # cross-origin resource sharing policy (optional)
     corsPolicy:
       allowOrigin:
         - example.com
+    # request timeout (optional)
+    timeout: 5s
   # promote the canary without analysing it (default false)
   skipAnalysis: false
   # define the canary analysis timing and KPIs

artifacts/flagger/account.yaml

@@ -59,6 +59,26 @@ rules:
       - virtualservices
       - virtualservices/status
     verbs: ["*"]
+  - apiGroups:
+      - split.smi-spec.io
+    resources:
+      - trafficsplits
+    verbs: ["*"]
+  - apiGroups:
+      - gloo.solo.io
+    resources:
+      - settings
+      - upstreams
+      - upstreamgroups
+      - proxies
+      - virtualservices
+    verbs: ["*"]
+  - apiGroups:
+      - gateway.solo.io
+    resources:
+      - virtualservices
+      - gateways
+    verbs: ["*"]
   - nonResourceURLs:
       - /version
     verbs:

artifacts/flagger/deployment.yaml

@@ -22,7 +22,7 @@ spec:
       serviceAccountName: flagger
       containers:
       - name: flagger
-        image: weaveworks/flagger:0.13.2
+        image: weaveworks/flagger:0.14.0
         imagePullPolicy: IfNotPresent
         ports:
         - name: http

artifacts/gloo/canary.yaml

@@ -0,0 +1,36 @@
+apiVersion: flagger.app/v1alpha3
+kind: Canary
+metadata:
+  name: podinfo
+  namespace: test
+spec:
+  targetRef:
+    apiVersion: apps/v1
+    kind: Deployment
+    name: podinfo
+  progressDeadlineSeconds: 60
+  autoscalerRef:
+    apiVersion: autoscaling/v2beta1
+    kind: HorizontalPodAutoscaler
+    name: podinfo
+  service:
+    port: 9898
+  canaryAnalysis:
+    interval: 10s
+    threshold: 10
+    maxWeight: 50
+    stepWeight: 5
+    metrics:
+    - name: request-success-rate
+      threshold: 99
+      interval: 1m
+    - name: request-duration
+      threshold: 500
+      interval: 30s
+    webhooks:
+      - name: load-test
+        url: http://flagger-loadtester.test/
+        timeout: 5s
+        metadata:
+          type: cmd
+          cmd: "hey -z 1m -q 10 -c 2 http://gloo.example.com/"

artifacts/gloo/deployment.yaml

@@ -0,0 +1,67 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: podinfo
+  namespace: test
+  labels:
+    app: podinfo
+spec:
+  minReadySeconds: 5
+  revisionHistoryLimit: 5
+  progressDeadlineSeconds: 60
+  strategy:
+    rollingUpdate:
+      maxUnavailable: 0
+    type: RollingUpdate
+  selector:
+    matchLabels:
+      app: podinfo
+  template:
+    metadata:
+      annotations:
+        prometheus.io/scrape: "true"
+      labels:
+        app: podinfo
+    spec:
+      containers:
+      - name: podinfod
+        image: quay.io/stefanprodan/podinfo:1.4.0
+        imagePullPolicy: IfNotPresent
+        ports:
+        - containerPort: 9898
+          name: http
+          protocol: TCP
+        command:
+        - ./podinfo
+        - --port=9898
+        - --level=info
+        - --random-delay=false
+        - --random-error=false
+        env:
+        - name: PODINFO_UI_COLOR
+          value: blue
+        livenessProbe:
+          exec:
+            command:
+            - podcli
+            - check
+            - http
+            - localhost:9898/healthz
+          initialDelaySeconds: 5
+          timeoutSeconds: 5
+        readinessProbe:
+          exec:
+            command:
+            - podcli
+            - check
+            - http
+            - localhost:9898/readyz
+          initialDelaySeconds: 5
+          timeoutSeconds: 5
+        resources:
+          limits:
+            cpu: 2000m
+            memory: 512Mi
+          requests:
+            cpu: 100m
+            memory: 64Mi

artifacts/gloo/hpa.yaml

@@ -0,0 +1,19 @@
+apiVersion: autoscaling/v2beta1
+kind: HorizontalPodAutoscaler
+metadata:
+  name: podinfo
+  namespace: test
+spec:
+  scaleTargetRef:
+    apiVersion: apps/v1
+    kind: Deployment
+    name: podinfo
+  minReplicas: 1
+  maxReplicas: 4
+  metrics:
+  - type: Resource
+    resource:
+      name: cpu
+      # scale up if usage is above
+      # 99% of the requested CPU (100m)
+      targetAverageUtilization: 99

artifacts/gloo/virtual-service.yaml

@@ -0,0 +1,17 @@
+apiVersion: gateway.solo.io/v1
+kind: VirtualService
+metadata:
+  name: podinfo
+  namespace: test
+spec:
+  virtualHost:
+    domains:
+      - '*'
+    name: podinfo.default
+    routes:
+      - matcher:
+          prefix: /
+        routeAction:
+          upstreamGroup:
+            name: podinfo
+            namespace: gloo

artifacts/smi/istio-adapter.yaml

@@ -0,0 +1,131 @@
+apiVersion: apiextensions.k8s.io/v1beta1
+kind: CustomResourceDefinition
+metadata:
+  name: trafficsplits.split.smi-spec.io
+spec:
+  additionalPrinterColumns:
+    - JSONPath: .spec.service
+      description: The service
+      name: Service
+      type: string
+  group: split.smi-spec.io
+  names:
+    kind: TrafficSplit
+    listKind: TrafficSplitList
+    plural: trafficsplits
+    singular: trafficsplit
+  scope: Namespaced
+  subresources:
+    status: {}
+  version: v1alpha1
+  versions:
+    - name: v1alpha1
+      served: true
+      storage: true
+---
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: smi-adapter-istio
+  namespace: istio-system
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRole
+metadata:
+  name: smi-adapter-istio
+rules:
+  - apiGroups:
+      - ""
+    resources:
+      - pods
+      - services
+      - endpoints
+      - persistentvolumeclaims
+      - events
+      - configmaps
+      - secrets
+    verbs:
+      - '*'
+  - apiGroups:
+      - apps
+    resources:
+      - deployments
+      - daemonsets
+      - replicasets
+      - statefulsets
+    verbs:
+      - '*'
+  - apiGroups:
+      - monitoring.coreos.com
+    resources:
+      - servicemonitors
+    verbs:
+      - get
+      - create
+  - apiGroups:
+      - apps
+    resourceNames:
+      - smi-adapter-istio
+    resources:
+      - deployments/finalizers
+    verbs:
+      - update
+  - apiGroups:
+      - split.smi-spec.io
+    resources:
+      - '*'
+    verbs:
+      - '*'
+  - apiGroups:
+      - networking.istio.io
+    resources:
+      - '*'
+    verbs:
+      - '*'
+---
+kind: ClusterRoleBinding
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  name: smi-adapter-istio
+subjects:
+  - kind: ServiceAccount
+    name: smi-adapter-istio
+    namespace: istio-system
+roleRef:
+  kind: ClusterRole
+  name: smi-adapter-istio
+  apiGroup: rbac.authorization.k8s.io
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: smi-adapter-istio
+  namespace: istio-system
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      name: smi-adapter-istio
+  template:
+    metadata:
+      labels:
+        name: smi-adapter-istio
+      annotations:
+        sidecar.istio.io/inject: "false"
+    spec:
+      serviceAccountName: smi-adapter-istio
+      containers:
+        - name: smi-adapter-istio
+          image: docker.io/stefanprodan/smi-adapter-istio:0.0.2-beta.1
+          command:
+          - smi-adapter-istio
+          imagePullPolicy: Always
+          env:
+            - name: WATCH_NAMESPACE
+              value: ""
+            - name: POD_NAME
+              valueFrom:
+                fieldRef:
+                  fieldPath: metadata.name
+            - name: OPERATOR_NAME
+              value: "smi-adapter-istio"

charts/flagger/Chart.yaml

@@ -1,7 +1,7 @@
 apiVersion: v1
 name: flagger
-version: 0.13.2
-appVersion: 0.13.2
+version: 0.14.0
+appVersion: 0.14.0
 kubeVersion: ">=1.11.0-0"
 engine: gotpl
 description: Flagger is a Kubernetes operator that automates the promotion of canary deployments using Istio, App Mesh or NGINX routing for traffic shifting and Prometheus metrics for canary analysis.

charts/flagger/templates/rbac.yaml

@@ -55,6 +55,26 @@ rules:
       - virtualservices
       - virtualservices/status
     verbs: ["*"]
+  - apiGroups:
+      - split.smi-spec.io
+    resources:
+      - trafficsplits
+    verbs: ["*"]
+  - apiGroups:
+      - gloo.solo.io
+    resources:
+      - settings
+      - upstreams
+      - upstreamgroups
+      - proxies
+      - virtualservices
+    verbs: ["*"]
+  - apiGroups:
+      - gateway.solo.io
+    resources:
+      - virtualservices
+      - gateways
+    verbs: ["*"]
   - nonResourceURLs:
       - /version
     verbs:

charts/flagger/values.yaml

@@ -2,7 +2,7 @@
 
 image:
   repository: weaveworks/flagger
-  tag: 0.13.2
+  tag: 0.14.0
   pullPolicy: IfNotPresent
 
 metricsServer: "http://prometheus:9090"

cmd/flagger/main.go

@@ -45,8 +45,8 @@ var (
 func init() {
 	flag.StringVar(&kubeconfig, "kubeconfig", "", "Path to a kubeconfig. Only required if out-of-cluster.")
 	flag.StringVar(&masterURL, "master", "", "The address of the Kubernetes API server. Overrides any value in kubeconfig. Only required if out-of-cluster.")
-	flag.StringVar(&metricsServer, "metrics-server", "http://prometheus:9090", "Prometheus URL")
-	flag.DurationVar(&controlLoopInterval, "control-loop-interval", 10*time.Second, "Kubernetes API sync interval")
+	flag.StringVar(&metricsServer, "metrics-server", "http://prometheus:9090", "Prometheus URL.")
+	flag.DurationVar(&controlLoopInterval, "control-loop-interval", 10*time.Second, "Kubernetes API sync interval.")
 	flag.StringVar(&logLevel, "log-level", "debug", "Log level can be: debug, info, warning, error.")
 	flag.StringVar(&port, "port", "8080", "Port to listen on.")
 	flag.StringVar(&slackURL, "slack-url", "", "Slack hook URL.")
@@ -55,9 +55,9 @@ func init() {
 	flag.IntVar(&threadiness, "threadiness", 2, "Worker concurrency.")
 	flag.BoolVar(&zapReplaceGlobals, "zap-replace-globals", false, "Whether to change the logging level of the global zap logger.")
 	flag.StringVar(&zapEncoding, "zap-encoding", "json", "Zap logger encoding.")
-	flag.StringVar(&namespace, "namespace", "", "Namespace that flagger would watch canary object")
-	flag.StringVar(&meshProvider, "mesh-provider", "istio", "Service mesh provider, can be istio or appmesh")
-	flag.StringVar(&selectorLabels, "selector-labels", "app,name,app.kubernetes.io/name", "List of pod labels that Flagger uses to create pod selectors")
+	flag.StringVar(&namespace, "namespace", "", "Namespace that flagger would watch canary object.")
+	flag.StringVar(&meshProvider, "mesh-provider", "istio", "Service mesh provider, can be istio, appmesh, supergloo, nginx or smi.")
+	flag.StringVar(&selectorLabels, "selector-labels", "app,name,app.kubernetes.io/name", "List of pod labels that Flagger uses to create pod selectors.")
 }
 
 func main() {
@@ -87,12 +87,12 @@ func main() {
 
 	meshClient, err := clientset.NewForConfig(cfg)
 	if err != nil {
-		logger.Fatalf("Error building istio clientset: %v", err)
+		logger.Fatalf("Error building mesh clientset: %v", err)
 	}
 
 	flaggerClient, err := clientset.NewForConfig(cfg)
 	if err != nil {
-		logger.Fatalf("Error building example clientset: %s", err.Error())
+		logger.Fatalf("Error building flagger clientset: %s", err.Error())
 	}
 
 	flaggerInformerFactory := informers.NewSharedInformerFactoryWithOptions(flaggerClient, time.Second*30, informers.WithNamespace(namespace))
@@ -116,7 +116,12 @@ func main() {
 		logger.Infof("Watching namespace %s", namespace)
 	}
 
-	ok, err := metrics.CheckMetricsServer(metricsServer)
+	observerFactory, err := metrics.NewFactory(metricsServer, meshProvider, 5*time.Second)
+	if err != nil {
+		logger.Fatalf("Error building prometheus client: %s", err.Error())
+	}
+
+	ok, err := observerFactory.Client.IsOnline()
 	if ok {
 		logger.Infof("Connected to metrics server %s", metricsServer)
 	} else {
@@ -148,6 +153,7 @@ func main() {
 		logger,
 		slack,
 		routerFactory,
+		observerFactory,
 		meshProvider,
 		version.VERSION,
 		labels,

docs/gitbook/SUMMARY.md

@@ -16,10 +16,12 @@
 * [Istio A/B Testing](usage/ab-testing.md)
 * [App Mesh Canary Deployments](usage/appmesh-progressive-delivery.md)
 * [NGINX Canary Deployments](usage/nginx-progressive-delivery.md)
+* [Gloo Canary Deployments](usage/gloo-progressive-delivery.md)
 * [Monitoring](usage/monitoring.md)
 * [Alerting](usage/alerting.md)
 
 ## Tutorials
 
+* [SMI Istio Canary Deployments](tutorials/flagger-smi-istio.md)
 * [Canaries with Helm charts and GitOps](tutorials/canary-helm-gitops.md)
 * [Zero downtime deployments](tutorials/zero-downtime-deployments.md)

docs/gitbook/how-it-works.md

@@ -38,7 +38,6 @@ spec:
     # Istio gateways (optional)
     gateways:
     - public-gateway.istio-system.svc.cluster.local
-    - mesh
     # Istio virtual service host names (optional)
     hosts:
     - podinfo.example.com

docs/gitbook/install/flagger-install-on-eks-appmesh.md

@@ -163,7 +163,7 @@ Deploy Grafana in the _**appmesh-system**_ namespace:
 ```bash
 helm upgrade -i flagger-grafana flagger/grafana \
 --namespace=appmesh-system \
---set url=http://prometheus.appmesh-system:9090
+--set url=http://flagger-prometheus.appmesh-system:9090
 ```
 
 You can access Grafana using port forwarding:

docs/gitbook/tutorials/flagger-smi-istio.md

@@ -0,0 +1,332 @@
+# Flagger SMI
+
+This guide shows you how to use the SMI Istio adapter and Flagger to automate canary deployments.
+
+### Prerequisites
+
+Flagger requires a Kubernetes cluster **v1.11** or newer with the following admission controllers enabled:
+
+* MutatingAdmissionWebhook
+* ValidatingAdmissionWebhook 
+
+Flagger depends on [Istio](https://istio.io/docs/setup/kubernetes/quick-start/) **v1.0.3** or newer 
+with traffic management, telemetry and Prometheus enabled. 
+
+A minimal Istio installation should contain the following services:
+
+* istio-pilot
+* istio-ingressgateway
+* istio-sidecar-injector
+* istio-telemetry
+* prometheus
+
+### Install Istio and the SMI adapter
+
+Add Istio Helm repository:
+
+```bash
+helm repo add istio.io https://storage.googleapis.com/istio-release/releases/1.1.5/charts
+```
+
+Install Istio CRDs:
+
+```bash
+helm upgrade -i istio-init istio.io/istio-init --wait --namespace istio-system
+
+kubectl -n istio-system wait --for=condition=complete job/istio-init-crd-11
+```
+
+Install Istio:
+
+```bash
+helm upgrade -i istio istio.io/istio --wait --namespace istio-system
+```
+
+Create a generic Istio gateway to expose services outside the mesh on HTTP:
+
+```yaml
+apiVersion: networking.istio.io/v1alpha3
+kind: Gateway
+metadata:
+  name: public-gateway
+  namespace: istio-system
+spec:
+  selector:
+    istio: ingressgateway
+  servers:
+    - port:
+        number: 80
+        name: http
+        protocol: HTTP
+      hosts:
+        - "*"
+```
+
+Save the above resource as public-gateway.yaml and then apply it:
+
+```bash
+kubectl apply -f ./public-gateway.yaml
+```
+
+Find the Gateway load balancer IP and add a DNS record for it:
+
+```bash
+kubectl -n istio-system get svc/istio-ingressgateway -ojson | jq -r .status.loadBalancer.ingress[0].ip
+```
+
+Install the SMI adapter:
+
+```bash
+REPO=https://raw.githubusercontent.com/weaveworks/flagger/master
+
+kubectl apply -f ${REPO}/artifacts/smi/istio-adapter.yaml
+```
+
+### Install Flagger and Grafana
+
+Add Flagger Helm repository:
+
+```bash
+helm repo add flagger https://flagger.app
+```
+
+Deploy Flagger in the _**istio-system**_ namespace:
+
+```bash
+helm upgrade -i flagger flagger/flagger \
+--namespace=istio-system \
+--set image.tag=master-12d84b2 \
+--set meshProvider=smi:istio
+```
+
+Flagger comes with a Grafana dashboard made for monitoring the canary deployments.
+
+Deploy Grafana in the _**istio-system**_ namespace:
+
+```bash
+helm upgrade -i flagger-grafana flagger/grafana \
+--namespace=istio-system \
+--set url=http://prometheus.istio-system:9090
+```
+
+You can access Grafana using port forwarding:
+
+```bash
+kubectl -n istio-system port-forward svc/flagger-grafana 3000:80
+```
+
+### Workloads bootstrap
+
+Create a test namespace with Istio sidecar injection enabled:
+
+```bash
+export REPO=https://raw.githubusercontent.com/weaveworks/flagger/master
+
+kubectl apply -f ${REPO}/artifacts/namespaces/test.yaml
+```
+
+Create a deployment and a horizontal pod autoscaler:
+
+```bash
+kubectl apply -f ${REPO}/artifacts/canaries/deployment.yaml
+kubectl apply -f ${REPO}/artifacts/canaries/hpa.yaml
+```
+
+Deploy the load testing service to generate traffic during the canary analysis:
+
+```bash
+kubectl -n test apply -f ${REPO}/artifacts/loadtester/deployment.yaml
+kubectl -n test apply -f ${REPO}/artifacts/loadtester/service.yaml
+```
+
+Create a canary custom resource (replace example.com with your own domain):
+
+```yaml
+apiVersion: flagger.app/v1alpha3
+kind: Canary
+metadata:
+  name: podinfo
+  namespace: test
+spec:
+  # deployment reference
+  targetRef:
+    apiVersion: apps/v1
+    kind: Deployment
+    name: podinfo
+  # the maximum time in seconds for the canary deployment
+  # to make progress before it is rollback (default 600s)
+  progressDeadlineSeconds: 60
+  # HPA reference (optional)
+  autoscalerRef:
+    apiVersion: autoscaling/v2beta1
+    kind: HorizontalPodAutoscaler
+    name: podinfo
+  service:
+    # container port
+    port: 9898
+    # Istio gateways (optional)
+    gateways:
+    - public-gateway.istio-system.svc.cluster.local
+    # Istio virtual service host names (optional)
+    hosts:
+    - app.example.com
+  canaryAnalysis:
+    # schedule interval (default 60s)
+    interval: 10s
+    # max number of failed metric checks before rollback
+    threshold: 5
+    # max traffic percentage routed to canary
+    # percentage (0-100)
+    maxWeight: 50
+    # canary increment step
+    # percentage (0-100)
+    stepWeight: 10
+    metrics:
+    - name: request-success-rate
+      # minimum req success rate (non 5xx responses)
+      # percentage (0-100)
+      threshold: 99
+      interval: 1m
+    - name: request-duration
+      # maximum req duration P99
+      # milliseconds
+      threshold: 500
+      interval: 30s
+    # generate traffic during analysis
+    webhooks:
+      - name: load-test
+        url: http://flagger-loadtester.test/
+        timeout: 5s
+        metadata:
+          cmd: "hey -z 1m -q 10 -c 2 http://podinfo.test:9898/"
+```
+
+Save the above resource as podinfo-canary.yaml and then apply it:
+
+```bash
+kubectl apply -f ./podinfo-canary.yaml
+```
+
+After a couple of seconds Flagger will create the canary objects:
+
+```bash
+# applied 
+deployment.apps/podinfo
+horizontalpodautoscaler.autoscaling/podinfo
+canary.flagger.app/podinfo
+
+# generated 
+deployment.apps/podinfo-primary
+horizontalpodautoscaler.autoscaling/podinfo-primary
+service/podinfo
+service/podinfo-canary
+service/podinfo-primary
+trafficsplits.split.smi-spec.io/podinfo
+```
+
+### Automated canary promotion
+
+Flagger implements a control loop that gradually shifts traffic to the canary while measuring key performance indicators 
+like HTTP requests success rate, requests average duration and pod health.
+Based on analysis of the KPIs a canary is promoted or aborted, and the analysis result is published to Slack.
+
+![Flagger Canary Stages](https://raw.githubusercontent.com/weaveworks/flagger/master/docs/diagrams/flagger-canary-steps.png)
+
+Trigger a canary deployment by updating the container image:
+
+```bash
+kubectl -n test set image deployment/podinfo \
+podinfod=quay.io/stefanprodan/podinfo:1.4.1
+```
+
+Flagger detects that the deployment revision changed and starts a new rollout:
+
+```text
+kubectl -n istio-system logs deployment/flagger -f | jq .msg
+
+
+New revision detected podinfo.test
+Scaling up podinfo.test
+Waiting for podinfo.test rollout to finish: 0 of 1 updated replicas are available
+Advance podinfo.test canary weight 5
+Advance podinfo.test canary weight 10
+Advance podinfo.test canary weight 15
+Advance podinfo.test canary weight 20
+Advance podinfo.test canary weight 25
+Advance podinfo.test canary weight 30
+Advance podinfo.test canary weight 35
+Advance podinfo.test canary weight 40
+Advance podinfo.test canary weight 45
+Advance podinfo.test canary weight 50
+Copying podinfo.test template spec to podinfo-primary.test
+Waiting for podinfo-primary.test rollout to finish: 1 of 2 updated replicas are available
+Promotion completed! Scaling down podinfo.test
+```
+
+**Note** that if you apply new changes to the deployment during the canary analysis, Flagger will restart the analysis.
+
+During the analysis the canary’s progress can be monitored with Grafana. The Istio dashboard URL is 
+http://localhost:3000/d/flagger-istio/istio-canary?refresh=10s&orgId=1&var-namespace=test&var-primary=podinfo-primary&var-canary=podinfo
+
+You can monitor all canaries with:
+
+```bash
+watch kubectl get canaries --all-namespaces
+
+NAMESPACE   NAME      STATUS        WEIGHT   LASTTRANSITIONTIME
+test        podinfo   Progressing   15       2019-05-16T14:05:07Z
+prod        frontend  Succeeded     0        2019-05-15T16:15:07Z
+prod        backend   Failed        0        2019-05-14T17:05:07Z
+```
+
+### Automated rollback
+
+During the canary analysis you can generate HTTP 500 errors and high latency to test if Flagger pauses the rollout.
+
+Create a tester pod and exec into it:
+
+```bash
+kubectl -n test run tester \
+--image=quay.io/stefanprodan/podinfo:1.2.1 \
+-- ./podinfo --port=9898
+
+kubectl -n test exec -it tester-xx-xx sh
+```
+
+Generate HTTP 500 errors:
+
+```bash
+watch curl http://podinfo-canary:9898/status/500
+```
+
+Generate latency:
+
+```bash
+watch curl http://podinfo-canary:9898/delay/1
+```
+
+When the number of failed checks reaches the canary analysis threshold, the traffic is routed back to the primary, 
+the canary is scaled to zero and the rollout is marked as failed.
+
+```text
+kubectl -n test describe canary/podinfo
+
+Status:
+  Canary Weight:         0
+  Failed Checks:         10
+  Phase:                 Failed
+Events:
+  Type     Reason  Age   From     Message
+  ----     ------  ----  ----     -------
+  Normal   Synced  3m    flagger  Starting canary deployment for podinfo.test
+  Normal   Synced  3m    flagger  Advance podinfo.test canary weight 5
+  Normal   Synced  3m    flagger  Advance podinfo.test canary weight 10
+  Normal   Synced  3m    flagger  Advance podinfo.test canary weight 15
+  Normal   Synced  3m    flagger  Halt podinfo.test advancement success rate 69.17% < 99%
+  Normal   Synced  2m    flagger  Halt podinfo.test advancement success rate 61.39% < 99%
+  Normal   Synced  2m    flagger  Halt podinfo.test advancement success rate 55.06% < 99%
+  Normal   Synced  2m    flagger  Halt podinfo.test advancement success rate 47.00% < 99%
+  Normal   Synced  2m    flagger  (combined from similar events): Halt podinfo.test advancement success rate 38.08% < 99%
+  Warning  Synced  1m    flagger  Rolling back podinfo.test failed checks threshold reached 10
+  Warning  Synced  1m    flagger  Canary failed! Scaling down podinfo.test
+```

docs/gitbook/usage/ab-testing.md

@@ -60,7 +60,6 @@ spec:
     # Istio gateways (optional)
     gateways:
     - public-gateway.istio-system.svc.cluster.local
-    - mesh
     # Istio virtual service host names (optional)
     hosts:
     - app.example.com

docs/gitbook/usage/gloo-progressive-delivery.md

@@ -0,0 +1,366 @@
+# NGNIX Ingress Controller Canary Deployments
+
+This guide shows you how to use the [Gloo](https://gloo.solo.io/) ingress controller and Flagger to automate canary deployments.
+
+![Flagger Gloo Ingress Controller](https://raw.githubusercontent.com/weaveworks/flagger/master/docs/diagrams/flagger-gloo-overview.png)
+
+### Prerequisites
+
+Flagger requires a Kubernetes cluster **v1.11** or newer and Gloo ingress **0.13.29** or newer.
+
+Install Gloo with Helm:
+
+```bash
+helm repo add gloo https://storage.googleapis.com/solo-public-helm
+
+helm upgrade -i gloo gloo/gloo \
+--namespace gloo-system
+```
+
+Install Flagger and the Prometheus add-on in the same namespace as Gloo:
+
+```bash
+helm repo add flagger https://flagger.app
+
+helm upgrade -i flagger flagger/flagger \
+--namespace gloo-system \
+--set prometheus.install=true \
+--set meshProvider=gloo
+```
+
+Optionally you can enable Slack notifications:
+
+```bash
+helm upgrade -i flagger flagger/flagger \
+--reuse-values \
+--namespace gloo-system \
+--set slack.url=https://hooks.slack.com/services/YOUR/SLACK/WEBHOOK \
+--set slack.channel=general \
+--set slack.user=flagger
+```
+
+### Bootstrap
+
+Flagger takes a Kubernetes deployment and optionally a horizontal pod autoscaler (HPA), 
+then creates a series of objects (Kubernetes deployments, ClusterIP services and Gloo upstream groups). 
+These objects expose the application outside the cluster and drive the canary analysis and promotion.
+
+Create a test namespace:
+
+```bash
+kubectl create ns test
+```
+
+Create a deployment and a horizontal pod autoscaler:
+
+```bash
+kubectl apply -f ${REPO}/artifacts/gloo/deployment.yaml
+kubectl apply -f ${REPO}/artifacts/gloo/hpa.yaml
+```
+
+Deploy the load testing service to generate traffic during the canary analysis:
+
+```bash
+helm upgrade -i flagger-loadtester flagger/loadtester \
+--namespace=test
+```
+
+Create an virtual service definition that references an upstream group that will be generated by Flagger
+(replace `app.example.com` with your own domain):
+
+```yaml
+apiVersion: gateway.solo.io/v1
+kind: VirtualService
+metadata:
+  name: podinfo
+  namespace: test
+spec:
+  virtualHost:
+    domains:
+      - 'app.example.com'
+    name: podinfo.test
+    routes:
+      - matcher:
+          prefix: /
+        routeAction:
+          upstreamGroup:
+            name: podinfo
+            namespace: test
+```
+
+Save the above resource as podinfo-virtualservice.yaml and then apply it:
+
+```bash
+kubectl apply -f ./podinfo-virtualservice.yaml
+```
+
+Create a canary custom resource (replace `app.example.com` with your own domain):
+
+```yaml
+apiVersion: flagger.app/v1alpha3
+kind: Canary
+metadata:
+  name: podinfo
+  namespace: test
+spec:
+  # deployment reference
+  targetRef:
+    apiVersion: apps/v1
+    kind: Deployment
+    name: podinfo
+  # HPA reference (optional)
+  autoscalerRef:
+    apiVersion: autoscaling/v2beta1
+    kind: HorizontalPodAutoscaler
+    name: podinfo
+  # the maximum time in seconds for the canary deployment
+  # to make progress before it is rollback (default 600s)
+  progressDeadlineSeconds: 60
+  service:
+    # container port
+    port: 9898
+  canaryAnalysis:
+    # schedule interval (default 60s)
+    interval: 10s
+    # max number of failed metric checks before rollback
+    threshold: 5
+    # max traffic percentage routed to canary
+    # percentage (0-100)
+    maxWeight: 50
+    # canary increment step
+    # percentage (0-100)
+    stepWeight: 5
+    # Gloo Prometheus checks
+    metrics:
+    - name: request-success-rate
+      # minimum req success rate (non 5xx responses)
+      # percentage (0-100)
+      threshold: 99
+      interval: 1m
+    - name: request-duration
+      # maximum req duration P99
+      # milliseconds
+      threshold: 500
+      interval: 30s
+    # load testing (optional)
+    webhooks:
+      - name: load-test
+        url: http://flagger-loadtester.test/
+        timeout: 5s
+        metadata:
+          type: cmd
+          cmd: "hey -z 1m -q 10 -c 2 http://app.example.com/"
+```
+
+Save the above resource as podinfo-canary.yaml and then apply it:
+
+```bash
+kubectl apply -f ./podinfo-canary.yaml
+```
+
+After a couple of seconds Flagger will create the canary objects:
+
+```bash
+# applied 
+deployment.apps/podinfo
+horizontalpodautoscaler.autoscaling/podinfo
+virtualservices.gateway.solo.io/podinfo
+canary.flagger.app/podinfo
+
+# generated 
+deployment.apps/podinfo-primary
+horizontalpodautoscaler.autoscaling/podinfo-primary
+service/podinfo
+service/podinfo-canary
+service/podinfo-primary
+upstreamgroups.gloo.solo.io/podinfo
+```
+
+When the bootstrap finishes Flagger will set the canary status to initialized:
+
+```bash
+kubectl -n test get canary podinfo
+
+NAME      STATUS        WEIGHT   LASTTRANSITIONTIME
+podinfo   Initialized   0        2019-05-17T08:09:51Z
+```
+
+### Automated canary promotion
+
+Flagger implements a control loop that gradually shifts traffic to the canary while measuring key performance indicators 
+like HTTP requests success rate, requests average duration and pod health.
+Based on analysis of the KPIs a canary is promoted or aborted, and the analysis result is published to Slack.
+
+![Flagger Canary Stages](https://raw.githubusercontent.com/weaveworks/flagger/master/docs/diagrams/flagger-canary-steps.png)
+
+Trigger a canary deployment by updating the container image:
+
+```bash
+kubectl -n test set image deployment/podinfo \
+podinfod=quay.io/stefanprodan/podinfo:1.4.1
+```
+
+Flagger detects that the deployment revision changed and starts a new rollout:
+
+```text
+kubectl -n test describe canary/podinfo
+
+Status:
+  Canary Weight:         0
+  Failed Checks:         0
+  Phase:                 Succeeded
+Events:
+  Type     Reason  Age   From     Message
+  ----     ------  ----  ----     -------
+  Normal   Synced  3m    flagger  New revision detected podinfo.test
+  Normal   Synced  3m    flagger  Scaling up podinfo.test
+  Warning  Synced  3m    flagger  Waiting for podinfo.test rollout to finish: 0 of 1 updated replicas are available
+  Normal   Synced  3m    flagger  Advance podinfo.test canary weight 5
+  Normal   Synced  3m    flagger  Advance podinfo.test canary weight 10
+  Normal   Synced  3m    flagger  Advance podinfo.test canary weight 15
+  Normal   Synced  2m    flagger  Advance podinfo.test canary weight 20
+  Normal   Synced  2m    flagger  Advance podinfo.test canary weight 25
+  Normal   Synced  1m    flagger  Advance podinfo.test canary weight 30
+  Normal   Synced  1m    flagger  Advance podinfo.test canary weight 35
+  Normal   Synced  55s   flagger  Advance podinfo.test canary weight 40
+  Normal   Synced  45s   flagger  Advance podinfo.test canary weight 45
+  Normal   Synced  35s   flagger  Advance podinfo.test canary weight 50
+  Normal   Synced  25s   flagger  Copying podinfo.test template spec to podinfo-primary.test
+  Warning  Synced  15s   flagger  Waiting for podinfo-primary.test rollout to finish: 1 of 2 updated replicas are available
+  Normal   Synced  5s    flagger  Promotion completed! Scaling down podinfo.test
+```
+
+**Note** that if you apply new changes to the deployment during the canary analysis, Flagger will restart the analysis.
+
+You can monitor all canaries with:
+
+```bash
+watch kubectl get canaries --all-namespaces
+
+NAMESPACE   NAME      STATUS        WEIGHT   LASTTRANSITIONTIME
+test        podinfo   Progressing   15       2019-05-17T14:05:07Z
+prod        frontend  Succeeded     0        2019-05-17T16:15:07Z
+prod        backend   Failed        0        2019-05-17T17:05:07Z
+```
+
+### Automated rollback
+
+During the canary analysis you can generate HTTP 500 errors and high latency to test if Flagger pauses and rolls back the faulted version.
+
+Trigger another canary deployment:
+
+```bash
+kubectl -n test set image deployment/podinfo \
+podinfod=quay.io/stefanprodan/podinfo:1.4.2
+```
+
+Generate HTTP 500 errors:
+
+```bash
+watch curl http://app.example.com/status/500
+```
+
+Generate high latency:
+
+```bash
+watch curl http://app.example.com/delay/2
+```
+
+When the number of failed checks reaches the canary analysis threshold, the traffic is routed back to the primary, 
+the canary is scaled to zero and the rollout is marked as failed.
+
+```text
+kubectl -n test describe canary/podinfo
+
+Status:
+  Canary Weight:         0
+  Failed Checks:         10
+  Phase:                 Failed
+Events:
+  Type     Reason  Age   From     Message
+  ----     ------  ----  ----     -------
+  Normal   Synced  3m    flagger  Starting canary deployment for podinfo.test
+  Normal   Synced  3m    flagger  Advance podinfo.test canary weight 5
+  Normal   Synced  3m    flagger  Advance podinfo.test canary weight 10
+  Normal   Synced  3m    flagger  Advance podinfo.test canary weight 15
+  Normal   Synced  3m    flagger  Halt podinfo.test advancement success rate 69.17% < 99%
+  Normal   Synced  2m    flagger  Halt podinfo.test advancement success rate 61.39% < 99%
+  Normal   Synced  2m    flagger  Halt podinfo.test advancement success rate 55.06% < 99%
+  Normal   Synced  2m    flagger  Halt podinfo.test advancement success rate 47.00% < 99%
+  Normal   Synced  2m    flagger  (combined from similar events): Halt podinfo.test advancement success rate 38.08% < 99%
+  Warning  Synced  1m    flagger  Rolling back podinfo.test failed checks threshold reached 10
+  Warning  Synced  1m    flagger  Canary failed! Scaling down podinfo.test
+```
+
+### Custom metrics
+
+The canary analysis can be extended with Prometheus queries.
+
+The demo app is instrumented with Prometheus so you can create a custom check that will use the HTTP request duration 
+histogram to validate the canary. 
+
+Edit the canary analysis and add the following metric:
+
+```yaml
+  canaryAnalysis:
+    metrics:
+    - name: "404s percentage"
+      threshold: 5
+      query: |
+        100 - sum(
+            rate(
+                http_request_duration_seconds_count{
+                  kubernetes_namespace="test",
+                  kubernetes_pod_name=~"podinfo-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"
+                  status!="404"
+                }[1m]
+            )
+        )
+        /
+        sum(
+            rate(
+                http_request_duration_seconds_count{
+                  kubernetes_namespace="test",
+                  kubernetes_pod_name=~"podinfo-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"
+                }[1m]
+            )
+        ) * 100
+```
+
+The above configuration validates the canary by checking if the HTTP 404 req/sec percentage is below 5 
+percent of the total traffic. If the 404s rate reaches the 5% threshold, then the canary fails.
+
+Trigger a canary deployment by updating the container image:
+
+```bash
+kubectl -n test set image deployment/podinfo \
+podinfod=quay.io/stefanprodan/podinfo:1.4.3
+```
+
+Generate 404s:
+
+```bash
+watch curl http://app.example.com/status/400
+```
+
+Watch Flagger logs:
+
+```
+kubectl -n gloo-system logs deployment/flagger -f | jq .msg
+
+Starting canary deployment for podinfo.test
+Advance podinfo.test canary weight 5
+Advance podinfo.test canary weight 10
+Advance podinfo.test canary weight 15
+Halt podinfo.test advancement 404s percentage 6.20 > 5
+Halt podinfo.test advancement 404s percentage 6.45 > 5
+Halt podinfo.test advancement 404s percentage 7.60 > 5
+Halt podinfo.test advancement 404s percentage 8.69 > 5
+Halt podinfo.test advancement 404s percentage 9.70 > 5
+Rolling back podinfo.test failed checks threshold reached 5
+Canary failed! Scaling down podinfo.test
+```
+
+If you have Slack configured, Flagger will send a notification with the reason why the canary failed.
+
+

docs/gitbook/usage/progressive-delivery.md

@@ -54,7 +54,6 @@ spec:
     # Istio gateways (optional)
     gateways:
     - public-gateway.istio-system.svc.cluster.local
-    - mesh
     # Istio virtual service host names (optional)
     hosts:
     - app.example.com

hack/update-codegen.sh

@@ -23,5 +23,5 @@ CODEGEN_PKG=${CODEGEN_PKG:-$(cd ${SCRIPT_ROOT}; ls -d -1 ./vendor/k8s.io/code-ge
 
 ${CODEGEN_PKG}/generate-groups.sh "deepcopy,client,informer,lister" \
   github.com/weaveworks/flagger/pkg/client github.com/weaveworks/flagger/pkg/apis \
-  "appmesh:v1beta1 istio:v1alpha3 flagger:v1alpha3" \
+  "appmesh:v1beta1 istio:v1alpha3 flagger:v1alpha3 smi:v1alpha1" \
   --go-header-file ${SCRIPT_ROOT}/hack/boilerplate.go.txt

pkg/apis/smi/register.go

@@ -0,0 +1,5 @@
+package smi
+
+const (
+	GroupName = "split.smi-spec.io"
+)

pkg/apis/smi/v1alpha1/doc.go

@@ -0,0 +1,4 @@
+// +k8s:deepcopy-gen=package
+// +groupName=split.smi-spec.io
+
+package v1alpha1

pkg/apis/smi/v1alpha1/register.go

@@ -0,0 +1,48 @@
+package v1alpha1
+
+import (
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/runtime"
+	"k8s.io/apimachinery/pkg/runtime/schema"
+
+	ts "github.com/weaveworks/flagger/pkg/apis/smi"
+)
+
+// SchemeGroupVersion is the identifier for the API which includes
+// the name of the group and the version of the API
+var SchemeGroupVersion = schema.GroupVersion{
+	Group:   ts.GroupName,
+	Version: "v1alpha1",
+}
+
+// Kind takes an unqualified kind and returns back a Group qualified GroupKind
+func Kind(kind string) schema.GroupKind {
+	return SchemeGroupVersion.WithKind(kind).GroupKind()
+}
+
+// Resource takes an unqualified resource and returns a Group qualified GroupResource
+func Resource(resource string) schema.GroupResource {
+	return SchemeGroupVersion.WithResource(resource).GroupResource()
+}
+
+var (
+	// SchemeBuilder collects functions that add things to a scheme. It's to allow
+	// code to compile without explicitly referencing generated types. You should
+	// declare one in each package that will have generated deep copy or conversion
+	// functions.
+	SchemeBuilder = runtime.NewSchemeBuilder(addKnownTypes)
+
+	// AddToScheme applies all the stored functions to the scheme. A non-nil error
+	// indicates that one function failed and the attempt was abandoned.
+	AddToScheme = SchemeBuilder.AddToScheme
+)
+
+// Adds the list of known types to Scheme.
+func addKnownTypes(scheme *runtime.Scheme) error {
+	scheme.AddKnownTypes(SchemeGroupVersion,
+		&TrafficSplit{},
+		&TrafficSplitList{},
+	)
+	metav1.AddToGroupVersion(scheme, SchemeGroupVersion)
+	return nil
+}

pkg/apis/smi/v1alpha1/traffic_split.go

@@ -0,0 +1,56 @@
+package v1alpha1
+
+import (
+	"k8s.io/apimachinery/pkg/api/resource"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+)
+
+// +genclient
+// +genclient:noStatus
+// +k8s:deepcopy-gen:interfaces=k8s.io/apimachinery/pkg/runtime.Object
+
+// TrafficSplit allows users to incrementally direct percentages of traffic
+// between various services. It will be used by clients such as ingress
+// controllers or service mesh sidecars to split the outgoing traffic to
+// different destinations.
+type TrafficSplit struct {
+	metav1.TypeMeta `json:",inline"`
+	// Standard object's metadata.
+	// More info: https://git.k8s.io/community/contributors/devel/api-conventions.md#metadata
+	// +optional
+	metav1.ObjectMeta `json:"metadata,omitempty" protobuf:"bytes,1,opt,name=metadata"`
+
+	// Specification of the desired behavior of the traffic split.
+	// More info: https://git.k8s.io/community/contributors/devel/api-conventions.md#spec-and-status
+	// +optional
+	Spec TrafficSplitSpec `json:"spec,omitempty" protobuf:"bytes,2,opt,name=spec"`
+
+	// Most recently observed status of the pod.
+	// This data may not be up to date.
+	// Populated by the system.
+	// Read-only.
+	// More info: https://git.k8s.io/community/contributors/devel/api-conventions.md#spec-and-status
+	// +optional
+	//Status Status `json:"status,omitempty" protobuf:"bytes,3,opt,name=status"`
+}
+
+// TrafficSplitSpec is the specification for a TrafficSplit
+type TrafficSplitSpec struct {
+	Service  string                `json:"service,omitempty"`
+	Backends []TrafficSplitBackend `json:"backends,omitempty"`
+}
+
+// TrafficSplitBackend defines a backend
+type TrafficSplitBackend struct {
+	Service string             `json:"service,omitempty"`
+	Weight  *resource.Quantity `json:"weight,omitempty"`
+}
+
+// +k8s:deepcopy-gen:interfaces=k8s.io/apimachinery/pkg/runtime.Object
+
+type TrafficSplitList struct {
+	metav1.TypeMeta `json:",inline"`
+	metav1.ListMeta `json:"metadata"`
+
+	Items []TrafficSplit `json:"items"`
+}

pkg/apis/smi/v1alpha1/zz_generated.deepcopy.go

@@ -0,0 +1,129 @@
+// +build !ignore_autogenerated
+
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by deepcopy-gen. DO NOT EDIT.
+
+package v1alpha1
+
+import (
+	runtime "k8s.io/apimachinery/pkg/runtime"
+)
+
+// DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.
+func (in *TrafficSplit) DeepCopyInto(out *TrafficSplit) {
+	*out = *in
+	out.TypeMeta = in.TypeMeta
+	in.ObjectMeta.DeepCopyInto(&out.ObjectMeta)
+	in.Spec.DeepCopyInto(&out.Spec)
+	return
+}
+
+// DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new TrafficSplit.
+func (in *TrafficSplit) DeepCopy() *TrafficSplit {
+	if in == nil {
+		return nil
+	}
+	out := new(TrafficSplit)
+	in.DeepCopyInto(out)
+	return out
+}
+
+// DeepCopyObject is an autogenerated deepcopy function, copying the receiver, creating a new runtime.Object.
+func (in *TrafficSplit) DeepCopyObject() runtime.Object {
+	if c := in.DeepCopy(); c != nil {
+		return c
+	}
+	return nil
+}
+
+// DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.
+func (in *TrafficSplitBackend) DeepCopyInto(out *TrafficSplitBackend) {
+	*out = *in
+	if in.Weight != nil {
+		in, out := &in.Weight, &out.Weight
+		x := (*in).DeepCopy()
+		*out = &x
+	}
+	return
+}
+
+// DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new TrafficSplitBackend.
+func (in *TrafficSplitBackend) DeepCopy() *TrafficSplitBackend {
+	if in == nil {
+		return nil
+	}
+	out := new(TrafficSplitBackend)
+	in.DeepCopyInto(out)
+	return out
+}
+
+// DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.
+func (in *TrafficSplitList) DeepCopyInto(out *TrafficSplitList) {
+	*out = *in
+	out.TypeMeta = in.TypeMeta
+	out.ListMeta = in.ListMeta
+	if in.Items != nil {
+		in, out := &in.Items, &out.Items
+		*out = make([]TrafficSplit, len(*in))
+		for i := range *in {
+			(*in)[i].DeepCopyInto(&(*out)[i])
+		}
+	}
+	return
+}
+
+// DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new TrafficSplitList.
+func (in *TrafficSplitList) DeepCopy() *TrafficSplitList {
+	if in == nil {
+		return nil
+	}
+	out := new(TrafficSplitList)
+	in.DeepCopyInto(out)
+	return out
+}
+
+// DeepCopyObject is an autogenerated deepcopy function, copying the receiver, creating a new runtime.Object.
+func (in *TrafficSplitList) DeepCopyObject() runtime.Object {
+	if c := in.DeepCopy(); c != nil {
+		return c
+	}
+	return nil
+}
+
+// DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.
+func (in *TrafficSplitSpec) DeepCopyInto(out *TrafficSplitSpec) {
+	*out = *in
+	if in.Backends != nil {
+		in, out := &in.Backends, &out.Backends
+		*out = make([]TrafficSplitBackend, len(*in))
+		for i := range *in {
+			(*in)[i].DeepCopyInto(&(*out)[i])
+		}
+	}
+	return
+}
+
+// DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new TrafficSplitSpec.
+func (in *TrafficSplitSpec) DeepCopy() *TrafficSplitSpec {
+	if in == nil {
+		return nil
+	}
+	out := new(TrafficSplitSpec)
+	in.DeepCopyInto(out)
+	return out
+}

pkg/client/clientset/versioned/clientset.go

@@ -22,6 +22,7 @@ import (
 	appmeshv1beta1 "github.com/weaveworks/flagger/pkg/client/clientset/versioned/typed/appmesh/v1beta1"
 	flaggerv1alpha3 "github.com/weaveworks/flagger/pkg/client/clientset/versioned/typed/flagger/v1alpha3"
 	networkingv1alpha3 "github.com/weaveworks/flagger/pkg/client/clientset/versioned/typed/istio/v1alpha3"
+	splitv1alpha1 "github.com/weaveworks/flagger/pkg/client/clientset/versioned/typed/smi/v1alpha1"
 	discovery "k8s.io/client-go/discovery"
 	rest "k8s.io/client-go/rest"
 	flowcontrol "k8s.io/client-go/util/flowcontrol"
@@ -38,6 +39,9 @@ type Interface interface {
 	NetworkingV1alpha3() networkingv1alpha3.NetworkingV1alpha3Interface
 	// Deprecated: please explicitly pick a version if possible.
 	Networking() networkingv1alpha3.NetworkingV1alpha3Interface
+	SplitV1alpha1() splitv1alpha1.SplitV1alpha1Interface
+	// Deprecated: please explicitly pick a version if possible.
+	Split() splitv1alpha1.SplitV1alpha1Interface
 }
 
 // Clientset contains the clients for groups. Each group has exactly one
@@ -47,6 +51,7 @@ type Clientset struct {
 	appmeshV1beta1     *appmeshv1beta1.AppmeshV1beta1Client
 	flaggerV1alpha3    *flaggerv1alpha3.FlaggerV1alpha3Client
 	networkingV1alpha3 *networkingv1alpha3.NetworkingV1alpha3Client
+	splitV1alpha1      *splitv1alpha1.SplitV1alpha1Client
 }
 
 // AppmeshV1beta1 retrieves the AppmeshV1beta1Client
@@ -82,6 +87,17 @@ func (c *Clientset) Networking() networkingv1alpha3.NetworkingV1alpha3Interface
 	return c.networkingV1alpha3
 }
 
+// SplitV1alpha1 retrieves the SplitV1alpha1Client
+func (c *Clientset) SplitV1alpha1() splitv1alpha1.SplitV1alpha1Interface {
+	return c.splitV1alpha1
+}
+
+// Deprecated: Split retrieves the default version of SplitClient.
+// Please explicitly pick a version.
+func (c *Clientset) Split() splitv1alpha1.SplitV1alpha1Interface {
+	return c.splitV1alpha1
+}
+
 // Discovery retrieves the DiscoveryClient
 func (c *Clientset) Discovery() discovery.DiscoveryInterface {
 	if c == nil {
@@ -110,6 +126,10 @@ func NewForConfig(c *rest.Config) (*Clientset, error) {
 	if err != nil {
 		return nil, err
 	}
+	cs.splitV1alpha1, err = splitv1alpha1.NewForConfig(&configShallowCopy)
+	if err != nil {
+		return nil, err
+	}
 
 	cs.DiscoveryClient, err = discovery.NewDiscoveryClientForConfig(&configShallowCopy)
 	if err != nil {
@@ -125,6 +145,7 @@ func NewForConfigOrDie(c *rest.Config) *Clientset {
 	cs.appmeshV1beta1 = appmeshv1beta1.NewForConfigOrDie(c)
 	cs.flaggerV1alpha3 = flaggerv1alpha3.NewForConfigOrDie(c)
 	cs.networkingV1alpha3 = networkingv1alpha3.NewForConfigOrDie(c)
+	cs.splitV1alpha1 = splitv1alpha1.NewForConfigOrDie(c)
 
 	cs.DiscoveryClient = discovery.NewDiscoveryClientForConfigOrDie(c)
 	return &cs
@@ -136,6 +157,7 @@ func New(c rest.Interface) *Clientset {
 	cs.appmeshV1beta1 = appmeshv1beta1.New(c)
 	cs.flaggerV1alpha3 = flaggerv1alpha3.New(c)
 	cs.networkingV1alpha3 = networkingv1alpha3.New(c)
+	cs.splitV1alpha1 = splitv1alpha1.New(c)
 
 	cs.DiscoveryClient = discovery.NewDiscoveryClient(c)
 	return &cs

pkg/client/clientset/versioned/fake/clientset_generated.go

@@ -26,6 +26,8 @@ import (
 	fakeflaggerv1alpha3 "github.com/weaveworks/flagger/pkg/client/clientset/versioned/typed/flagger/v1alpha3/fake"
 	networkingv1alpha3 "github.com/weaveworks/flagger/pkg/client/clientset/versioned/typed/istio/v1alpha3"
 	fakenetworkingv1alpha3 "github.com/weaveworks/flagger/pkg/client/clientset/versioned/typed/istio/v1alpha3/fake"
+	splitv1alpha1 "github.com/weaveworks/flagger/pkg/client/clientset/versioned/typed/smi/v1alpha1"
+	fakesplitv1alpha1 "github.com/weaveworks/flagger/pkg/client/clientset/versioned/typed/smi/v1alpha1/fake"
 	"k8s.io/apimachinery/pkg/runtime"
 	"k8s.io/apimachinery/pkg/watch"
 	"k8s.io/client-go/discovery"
@@ -104,3 +106,13 @@ func (c *Clientset) NetworkingV1alpha3() networkingv1alpha3.NetworkingV1alpha3In
 func (c *Clientset) Networking() networkingv1alpha3.NetworkingV1alpha3Interface {
 	return &fakenetworkingv1alpha3.FakeNetworkingV1alpha3{Fake: &c.Fake}
 }
+
+// SplitV1alpha1 retrieves the SplitV1alpha1Client
+func (c *Clientset) SplitV1alpha1() splitv1alpha1.SplitV1alpha1Interface {
+	return &fakesplitv1alpha1.FakeSplitV1alpha1{Fake: &c.Fake}
+}
+
+// Split retrieves the SplitV1alpha1Client
+func (c *Clientset) Split() splitv1alpha1.SplitV1alpha1Interface {
+	return &fakesplitv1alpha1.FakeSplitV1alpha1{Fake: &c.Fake}
+}

pkg/client/clientset/versioned/fake/register.go

@@ -22,6 +22,7 @@ import (
 	appmeshv1beta1 "github.com/weaveworks/flagger/pkg/apis/appmesh/v1beta1"
 	flaggerv1alpha3 "github.com/weaveworks/flagger/pkg/apis/flagger/v1alpha3"
 	networkingv1alpha3 "github.com/weaveworks/flagger/pkg/apis/istio/v1alpha3"
+	splitv1alpha1 "github.com/weaveworks/flagger/pkg/apis/smi/v1alpha1"
 	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	runtime "k8s.io/apimachinery/pkg/runtime"
 	schema "k8s.io/apimachinery/pkg/runtime/schema"
@@ -36,6 +37,7 @@ var localSchemeBuilder = runtime.SchemeBuilder{
 	appmeshv1beta1.AddToScheme,
 	flaggerv1alpha3.AddToScheme,
 	networkingv1alpha3.AddToScheme,
+	splitv1alpha1.AddToScheme,
 }
 
 // AddToScheme adds all types of this clientset into the given scheme. This allows composition

pkg/client/clientset/versioned/scheme/register.go

@@ -22,6 +22,7 @@ import (
 	appmeshv1beta1 "github.com/weaveworks/flagger/pkg/apis/appmesh/v1beta1"
 	flaggerv1alpha3 "github.com/weaveworks/flagger/pkg/apis/flagger/v1alpha3"
 	networkingv1alpha3 "github.com/weaveworks/flagger/pkg/apis/istio/v1alpha3"
+	splitv1alpha1 "github.com/weaveworks/flagger/pkg/apis/smi/v1alpha1"
 	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	runtime "k8s.io/apimachinery/pkg/runtime"
 	schema "k8s.io/apimachinery/pkg/runtime/schema"
@@ -36,6 +37,7 @@ var localSchemeBuilder = runtime.SchemeBuilder{
 	appmeshv1beta1.AddToScheme,
 	flaggerv1alpha3.AddToScheme,
 	networkingv1alpha3.AddToScheme,
+	splitv1alpha1.AddToScheme,
 }
 
 // AddToScheme adds all types of this clientset into the given scheme. This allows composition

pkg/client/clientset/versioned/typed/smi/v1alpha1/doc.go

@@ -0,0 +1,20 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by client-gen. DO NOT EDIT.
+
+// This package has the automatically generated typed clients.
+package v1alpha1

pkg/client/clientset/versioned/typed/smi/v1alpha1/fake/doc.go

@@ -0,0 +1,20 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by client-gen. DO NOT EDIT.
+
+// Package fake has the automatically generated clients.
+package fake

pkg/client/clientset/versioned/typed/smi/v1alpha1/fake/fake_smi_client.go

@@ -0,0 +1,40 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by client-gen. DO NOT EDIT.
+
+package fake
+
+import (
+	v1alpha1 "github.com/weaveworks/flagger/pkg/client/clientset/versioned/typed/smi/v1alpha1"
+	rest "k8s.io/client-go/rest"
+	testing "k8s.io/client-go/testing"
+)
+
+type FakeSplitV1alpha1 struct {
+	*testing.Fake
+}
+
+func (c *FakeSplitV1alpha1) TrafficSplits(namespace string) v1alpha1.TrafficSplitInterface {
+	return &FakeTrafficSplits{c, namespace}
+}
+
+// RESTClient returns a RESTClient that is used to communicate
+// with API server by this client implementation.
+func (c *FakeSplitV1alpha1) RESTClient() rest.Interface {
+	var ret *rest.RESTClient
+	return ret
+}

pkg/client/clientset/versioned/typed/smi/v1alpha1/fake/fake_trafficsplit.go

@@ -0,0 +1,128 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by client-gen. DO NOT EDIT.
+
+package fake
+
+import (
+	v1alpha1 "github.com/weaveworks/flagger/pkg/apis/smi/v1alpha1"
+	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	labels "k8s.io/apimachinery/pkg/labels"
+	schema "k8s.io/apimachinery/pkg/runtime/schema"
+	types "k8s.io/apimachinery/pkg/types"
+	watch "k8s.io/apimachinery/pkg/watch"
+	testing "k8s.io/client-go/testing"
+)
+
+// FakeTrafficSplits implements TrafficSplitInterface
+type FakeTrafficSplits struct {
+	Fake *FakeSplitV1alpha1
+	ns   string
+}
+
+var trafficsplitsResource = schema.GroupVersionResource{Group: "split.smi-spec.io", Version: "v1alpha1", Resource: "trafficsplits"}
+
+var trafficsplitsKind = schema.GroupVersionKind{Group: "split.smi-spec.io", Version: "v1alpha1", Kind: "TrafficSplit"}
+
+// Get takes name of the trafficSplit, and returns the corresponding trafficSplit object, and an error if there is any.
+func (c *FakeTrafficSplits) Get(name string, options v1.GetOptions) (result *v1alpha1.TrafficSplit, err error) {
+	obj, err := c.Fake.
+		Invokes(testing.NewGetAction(trafficsplitsResource, c.ns, name), &v1alpha1.TrafficSplit{})
+
+	if obj == nil {
+		return nil, err
+	}
+	return obj.(*v1alpha1.TrafficSplit), err
+}
+
+// List takes label and field selectors, and returns the list of TrafficSplits that match those selectors.
+func (c *FakeTrafficSplits) List(opts v1.ListOptions) (result *v1alpha1.TrafficSplitList, err error) {
+	obj, err := c.Fake.
+		Invokes(testing.NewListAction(trafficsplitsResource, trafficsplitsKind, c.ns, opts), &v1alpha1.TrafficSplitList{})
+
+	if obj == nil {
+		return nil, err
+	}
+
+	label, _, _ := testing.ExtractFromListOptions(opts)
+	if label == nil {
+		label = labels.Everything()
+	}
+	list := &v1alpha1.TrafficSplitList{ListMeta: obj.(*v1alpha1.TrafficSplitList).ListMeta}
+	for _, item := range obj.(*v1alpha1.TrafficSplitList).Items {
+		if label.Matches(labels.Set(item.Labels)) {
+			list.Items = append(list.Items, item)
+		}
+	}
+	return list, err
+}
+
+// Watch returns a watch.Interface that watches the requested trafficSplits.
+func (c *FakeTrafficSplits) Watch(opts v1.ListOptions) (watch.Interface, error) {
+	return c.Fake.
+		InvokesWatch(testing.NewWatchAction(trafficsplitsResource, c.ns, opts))
+
+}
+
+// Create takes the representation of a trafficSplit and creates it.  Returns the server's representation of the trafficSplit, and an error, if there is any.
+func (c *FakeTrafficSplits) Create(trafficSplit *v1alpha1.TrafficSplit) (result *v1alpha1.TrafficSplit, err error) {
+	obj, err := c.Fake.
+		Invokes(testing.NewCreateAction(trafficsplitsResource, c.ns, trafficSplit), &v1alpha1.TrafficSplit{})
+
+	if obj == nil {
+		return nil, err
+	}
+	return obj.(*v1alpha1.TrafficSplit), err
+}
+
+// Update takes the representation of a trafficSplit and updates it. Returns the server's representation of the trafficSplit, and an error, if there is any.
+func (c *FakeTrafficSplits) Update(trafficSplit *v1alpha1.TrafficSplit) (result *v1alpha1.TrafficSplit, err error) {
+	obj, err := c.Fake.
+		Invokes(testing.NewUpdateAction(trafficsplitsResource, c.ns, trafficSplit), &v1alpha1.TrafficSplit{})
+
+	if obj == nil {
+		return nil, err
+	}
+	return obj.(*v1alpha1.TrafficSplit), err
+}
+
+// Delete takes name of the trafficSplit and deletes it. Returns an error if one occurs.
+func (c *FakeTrafficSplits) Delete(name string, options *v1.DeleteOptions) error {
+	_, err := c.Fake.
+		Invokes(testing.NewDeleteAction(trafficsplitsResource, c.ns, name), &v1alpha1.TrafficSplit{})
+
+	return err
+}
+
+// DeleteCollection deletes a collection of objects.
+func (c *FakeTrafficSplits) DeleteCollection(options *v1.DeleteOptions, listOptions v1.ListOptions) error {
+	action := testing.NewDeleteCollectionAction(trafficsplitsResource, c.ns, listOptions)
+
+	_, err := c.Fake.Invokes(action, &v1alpha1.TrafficSplitList{})
+	return err
+}
+
+// Patch applies the patch and returns the patched trafficSplit.
+func (c *FakeTrafficSplits) Patch(name string, pt types.PatchType, data []byte, subresources ...string) (result *v1alpha1.TrafficSplit, err error) {
+	obj, err := c.Fake.
+		Invokes(testing.NewPatchSubresourceAction(trafficsplitsResource, c.ns, name, pt, data, subresources...), &v1alpha1.TrafficSplit{})
+
+	if obj == nil {
+		return nil, err
+	}
+	return obj.(*v1alpha1.TrafficSplit), err
+}

pkg/client/clientset/versioned/typed/smi/v1alpha1/generated_expansion.go

@@ -0,0 +1,21 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by client-gen. DO NOT EDIT.
+
+package v1alpha1
+
+type TrafficSplitExpansion interface{}

pkg/client/clientset/versioned/typed/smi/v1alpha1/smi_client.go

@@ -0,0 +1,90 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by client-gen. DO NOT EDIT.
+
+package v1alpha1
+
+import (
+	v1alpha1 "github.com/weaveworks/flagger/pkg/apis/smi/v1alpha1"
+	"github.com/weaveworks/flagger/pkg/client/clientset/versioned/scheme"
+	serializer "k8s.io/apimachinery/pkg/runtime/serializer"
+	rest "k8s.io/client-go/rest"
+)
+
+type SplitV1alpha1Interface interface {
+	RESTClient() rest.Interface
+	TrafficSplitsGetter
+}
+
+// SplitV1alpha1Client is used to interact with features provided by the split.smi-spec.io group.
+type SplitV1alpha1Client struct {
+	restClient rest.Interface
+}
+
+func (c *SplitV1alpha1Client) TrafficSplits(namespace string) TrafficSplitInterface {
+	return newTrafficSplits(c, namespace)
+}
+
+// NewForConfig creates a new SplitV1alpha1Client for the given config.
+func NewForConfig(c *rest.Config) (*SplitV1alpha1Client, error) {
+	config := *c
+	if err := setConfigDefaults(&config); err != nil {
+		return nil, err
+	}
+	client, err := rest.RESTClientFor(&config)
+	if err != nil {
+		return nil, err
+	}
+	return &SplitV1alpha1Client{client}, nil
+}
+
+// NewForConfigOrDie creates a new SplitV1alpha1Client for the given config and
+// panics if there is an error in the config.
+func NewForConfigOrDie(c *rest.Config) *SplitV1alpha1Client {
+	client, err := NewForConfig(c)
+	if err != nil {
+		panic(err)
+	}
+	return client
+}
+
+// New creates a new SplitV1alpha1Client for the given RESTClient.
+func New(c rest.Interface) *SplitV1alpha1Client {
+	return &SplitV1alpha1Client{c}
+}
+
+func setConfigDefaults(config *rest.Config) error {
+	gv := v1alpha1.SchemeGroupVersion
+	config.GroupVersion = &gv
+	config.APIPath = "/apis"
+	config.NegotiatedSerializer = serializer.DirectCodecFactory{CodecFactory: scheme.Codecs}
+
+	if config.UserAgent == "" {
+		config.UserAgent = rest.DefaultKubernetesUserAgent()
+	}
+
+	return nil
+}
+
+// RESTClient returns a RESTClient that is used to communicate
+// with API server by this client implementation.
+func (c *SplitV1alpha1Client) RESTClient() rest.Interface {
+	if c == nil {
+		return nil
+	}
+	return c.restClient
+}

pkg/client/clientset/versioned/typed/smi/v1alpha1/trafficsplit.go

@@ -0,0 +1,174 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by client-gen. DO NOT EDIT.
+
+package v1alpha1
+
+import (
+	"time"
+
+	v1alpha1 "github.com/weaveworks/flagger/pkg/apis/smi/v1alpha1"
+	scheme "github.com/weaveworks/flagger/pkg/client/clientset/versioned/scheme"
+	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	types "k8s.io/apimachinery/pkg/types"
+	watch "k8s.io/apimachinery/pkg/watch"
+	rest "k8s.io/client-go/rest"
+)
+
+// TrafficSplitsGetter has a method to return a TrafficSplitInterface.
+// A group's client should implement this interface.
+type TrafficSplitsGetter interface {
+	TrafficSplits(namespace string) TrafficSplitInterface
+}
+
+// TrafficSplitInterface has methods to work with TrafficSplit resources.
+type TrafficSplitInterface interface {
+	Create(*v1alpha1.TrafficSplit) (*v1alpha1.TrafficSplit, error)
+	Update(*v1alpha1.TrafficSplit) (*v1alpha1.TrafficSplit, error)
+	Delete(name string, options *v1.DeleteOptions) error
+	DeleteCollection(options *v1.DeleteOptions, listOptions v1.ListOptions) error
+	Get(name string, options v1.GetOptions) (*v1alpha1.TrafficSplit, error)
+	List(opts v1.ListOptions) (*v1alpha1.TrafficSplitList, error)
+	Watch(opts v1.ListOptions) (watch.Interface, error)
+	Patch(name string, pt types.PatchType, data []byte, subresources ...string) (result *v1alpha1.TrafficSplit, err error)
+	TrafficSplitExpansion
+}
+
+// trafficSplits implements TrafficSplitInterface
+type trafficSplits struct {
+	client rest.Interface
+	ns     string
+}
+
+// newTrafficSplits returns a TrafficSplits
+func newTrafficSplits(c *SplitV1alpha1Client, namespace string) *trafficSplits {
+	return &trafficSplits{
+		client: c.RESTClient(),
+		ns:     namespace,
+	}
+}
+
+// Get takes name of the trafficSplit, and returns the corresponding trafficSplit object, and an error if there is any.
+func (c *trafficSplits) Get(name string, options v1.GetOptions) (result *v1alpha1.TrafficSplit, err error) {
+	result = &v1alpha1.TrafficSplit{}
+	err = c.client.Get().
+		Namespace(c.ns).
+		Resource("trafficsplits").
+		Name(name).
+		VersionedParams(&options, scheme.ParameterCodec).
+		Do().
+		Into(result)
+	return
+}
+
+// List takes label and field selectors, and returns the list of TrafficSplits that match those selectors.
+func (c *trafficSplits) List(opts v1.ListOptions) (result *v1alpha1.TrafficSplitList, err error) {
+	var timeout time.Duration
+	if opts.TimeoutSeconds != nil {
+		timeout = time.Duration(*opts.TimeoutSeconds) * time.Second
+	}
+	result = &v1alpha1.TrafficSplitList{}
+	err = c.client.Get().
+		Namespace(c.ns).
+		Resource("trafficsplits").
+		VersionedParams(&opts, scheme.ParameterCodec).
+		Timeout(timeout).
+		Do().
+		Into(result)
+	return
+}
+
+// Watch returns a watch.Interface that watches the requested trafficSplits.
+func (c *trafficSplits) Watch(opts v1.ListOptions) (watch.Interface, error) {
+	var timeout time.Duration
+	if opts.TimeoutSeconds != nil {
+		timeout = time.Duration(*opts.TimeoutSeconds) * time.Second
+	}
+	opts.Watch = true
+	return c.client.Get().
+		Namespace(c.ns).
+		Resource("trafficsplits").
+		VersionedParams(&opts, scheme.ParameterCodec).
+		Timeout(timeout).
+		Watch()
+}
+
+// Create takes the representation of a trafficSplit and creates it.  Returns the server's representation of the trafficSplit, and an error, if there is any.
+func (c *trafficSplits) Create(trafficSplit *v1alpha1.TrafficSplit) (result *v1alpha1.TrafficSplit, err error) {
+	result = &v1alpha1.TrafficSplit{}
+	err = c.client.Post().
+		Namespace(c.ns).
+		Resource("trafficsplits").
+		Body(trafficSplit).
+		Do().
+		Into(result)
+	return
+}
+
+// Update takes the representation of a trafficSplit and updates it. Returns the server's representation of the trafficSplit, and an error, if there is any.
+func (c *trafficSplits) Update(trafficSplit *v1alpha1.TrafficSplit) (result *v1alpha1.TrafficSplit, err error) {
+	result = &v1alpha1.TrafficSplit{}
+	err = c.client.Put().
+		Namespace(c.ns).
+		Resource("trafficsplits").
+		Name(trafficSplit.Name).
+		Body(trafficSplit).
+		Do().
+		Into(result)
+	return
+}
+
+// Delete takes name of the trafficSplit and deletes it. Returns an error if one occurs.
+func (c *trafficSplits) Delete(name string, options *v1.DeleteOptions) error {
+	return c.client.Delete().
+		Namespace(c.ns).
+		Resource("trafficsplits").
+		Name(name).
+		Body(options).
+		Do().
+		Error()
+}
+
+// DeleteCollection deletes a collection of objects.
+func (c *trafficSplits) DeleteCollection(options *v1.DeleteOptions, listOptions v1.ListOptions) error {
+	var timeout time.Duration
+	if listOptions.TimeoutSeconds != nil {
+		timeout = time.Duration(*listOptions.TimeoutSeconds) * time.Second
+	}
+	return c.client.Delete().
+		Namespace(c.ns).
+		Resource("trafficsplits").
+		VersionedParams(&listOptions, scheme.ParameterCodec).
+		Timeout(timeout).
+		Body(options).
+		Do().
+		Error()
+}
+
+// Patch applies the patch and returns the patched trafficSplit.
+func (c *trafficSplits) Patch(name string, pt types.PatchType, data []byte, subresources ...string) (result *v1alpha1.TrafficSplit, err error) {
+	result = &v1alpha1.TrafficSplit{}
+	err = c.client.Patch(pt).
+		Namespace(c.ns).
+		Resource("trafficsplits").
+		SubResource(subresources...).
+		Name(name).
+		Body(data).
+		Do().
+		Into(result)
+	return
+}

pkg/client/informers/externalversions/factory.go

@@ -28,6 +28,7 @@ import (
 	flagger "github.com/weaveworks/flagger/pkg/client/informers/externalversions/flagger"
 	internalinterfaces "github.com/weaveworks/flagger/pkg/client/informers/externalversions/internalinterfaces"
 	istio "github.com/weaveworks/flagger/pkg/client/informers/externalversions/istio"
+	smi "github.com/weaveworks/flagger/pkg/client/informers/externalversions/smi"
 	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	runtime "k8s.io/apimachinery/pkg/runtime"
 	schema "k8s.io/apimachinery/pkg/runtime/schema"
@@ -177,6 +178,7 @@ type SharedInformerFactory interface {
 	Appmesh() appmesh.Interface
 	Flagger() flagger.Interface
 	Networking() istio.Interface
+	Split() smi.Interface
 }
 
 func (f *sharedInformerFactory) Appmesh() appmesh.Interface {
@@ -190,3 +192,7 @@ func (f *sharedInformerFactory) Flagger() flagger.Interface {
 func (f *sharedInformerFactory) Networking() istio.Interface {
 	return istio.New(f, f.namespace, f.tweakListOptions)
 }
+
+func (f *sharedInformerFactory) Split() smi.Interface {
+	return smi.New(f, f.namespace, f.tweakListOptions)
+}

pkg/client/informers/externalversions/generic.go

@@ -24,6 +24,7 @@ import (
 	v1beta1 "github.com/weaveworks/flagger/pkg/apis/appmesh/v1beta1"
 	v1alpha3 "github.com/weaveworks/flagger/pkg/apis/flagger/v1alpha3"
 	istiov1alpha3 "github.com/weaveworks/flagger/pkg/apis/istio/v1alpha3"
+	v1alpha1 "github.com/weaveworks/flagger/pkg/apis/smi/v1alpha1"
 	schema "k8s.io/apimachinery/pkg/runtime/schema"
 	cache "k8s.io/client-go/tools/cache"
 )
@@ -70,6 +71,10 @@ func (f *sharedInformerFactory) ForResource(resource schema.GroupVersionResource
 	case istiov1alpha3.SchemeGroupVersion.WithResource("virtualservices"):
 		return &genericInformer{resource: resource.GroupResource(), informer: f.Networking().V1alpha3().VirtualServices().Informer()}, nil
 
+		// Group=split.smi-spec.io, Version=v1alpha1
+	case v1alpha1.SchemeGroupVersion.WithResource("trafficsplits"):
+		return &genericInformer{resource: resource.GroupResource(), informer: f.Split().V1alpha1().TrafficSplits().Informer()}, nil
+
 	}
 
 	return nil, fmt.Errorf("no informer found for %v", resource)

pkg/client/informers/externalversions/smi/interface.go

@@ -0,0 +1,46 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by informer-gen. DO NOT EDIT.
+
+package split
+
+import (
+	internalinterfaces "github.com/weaveworks/flagger/pkg/client/informers/externalversions/internalinterfaces"
+	v1alpha1 "github.com/weaveworks/flagger/pkg/client/informers/externalversions/smi/v1alpha1"
+)
+
+// Interface provides access to each of this group's versions.
+type Interface interface {
+	// V1alpha1 provides access to shared informers for resources in V1alpha1.
+	V1alpha1() v1alpha1.Interface
+}
+
+type group struct {
+	factory          internalinterfaces.SharedInformerFactory
+	namespace        string
+	tweakListOptions internalinterfaces.TweakListOptionsFunc
+}
+
+// New returns a new Interface.
+func New(f internalinterfaces.SharedInformerFactory, namespace string, tweakListOptions internalinterfaces.TweakListOptionsFunc) Interface {
+	return &group{factory: f, namespace: namespace, tweakListOptions: tweakListOptions}
+}
+
+// V1alpha1 returns a new v1alpha1.Interface.
+func (g *group) V1alpha1() v1alpha1.Interface {
+	return v1alpha1.New(g.factory, g.namespace, g.tweakListOptions)
+}

pkg/client/informers/externalversions/smi/v1alpha1/interface.go

@@ -0,0 +1,45 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by informer-gen. DO NOT EDIT.
+
+package v1alpha1
+
+import (
+	internalinterfaces "github.com/weaveworks/flagger/pkg/client/informers/externalversions/internalinterfaces"
+)
+
+// Interface provides access to all the informers in this group version.
+type Interface interface {
+	// TrafficSplits returns a TrafficSplitInformer.
+	TrafficSplits() TrafficSplitInformer
+}
+
+type version struct {
+	factory          internalinterfaces.SharedInformerFactory
+	namespace        string
+	tweakListOptions internalinterfaces.TweakListOptionsFunc
+}
+
+// New returns a new Interface.
+func New(f internalinterfaces.SharedInformerFactory, namespace string, tweakListOptions internalinterfaces.TweakListOptionsFunc) Interface {
+	return &version{factory: f, namespace: namespace, tweakListOptions: tweakListOptions}
+}
+
+// TrafficSplits returns a TrafficSplitInformer.
+func (v *version) TrafficSplits() TrafficSplitInformer {
+	return &trafficSplitInformer{factory: v.factory, namespace: v.namespace, tweakListOptions: v.tweakListOptions}
+}

pkg/client/informers/externalversions/smi/v1alpha1/trafficsplit.go

@@ -0,0 +1,89 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by informer-gen. DO NOT EDIT.
+
+package v1alpha1
+
+import (
+	time "time"
+
+	smiv1alpha1 "github.com/weaveworks/flagger/pkg/apis/smi/v1alpha1"
+	versioned "github.com/weaveworks/flagger/pkg/client/clientset/versioned"
+	internalinterfaces "github.com/weaveworks/flagger/pkg/client/informers/externalversions/internalinterfaces"
+	v1alpha1 "github.com/weaveworks/flagger/pkg/client/listers/smi/v1alpha1"
+	v1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	runtime "k8s.io/apimachinery/pkg/runtime"
+	watch "k8s.io/apimachinery/pkg/watch"
+	cache "k8s.io/client-go/tools/cache"
+)
+
+// TrafficSplitInformer provides access to a shared informer and lister for
+// TrafficSplits.
+type TrafficSplitInformer interface {
+	Informer() cache.SharedIndexInformer
+	Lister() v1alpha1.TrafficSplitLister
+}
+
+type trafficSplitInformer struct {
+	factory          internalinterfaces.SharedInformerFactory
+	tweakListOptions internalinterfaces.TweakListOptionsFunc
+	namespace        string
+}
+
+// NewTrafficSplitInformer constructs a new informer for TrafficSplit type.
+// Always prefer using an informer factory to get a shared informer instead of getting an independent
+// one. This reduces memory footprint and number of connections to the server.
+func NewTrafficSplitInformer(client versioned.Interface, namespace string, resyncPeriod time.Duration, indexers cache.Indexers) cache.SharedIndexInformer {
+	return NewFilteredTrafficSplitInformer(client, namespace, resyncPeriod, indexers, nil)
+}
+
+// NewFilteredTrafficSplitInformer constructs a new informer for TrafficSplit type.
+// Always prefer using an informer factory to get a shared informer instead of getting an independent
+// one. This reduces memory footprint and number of connections to the server.
+func NewFilteredTrafficSplitInformer(client versioned.Interface, namespace string, resyncPeriod time.Duration, indexers cache.Indexers, tweakListOptions internalinterfaces.TweakListOptionsFunc) cache.SharedIndexInformer {
+	return cache.NewSharedIndexInformer(
+		&cache.ListWatch{
+			ListFunc: func(options v1.ListOptions) (runtime.Object, error) {
+				if tweakListOptions != nil {
+					tweakListOptions(&options)
+				}
+				return client.SplitV1alpha1().TrafficSplits(namespace).List(options)
+			},
+			WatchFunc: func(options v1.ListOptions) (watch.Interface, error) {
+				if tweakListOptions != nil {
+					tweakListOptions(&options)
+				}
+				return client.SplitV1alpha1().TrafficSplits(namespace).Watch(options)
+			},
+		},
+		&smiv1alpha1.TrafficSplit{},
+		resyncPeriod,
+		indexers,
+	)
+}
+
+func (f *trafficSplitInformer) defaultInformer(client versioned.Interface, resyncPeriod time.Duration) cache.SharedIndexInformer {
+	return NewFilteredTrafficSplitInformer(client, f.namespace, resyncPeriod, cache.Indexers{cache.NamespaceIndex: cache.MetaNamespaceIndexFunc}, f.tweakListOptions)
+}
+
+func (f *trafficSplitInformer) Informer() cache.SharedIndexInformer {
+	return f.factory.InformerFor(&smiv1alpha1.TrafficSplit{}, f.defaultInformer)
+}
+
+func (f *trafficSplitInformer) Lister() v1alpha1.TrafficSplitLister {
+	return v1alpha1.NewTrafficSplitLister(f.Informer().GetIndexer())
+}

pkg/client/listers/smi/v1alpha1/expansion_generated.go

@@ -0,0 +1,27 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by lister-gen. DO NOT EDIT.
+
+package v1alpha1
+
+// TrafficSplitListerExpansion allows custom methods to be added to
+// TrafficSplitLister.
+type TrafficSplitListerExpansion interface{}
+
+// TrafficSplitNamespaceListerExpansion allows custom methods to be added to
+// TrafficSplitNamespaceLister.
+type TrafficSplitNamespaceListerExpansion interface{}

pkg/client/listers/smi/v1alpha1/trafficsplit.go

@@ -0,0 +1,94 @@
+/*
+Copyright The Flagger Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+// Code generated by lister-gen. DO NOT EDIT.
+
+package v1alpha1
+
+import (
+	v1alpha1 "github.com/weaveworks/flagger/pkg/apis/smi/v1alpha1"
+	"k8s.io/apimachinery/pkg/api/errors"
+	"k8s.io/apimachinery/pkg/labels"
+	"k8s.io/client-go/tools/cache"
+)
+
+// TrafficSplitLister helps list TrafficSplits.
+type TrafficSplitLister interface {
+	// List lists all TrafficSplits in the indexer.
+	List(selector labels.Selector) (ret []*v1alpha1.TrafficSplit, err error)
+	// TrafficSplits returns an object that can list and get TrafficSplits.
+	TrafficSplits(namespace string) TrafficSplitNamespaceLister
+	TrafficSplitListerExpansion
+}
+
+// trafficSplitLister implements the TrafficSplitLister interface.
+type trafficSplitLister struct {
+	indexer cache.Indexer
+}
+
+// NewTrafficSplitLister returns a new TrafficSplitLister.
+func NewTrafficSplitLister(indexer cache.Indexer) TrafficSplitLister {
+	return &trafficSplitLister{indexer: indexer}
+}
+
+// List lists all TrafficSplits in the indexer.
+func (s *trafficSplitLister) List(selector labels.Selector) (ret []*v1alpha1.TrafficSplit, err error) {
+	err = cache.ListAll(s.indexer, selector, func(m interface{}) {
+		ret = append(ret, m.(*v1alpha1.TrafficSplit))
+	})
+	return ret, err
+}
+
+// TrafficSplits returns an object that can list and get TrafficSplits.
+func (s *trafficSplitLister) TrafficSplits(namespace string) TrafficSplitNamespaceLister {
+	return trafficSplitNamespaceLister{indexer: s.indexer, namespace: namespace}
+}
+
+// TrafficSplitNamespaceLister helps list and get TrafficSplits.
+type TrafficSplitNamespaceLister interface {
+	// List lists all TrafficSplits in the indexer for a given namespace.
+	List(selector labels.Selector) (ret []*v1alpha1.TrafficSplit, err error)
+	// Get retrieves the TrafficSplit from the indexer for a given namespace and name.
+	Get(name string) (*v1alpha1.TrafficSplit, error)
+	TrafficSplitNamespaceListerExpansion
+}
+
+// trafficSplitNamespaceLister implements the TrafficSplitNamespaceLister
+// interface.
+type trafficSplitNamespaceLister struct {
+	indexer   cache.Indexer
+	namespace string
+}
+
+// List lists all TrafficSplits in the indexer for a given namespace.
+func (s trafficSplitNamespaceLister) List(selector labels.Selector) (ret []*v1alpha1.TrafficSplit, err error) {
+	err = cache.ListAllByNamespace(s.indexer, s.namespace, selector, func(m interface{}) {
+		ret = append(ret, m.(*v1alpha1.TrafficSplit))
+	})
+	return ret, err
+}
+
+// Get retrieves the TrafficSplit from the indexer for a given namespace and name.
+func (s trafficSplitNamespaceLister) Get(name string) (*v1alpha1.TrafficSplit, error) {
+	obj, exists, err := s.indexer.GetByKey(s.namespace + "/" + name)
+	if err != nil {
+		return nil, err
+	}
+	if !exists {
+		return nil, errors.NewNotFound(v1alpha1.Resource("trafficsplit"), name)
+	}
+	return obj.(*v1alpha1.TrafficSplit), nil
+}

pkg/controller/controller.go

@@ -33,23 +33,23 @@ const controllerAgentName = "flagger"
 
 // Controller is managing the canary objects and schedules canary deployments
 type Controller struct {
-	kubeClient    kubernetes.Interface
-	istioClient   clientset.Interface
-	flaggerClient clientset.Interface
-	flaggerLister flaggerlisters.CanaryLister
-	flaggerSynced cache.InformerSynced
-	flaggerWindow time.Duration
-	workqueue     workqueue.RateLimitingInterface
-	eventRecorder record.EventRecorder
-	logger        *zap.SugaredLogger
-	canaries      *sync.Map
-	jobs          map[string]CanaryJob
-	deployer      canary.Deployer
-	observer      metrics.Observer
-	recorder      metrics.Recorder
-	notifier      *notifier.Slack
-	routerFactory *router.Factory
-	meshProvider  string
+	kubeClient      kubernetes.Interface
+	istioClient     clientset.Interface
+	flaggerClient   clientset.Interface
+	flaggerLister   flaggerlisters.CanaryLister
+	flaggerSynced   cache.InformerSynced
+	flaggerWindow   time.Duration
+	workqueue       workqueue.RateLimitingInterface
+	eventRecorder   record.EventRecorder
+	logger          *zap.SugaredLogger
+	canaries        *sync.Map
+	jobs            map[string]CanaryJob
+	deployer        canary.Deployer
+	recorder        metrics.Recorder
+	notifier        *notifier.Slack
+	routerFactory   *router.Factory
+	observerFactory *metrics.Factory
+	meshProvider    string
 }
 
 func NewController(
@@ -62,6 +62,7 @@ func NewController(
 	logger *zap.SugaredLogger,
 	notifier *notifier.Slack,
 	routerFactory *router.Factory,
+	observerFactory *metrics.Factory,
 	meshProvider string,
 	version string,
 	labels []string,
@@ -92,23 +93,23 @@ func NewController(
 	recorder.SetInfo(version, meshProvider)
 
 	ctrl := &Controller{
-		kubeClient:    kubeClient,
-		istioClient:   istioClient,
-		flaggerClient: flaggerClient,
-		flaggerLister: flaggerInformer.Lister(),
-		flaggerSynced: flaggerInformer.Informer().HasSynced,
-		workqueue:     workqueue.NewNamedRateLimitingQueue(workqueue.DefaultControllerRateLimiter(), controllerAgentName),
-		eventRecorder: eventRecorder,
-		logger:        logger,
-		canaries:      new(sync.Map),
-		jobs:          map[string]CanaryJob{},
-		flaggerWindow: flaggerWindow,
-		deployer:      deployer,
-		observer:      metrics.NewObserver(metricServer),
-		recorder:      recorder,
-		notifier:      notifier,
-		routerFactory: routerFactory,
-		meshProvider:  meshProvider,
+		kubeClient:      kubeClient,
+		istioClient:     istioClient,
+		flaggerClient:   flaggerClient,
+		flaggerLister:   flaggerInformer.Lister(),
+		flaggerSynced:   flaggerInformer.Informer().HasSynced,
+		workqueue:       workqueue.NewNamedRateLimitingQueue(workqueue.DefaultControllerRateLimiter(), controllerAgentName),
+		eventRecorder:   eventRecorder,
+		logger:          logger,
+		canaries:        new(sync.Map),
+		jobs:            map[string]CanaryJob{},
+		flaggerWindow:   flaggerWindow,
+		deployer:        deployer,
+		observerFactory: observerFactory,
+		recorder:        recorder,
+		notifier:        notifier,
+		routerFactory:   routerFactory,
+		meshProvider:    meshProvider,
 	}
 
 	flaggerInformer.Informer().AddEventHandler(cache.ResourceEventHandlerFuncs{

pkg/controller/controller_test.go

@@ -37,7 +37,6 @@ type Mocks struct {
 	meshClient    clientset.Interface
 	flaggerClient clientset.Interface
 	deployer      canary.Deployer
-	observer      metrics.Observer
 	ctrl          *Controller
 	logger        *zap.SugaredLogger
 	router        router.Interface
@@ -77,7 +76,6 @@ func SetupMocks(abtest bool) Mocks {
 			FlaggerClient: flaggerClient,
 		},
 	}
-	observer := metrics.NewObserver("fake")
 
 	// init controller
 	flaggerInformerFactory := informers.NewSharedInformerFactory(flaggerClient, noResyncPeriodFunc())
@@ -86,21 +84,24 @@ func SetupMocks(abtest bool) Mocks {
 	// init router
 	rf := router.NewFactory(nil, kubeClient, flaggerClient, logger, flaggerClient)
 
+	// init observer
+	observerFactory, _ := metrics.NewFactory("fake", "istio", 5*time.Second)
+
 	ctrl := &Controller{
-		kubeClient:    kubeClient,
-		istioClient:   flaggerClient,
-		flaggerClient: flaggerClient,
-		flaggerLister: flaggerInformer.Lister(),
-		flaggerSynced: flaggerInformer.Informer().HasSynced,
-		workqueue:     workqueue.NewNamedRateLimitingQueue(workqueue.DefaultControllerRateLimiter(), controllerAgentName),
-		eventRecorder: &record.FakeRecorder{},
-		logger:        logger,
-		canaries:      new(sync.Map),
-		flaggerWindow: time.Second,
-		deployer:      deployer,
-		observer:      observer,
-		recorder:      metrics.NewRecorder(controllerAgentName, false),
-		routerFactory: rf,
+		kubeClient:      kubeClient,
+		istioClient:     flaggerClient,
+		flaggerClient:   flaggerClient,
+		flaggerLister:   flaggerInformer.Lister(),
+		flaggerSynced:   flaggerInformer.Informer().HasSynced,
+		workqueue:       workqueue.NewNamedRateLimitingQueue(workqueue.DefaultControllerRateLimiter(), controllerAgentName),
+		eventRecorder:   &record.FakeRecorder{},
+		logger:          logger,
+		canaries:        new(sync.Map),
+		flaggerWindow:   time.Second,
+		deployer:        deployer,
+		observerFactory: observerFactory,
+		recorder:        metrics.NewRecorder(controllerAgentName, false),
+		routerFactory:   rf,
 	}
 	ctrl.flaggerSynced = alwaysReady
 
@@ -108,7 +109,6 @@ func SetupMocks(abtest bool) Mocks {
 
 	return Mocks{
 		canary:        c,
-		observer:      observer,
 		deployer:      deployer,
 		logger:        logger,
 		flaggerClient: flaggerClient,

pkg/controller/scheduler.go

@@ -556,126 +556,65 @@ func (c *Controller) analyseCanary(r *flaggerv1.Canary) bool {
 		}
 	}
 
+	// create observer based on the mesh provider
+	observer := c.observerFactory.Observer()
+
 	// run metrics checks
 	for _, metric := range r.Spec.CanaryAnalysis.Metrics {
 		if metric.Interval == "" {
 			metric.Interval = r.GetMetricInterval()
 		}
 
-		// App Mesh checks
-		if c.meshProvider == "appmesh" {
-			if metric.Name == "request-success-rate" || metric.Name == "envoy_cluster_upstream_rq" {
-				val, err := c.observer.GetEnvoySuccessRate(r.Spec.TargetRef.Name, r.Namespace, metric.Name, metric.Interval)
-				if err != nil {
-					if strings.Contains(err.Error(), "no values found") {
-						c.recordEventWarningf(r, "Halt advancement no values found for metric %s probably %s.%s is not receiving traffic",
-							metric.Name, r.Spec.TargetRef.Name, r.Namespace)
-					} else {
-						c.recordEventErrorf(r, "Metrics server %s query failed: %v", c.observer.GetMetricsServer(), err)
-					}
-					return false
-				}
-				if float64(metric.Threshold) > val {
-					c.recordEventWarningf(r, "Halt %s.%s advancement success rate %.2f%% < %v%%",
-						r.Name, r.Namespace, val, metric.Threshold)
-					return false
-				}
-			}
-
-			if metric.Name == "request-duration" || metric.Name == "envoy_cluster_upstream_rq_time_bucket" {
-				val, err := c.observer.GetEnvoyRequestDuration(r.Spec.TargetRef.Name, r.Namespace, metric.Name, metric.Interval)
-				if err != nil {
-					c.recordEventErrorf(r, "Metrics server %s query failed: %v", c.observer.GetMetricsServer(), err)
-					return false
-				}
-				t := time.Duration(metric.Threshold) * time.Millisecond
-				if val > t {
-					c.recordEventWarningf(r, "Halt %s.%s advancement request duration %v > %v",
-						r.Name, r.Namespace, val, t)
-					return false
-				}
-			}
-		}
-
-		// Istio checks
-		if c.meshProvider == "istio" {
-			if metric.Name == "request-success-rate" || metric.Name == "istio_requests_total" {
-				val, err := c.observer.GetIstioSuccessRate(r.Spec.TargetRef.Name, r.Namespace, metric.Name, metric.Interval)
-				if err != nil {
-					if strings.Contains(err.Error(), "no values found") {
-						c.recordEventWarningf(r, "Halt advancement no values found for metric %s probably %s.%s is not receiving traffic",
-							metric.Name, r.Spec.TargetRef.Name, r.Namespace)
-					} else {
-						c.recordEventErrorf(r, "Metrics server %s query failed: %v", c.observer.GetMetricsServer(), err)
-					}
-					return false
-				}
-				if float64(metric.Threshold) > val {
-					c.recordEventWarningf(r, "Halt %s.%s advancement success rate %.2f%% < %v%%",
-						r.Name, r.Namespace, val, metric.Threshold)
-					return false
-				}
-			}
-
-			if metric.Name == "request-duration" || metric.Name == "istio_request_duration_seconds_bucket" {
-				val, err := c.observer.GetIstioRequestDuration(r.Spec.TargetRef.Name, r.Namespace, metric.Name, metric.Interval)
-				if err != nil {
-					c.recordEventErrorf(r, "Metrics server %s query failed: %v", c.observer.GetMetricsServer(), err)
-					return false
-				}
-				t := time.Duration(metric.Threshold) * time.Millisecond
-				if val > t {
-					c.recordEventWarningf(r, "Halt %s.%s advancement request duration %v > %v",
-						r.Name, r.Namespace, val, t)
-					return false
-				}
-			}
-		}
-
-		// NGINX checks
-		if c.meshProvider == "nginx" {
-			if metric.Name == "request-success-rate" {
-				val, err := c.observer.GetNginxSuccessRate(r.Spec.IngressRef.Name, r.Namespace, metric.Name, metric.Interval)
-				if err != nil {
-					if strings.Contains(err.Error(), "no values found") {
-						c.recordEventWarningf(r, "Halt advancement no values found for metric %s probably %s.%s is not receiving traffic",
-							metric.Name, r.Spec.TargetRef.Name, r.Namespace)
-					} else {
-						c.recordEventErrorf(r, "Metrics server %s query failed: %v", c.observer.GetMetricsServer(), err)
-					}
-					return false
-				}
-				if float64(metric.Threshold) > val {
-					c.recordEventWarningf(r, "Halt %s.%s advancement success rate %.2f%% < %v%%",
-						r.Name, r.Namespace, val, metric.Threshold)
-					return false
-				}
-			}
-
-			if metric.Name == "request-duration" {
-				val, err := c.observer.GetNginxRequestDuration(r.Spec.IngressRef.Name, r.Namespace, metric.Name, metric.Interval)
-				if err != nil {
-					c.recordEventErrorf(r, "Metrics server %s query failed: %v", c.observer.GetMetricsServer(), err)
-					return false
-				}
-				t := time.Duration(metric.Threshold) * time.Millisecond
-				if val > t {
-					c.recordEventWarningf(r, "Halt %s.%s advancement request duration %v > %v",
-						r.Name, r.Namespace, val, t)
-					return false
-				}
-			}
-		}
-
-		// custom checks
-		if metric.Query != "" {
-			val, err := c.observer.GetScalar(metric.Query)
+		if metric.Name == "request-success-rate" {
+			val, err := observer.GetRequestSuccessRate(r.Spec.TargetRef.Name, r.Namespace, metric.Interval)
 			if err != nil {
 				if strings.Contains(err.Error(), "no values found") {
 					c.recordEventWarningf(r, "Halt advancement no values found for metric %s probably %s.%s is not receiving traffic",
 						metric.Name, r.Spec.TargetRef.Name, r.Namespace)
 				} else {
-					c.recordEventErrorf(r, "Metrics server %s query failed: %v", c.observer.GetMetricsServer(), err)
+					c.recordEventErrorf(r, "Metrics server %s query failed: %v", c.observerFactory.Client.GetMetricsServer(), err)
+				}
+				return false
+			}
+			if float64(metric.Threshold) > val {
+				c.recordEventWarningf(r, "Halt %s.%s advancement success rate %.2f%% < %v%%",
+					r.Name, r.Namespace, val, metric.Threshold)
+				return false
+			}
+
+			//c.recordEventInfof(r, "Check %s passed %.2f%% > %v%%", metric.Name, val, metric.Threshold)
+		}
+
+		if metric.Name == "request-duration" {
+			val, err := observer.GetRequestDuration(r.Spec.TargetRef.Name, r.Namespace, metric.Interval)
+			if err != nil {
+				if strings.Contains(err.Error(), "no values found") {
+					c.recordEventWarningf(r, "Halt advancement no values found for metric %s probably %s.%s is not receiving traffic",
+						metric.Name, r.Spec.TargetRef.Name, r.Namespace)
+				} else {
+					c.recordEventErrorf(r, "Metrics server %s query failed: %v", c.observerFactory.Client.GetMetricsServer(), err)
+				}
+				return false
+			}
+			t := time.Duration(metric.Threshold) * time.Millisecond
+			if val > t {
+				c.recordEventWarningf(r, "Halt %s.%s advancement request duration %v > %v",
+					r.Name, r.Namespace, val, t)
+				return false
+			}
+
+			//c.recordEventInfof(r, "Check %s passed %v < %v", metric.Name, val, metric.Threshold)
+		}
+
+		// custom checks
+		if metric.Query != "" {
+			val, err := c.observerFactory.Client.RunQuery(metric.Query)
+			if err != nil {
+				if strings.Contains(err.Error(), "no values found") {
+					c.recordEventWarningf(r, "Halt advancement no values found for metric %s probably %s.%s is not receiving traffic",
+						metric.Name, r.Spec.TargetRef.Name, r.Namespace)
+				} else {
+					c.recordEventErrorf(r, "Metrics server %s query failed: %v", c.observerFactory.Client.GetMetricsServer(), err)
 				}
 				return false
 			}

pkg/metrics/client.go

@@ -0,0 +1,184 @@
+package metrics
+
+import (
+	"bufio"
+	"bytes"
+	"context"
+	"encoding/json"
+	"fmt"
+	"io/ioutil"
+	"net/http"
+	"net/url"
+	"strconv"
+	"strings"
+	"text/template"
+	"time"
+)
+
+// PrometheusClient is executing promql queries
+type PrometheusClient struct {
+	timeout time.Duration
+	url     url.URL
+}
+
+type prometheusResponse struct {
+	Data struct {
+		Result []struct {
+			Metric struct {
+				Name string `json:"name"`
+			}
+			Value []interface{} `json:"value"`
+		}
+	}
+}
+
+// NewPrometheusClient creates a Prometheus client for the provided URL address
+func NewPrometheusClient(address string, timeout time.Duration) (*PrometheusClient, error) {
+	promURL, err := url.Parse(address)
+	if err != nil {
+		return nil, err
+	}
+
+	return &PrometheusClient{timeout: timeout, url: *promURL}, nil
+}
+
+// RenderQuery renders the promql query using the provided text template
+func (p *PrometheusClient) RenderQuery(name string, namespace string, interval string, tmpl string) (string, error) {
+	meta := struct {
+		Name      string
+		Namespace string
+		Interval  string
+	}{
+		name,
+		namespace,
+		interval,
+	}
+
+	t, err := template.New("tmpl").Parse(tmpl)
+	if err != nil {
+		return "", err
+	}
+	var data bytes.Buffer
+	b := bufio.NewWriter(&data)
+
+	if err := t.Execute(b, meta); err != nil {
+		return "", err
+	}
+
+	err = b.Flush()
+	if err != nil {
+		return "", err
+	}
+
+	return data.String(), nil
+}
+
+// RunQuery executes the promql and converts the result to float64
+func (p *PrometheusClient) RunQuery(query string) (float64, error) {
+	if p.url.Host == "fake" {
+		return 100, nil
+	}
+
+	query = url.QueryEscape(p.TrimQuery(query))
+	u, err := url.Parse(fmt.Sprintf("./api/v1/query?query=%s", query))
+	if err != nil {
+		return 0, err
+	}
+
+	u = p.url.ResolveReference(u)
+
+	req, err := http.NewRequest("GET", u.String(), nil)
+	if err != nil {
+		return 0, err
+	}
+
+	ctx, cancel := context.WithTimeout(req.Context(), p.timeout)
+	defer cancel()
+
+	r, err := http.DefaultClient.Do(req.WithContext(ctx))
+	if err != nil {
+		return 0, err
+	}
+	defer r.Body.Close()
+
+	b, err := ioutil.ReadAll(r.Body)
+	if err != nil {
+		return 0, fmt.Errorf("error reading body: %s", err.Error())
+	}
+
+	if 400 <= r.StatusCode {
+		return 0, fmt.Errorf("error response: %s", string(b))
+	}
+
+	var result prometheusResponse
+	err = json.Unmarshal(b, &result)
+	if err != nil {
+		return 0, fmt.Errorf("error unmarshaling result: %s, '%s'", err.Error(), string(b))
+	}
+
+	var value *float64
+	for _, v := range result.Data.Result {
+		metricValue := v.Value[1]
+		switch metricValue.(type) {
+		case string:
+			f, err := strconv.ParseFloat(metricValue.(string), 64)
+			if err != nil {
+				return 0, err
+			}
+			value = &f
+		}
+	}
+	if value == nil {
+		return 0, fmt.Errorf("no values found")
+	}
+
+	return *value, nil
+}
+
+// TrimQuery takes a promql query and removes spaces, tabs and new lines
+func (p *PrometheusClient) TrimQuery(query string) string {
+	query = strings.Replace(query, "\n", "", -1)
+	query = strings.Replace(query, "\t", "", -1)
+	query = strings.Replace(query, " ", "", -1)
+
+	return query
+}
+
+// IsOnline call Prometheus status endpoint and returns an error if the API is unreachable
+func (p *PrometheusClient) IsOnline() (bool, error) {
+	u, err := url.Parse("./api/v1/status/flags")
+	if err != nil {
+		return false, err
+	}
+
+	u = p.url.ResolveReference(u)
+
+	req, err := http.NewRequest("GET", u.String(), nil)
+	if err != nil {
+		return false, err
+	}
+
+	ctx, cancel := context.WithTimeout(req.Context(), p.timeout)
+	defer cancel()
+
+	r, err := http.DefaultClient.Do(req.WithContext(ctx))
+	if err != nil {
+		return false, err
+	}
+	defer r.Body.Close()
+
+	b, err := ioutil.ReadAll(r.Body)
+	if err != nil {
+		return false, fmt.Errorf("error reading body: %s", err.Error())
+	}
+
+	if 400 <= r.StatusCode {
+		return false, fmt.Errorf("error response: %s", string(b))
+	}
+
+	return true, nil
+}
+
+func (p *PrometheusClient) GetMetricsServer() string {
+	return p.url.RawQuery
+}

pkg/metrics/client_test.go

@@ -0,0 +1,85 @@
+package metrics
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"testing"
+	"time"
+)
+
+func TestPrometheusClient_RunQuery(t *testing.T) {
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1545905245.458,"100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	query := `
+        histogram_quantile(0.99,
+          sum(
+            rate(
+              http_request_duration_seconds_bucket{
+                kubernetes_namespace="test",
+                kubernetes_pod_name=~"podinfo-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"
+              }[1m]
+            )
+          ) by (le)
+        )`
+
+	val, err := client.RunQuery(query)
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100 {
+		t.Errorf("Got %v wanted %v", val, 100)
+	}
+}
+
+func TestPrometheusClient_IsOnline(t *testing.T) {
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		json := `{"status":"success","data":{"config.file":"/etc/prometheus/prometheus.yml"}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	ok, err := client.IsOnline()
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if !ok {
+		t.Errorf("Got %v wanted %v", ok, true)
+	}
+}
+
+func TestPrometheusClient_IsOffline(t *testing.T) {
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(http.StatusBadGateway)
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	ok, err := client.IsOnline()
+	if err == nil {
+		t.Errorf("Got no error wanted %v", http.StatusBadGateway)
+	}
+
+	if ok {
+		t.Errorf("Got %v wanted %v", ok, false)
+	}
+}

pkg/metrics/envoy.go

@@ -1,119 +1,73 @@
 package metrics
 
 import (
-	"fmt"
-	"net/url"
-	"strconv"
 	"time"
 )
 
-const envoySuccessRateQuery = `
-sum(rate(
-envoy_cluster_upstream_rq{kubernetes_namespace="{{ .Namespace }}",
-kubernetes_pod_name=~"{{ .Name }}-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)",
-envoy_response_code!~"5.*"}
-[{{ .Interval }}])) 
-/ 
-sum(rate(
-envoy_cluster_upstream_rq{kubernetes_namespace="{{ .Namespace }}",
-kubernetes_pod_name=~"{{ .Name }}-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"}
-[{{ .Interval }}])) 
-* 100
-`
-
-func (c *Observer) GetEnvoySuccessRate(name string, namespace string, metric string, interval string) (float64, error) {
-	if c.metricsServer == "fake" {
-		return 100, nil
-	}
-
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		name,
-		namespace,
-		interval,
-	}
-
-	query, err := render(meta, envoySuccessRateQuery)
-	if err != nil {
-		return 0, err
-	}
-
-	var rate *float64
-	querySt := url.QueryEscape(query)
-	result, err := c.queryMetric(querySt)
-	if err != nil {
-		return 0, err
-	}
-
-	for _, v := range result.Data.Result {
-		metricValue := v.Value[1]
-		switch metricValue.(type) {
-		case string:
-			f, err := strconv.ParseFloat(metricValue.(string), 64)
-			if err != nil {
-				return 0, err
-			}
-			rate = &f
-		}
-	}
-	if rate == nil {
-		return 0, fmt.Errorf("no values found for metric %s", metric)
-	}
-	return *rate, nil
+var envoyQueries = map[string]string{
+	"request-success-rate": `
+	sum(
+		rate(
+			envoy_cluster_upstream_rq{
+				kubernetes_namespace="{{ .Namespace }}",
+				kubernetes_pod_name=~"{{ .Name }}-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)",
+				envoy_response_code!~"5.*"
+			}[{{ .Interval }}]
+		)
+	) 
+	/ 
+	sum(
+		rate(
+			envoy_cluster_upstream_rq{
+				kubernetes_namespace="{{ .Namespace }}",
+				kubernetes_pod_name=~"{{ .Name }}-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"
+			}[{{ .Interval }}]
+		)
+	) 
+	* 100`,
+	"request-duration": `
+	histogram_quantile(
+		0.99,
+		sum(
+			rate(
+				envoy_cluster_upstream_rq_time_bucket{
+					kubernetes_namespace="{{ .Namespace }}",
+					kubernetes_pod_name=~"{{ .Name }}-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"
+				}[{{ .Interval }}]
+			)
+		) by (le)
+	)`,
 }
 
-const envoyRequestDurationQuery = `
-histogram_quantile(0.99, sum(rate(
-envoy_cluster_upstream_rq_time_bucket{kubernetes_namespace="{{ .Namespace }}",
-kubernetes_pod_name=~"{{ .Name }}-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"}
-[{{ .Interval }}])) by (le))
-`
+type EnvoyObserver struct {
+	client *PrometheusClient
+}
 
-// GetEnvoyRequestDuration returns the 99P requests delay using envoy_cluster_upstream_rq_time_bucket metrics
-func (c *Observer) GetEnvoyRequestDuration(name string, namespace string, metric string, interval string) (time.Duration, error) {
-	if c.metricsServer == "fake" {
-		return 1, nil
-	}
-
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		name,
-		namespace,
-		interval,
-	}
-
-	query, err := render(meta, envoyRequestDurationQuery)
+func (ob *EnvoyObserver) GetRequestSuccessRate(name string, namespace string, interval string) (float64, error) {
+	query, err := ob.client.RenderQuery(name, namespace, interval, envoyQueries["request-success-rate"])
 	if err != nil {
 		return 0, err
 	}
 
-	var rate *float64
-	querySt := url.QueryEscape(query)
-	result, err := c.queryMetric(querySt)
+	value, err := ob.client.RunQuery(query)
 	if err != nil {
 		return 0, err
 	}
 
-	for _, v := range result.Data.Result {
-		metricValue := v.Value[1]
-		switch metricValue.(type) {
-		case string:
-			f, err := strconv.ParseFloat(metricValue.(string), 64)
-			if err != nil {
-				return 0, err
-			}
-			rate = &f
-		}
+	return value, nil
+}
+
+func (ob *EnvoyObserver) GetRequestDuration(name string, namespace string, interval string) (time.Duration, error) {
+	query, err := ob.client.RenderQuery(name, namespace, interval, envoyQueries["request-duration"])
+	if err != nil {
+		return 0, err
 	}
-	if rate == nil {
-		return 0, fmt.Errorf("no values found for metric %s", metric)
+
+	value, err := ob.client.RunQuery(query)
+	if err != nil {
+		return 0, err
 	}
-	ms := time.Duration(int64(*rate)) * time.Millisecond
+
+	ms := time.Duration(int64(value)) * time.Millisecond
 	return ms, nil
 }

pkg/metrics/envoy_test.go

@@ -1,51 +1,74 @@
 package metrics
 
 import (
+	"net/http"
+	"net/http/httptest"
 	"testing"
+	"time"
 )
 
-func Test_EnvoySuccessRateQueryRender(t *testing.T) {
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		"podinfo",
-		"default",
-		"1m",
-	}
+func TestEnvoyObserver_GetRequestSuccessRate(t *testing.T) {
+	expected := `sum(rate(envoy_cluster_upstream_rq{kubernetes_namespace="default",kubernetes_pod_name=~"podinfo-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)",envoy_response_code!~"5.*"}[1m]))/sum(rate(envoy_cluster_upstream_rq{kubernetes_namespace="default",kubernetes_pod_name=~"podinfo-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"}[1m]))*100`
 
-	query, err := render(meta, envoySuccessRateQuery)
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		promql := r.URL.Query()["query"][0]
+		if promql != expected {
+			t.Errorf("\nGot %s \nWanted %s", promql, expected)
+		}
+
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1,"100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
 	if err != nil {
 		t.Fatal(err)
 	}
 
-	expected := `sum(rate(envoy_cluster_upstream_rq{kubernetes_namespace="default",kubernetes_pod_name=~"podinfo-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)",envoy_response_code!~"5.*"}[1m])) / sum(rate(envoy_cluster_upstream_rq{kubernetes_namespace="default",kubernetes_pod_name=~"podinfo-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"}[1m])) * 100`
+	observer := &EnvoyObserver{
+		client: client,
+	}
 
-	if query != expected {
-		t.Errorf("\nGot %s \nWanted %s", query, expected)
+	val, err := observer.GetRequestSuccessRate("podinfo", "default", "1m")
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100 {
+		t.Errorf("Got %v wanted %v", val, 100)
 	}
 }
 
-func Test_EnvoyRequestDurationQueryRender(t *testing.T) {
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		"podinfo",
-		"default",
-		"1m",
-	}
+func TestEnvoyObserver_GetRequestDuration(t *testing.T) {
+	expected := `histogram_quantile(0.99,sum(rate(envoy_cluster_upstream_rq_time_bucket{kubernetes_namespace="default",kubernetes_pod_name=~"podinfo-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"}[1m]))by(le))`
 
-	query, err := render(meta, envoyRequestDurationQuery)
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		promql := r.URL.Query()["query"][0]
+		if promql != expected {
+			t.Errorf("\nGot %s \nWanted %s", promql, expected)
+		}
+
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1,"100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
 	if err != nil {
 		t.Fatal(err)
 	}
 
-	expected := `histogram_quantile(0.99, sum(rate(envoy_cluster_upstream_rq_time_bucket{kubernetes_namespace="default",kubernetes_pod_name=~"podinfo-[0-9a-zA-Z]+(-[0-9a-zA-Z]+)"}[1m])) by (le))`
+	observer := &EnvoyObserver{
+		client: client,
+	}
 
-	if query != expected {
-		t.Errorf("\nGot %s \nWanted %s", query, expected)
+	val, err := observer.GetRequestDuration("podinfo", "default", "1m")
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100*time.Millisecond {
+		t.Errorf("Got %v wanted %v", val, 100*time.Millisecond)
 	}
 }

pkg/metrics/factory.go

@@ -0,0 +1,48 @@
+package metrics
+
+import (
+	"strings"
+	"time"
+)
+
+type Factory struct {
+	MeshProvider string
+	Client       *PrometheusClient
+}
+
+func NewFactory(metricsServer string, meshProvider string, timeout time.Duration) (*Factory, error) {
+	client, err := NewPrometheusClient(metricsServer, timeout)
+	if err != nil {
+		return nil, err
+	}
+
+	return &Factory{
+		MeshProvider: meshProvider,
+		Client:       client,
+	}, nil
+}
+
+func (factory Factory) Observer() Interface {
+	switch {
+	case factory.MeshProvider == "appmesh":
+		return &EnvoyObserver{
+			client: factory.Client,
+		}
+	case factory.MeshProvider == "nginx":
+		return &NginxObserver{
+			client: factory.Client,
+		}
+	case strings.HasPrefix(factory.MeshProvider, "gloo"):
+		return &GlooObserver{
+			client: factory.Client,
+		}
+	case factory.MeshProvider == "smi:linkerd":
+		return &LinkerdObserver{
+			client: factory.Client,
+		}
+	default:
+		return &IstioObserver{
+			client: factory.Client,
+		}
+	}
+}

pkg/metrics/gloo.go

@@ -0,0 +1,72 @@
+package metrics
+
+import (
+	"time"
+)
+
+//envoy_cluster_name="test-podinfo-primary-9898_gloo-system"
+
+var glooQueries = map[string]string{
+	"request-success-rate": `
+	sum(
+		rate(
+			envoy_cluster_upstream_rq{
+				envoy_cluster_name=~"{{ .Namespace }}-{{ .Name }}-canary-[0-9a-zA-Z-]+_[0-9a-zA-Z-]+",
+				envoy_response_code!~"5.*"
+			}[{{ .Interval }}]
+		)
+	) 
+	/ 
+	sum(
+		rate(
+			envoy_cluster_upstream_rq{
+				envoy_cluster_name=~"{{ .Namespace }}-{{ .Name }}-canary-[0-9a-zA-Z-]+_[0-9a-zA-Z-]+",
+			}[{{ .Interval }}]
+		)
+	) 
+	* 100`,
+	"request-duration": `
+	histogram_quantile(
+		0.99,
+		sum(
+			rate(
+				envoy_cluster_upstream_rq_time_bucket{
+					envoy_cluster_name=~"{{ .Namespace }}-{{ .Name }}-canary-[0-9a-zA-Z-]+_[0-9a-zA-Z-]+",
+				}[{{ .Interval }}]
+			)
+		) by (le)
+	)`,
+}
+
+type GlooObserver struct {
+	client *PrometheusClient
+}
+
+func (ob *GlooObserver) GetRequestSuccessRate(name string, namespace string, interval string) (float64, error) {
+	query, err := ob.client.RenderQuery(name, namespace, interval, glooQueries["request-success-rate"])
+	if err != nil {
+		return 0, err
+	}
+
+	value, err := ob.client.RunQuery(query)
+	if err != nil {
+		return 0, err
+	}
+
+	return value, nil
+}
+
+func (ob *GlooObserver) GetRequestDuration(name string, namespace string, interval string) (time.Duration, error) {
+	query, err := ob.client.RenderQuery(name, namespace, interval, glooQueries["request-duration"])
+	if err != nil {
+		return 0, err
+	}
+
+	value, err := ob.client.RunQuery(query)
+	if err != nil {
+		return 0, err
+	}
+
+	ms := time.Duration(int64(value)) * time.Millisecond
+	return ms, nil
+}

pkg/metrics/gloo_test.go

@@ -0,0 +1,74 @@
+package metrics
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"testing"
+	"time"
+)
+
+func TestGlooObserver_GetRequestSuccessRate(t *testing.T) {
+	expected := `sum(rate(envoy_cluster_upstream_rq{envoy_cluster_name=~"default-podinfo-canary-[0-9a-zA-Z-]+_[0-9a-zA-Z-]+",envoy_response_code!~"5.*"}[1m]))/sum(rate(envoy_cluster_upstream_rq{envoy_cluster_name=~"default-podinfo-canary-[0-9a-zA-Z-]+_[0-9a-zA-Z-]+",}[1m]))*100`
+
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		promql := r.URL.Query()["query"][0]
+		if promql != expected {
+			t.Errorf("\nGot %s \nWanted %s", promql, expected)
+		}
+
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1,"100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	observer := &GlooObserver{
+		client: client,
+	}
+
+	val, err := observer.GetRequestSuccessRate("podinfo", "default", "1m")
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100 {
+		t.Errorf("Got %v wanted %v", val, 100)
+	}
+}
+
+func TestGlooObserver_GetRequestDuration(t *testing.T) {
+	expected := `histogram_quantile(0.99,sum(rate(envoy_cluster_upstream_rq_time_bucket{envoy_cluster_name=~"default-podinfo-canary-[0-9a-zA-Z-]+_[0-9a-zA-Z-]+",}[1m]))by(le))`
+
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		promql := r.URL.Query()["query"][0]
+		if promql != expected {
+			t.Errorf("\nGot %s \nWanted %s", promql, expected)
+		}
+
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1,"100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	observer := &GlooObserver{
+		client: client,
+	}
+
+	val, err := observer.GetRequestDuration("podinfo", "default", "1m")
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100*time.Millisecond {
+		t.Errorf("Got %v wanted %v", val, 100*time.Millisecond)
+	}
+}

pkg/metrics/istio.go

@@ -1,123 +1,76 @@
 package metrics
 
 import (
-	"fmt"
-	"net/url"
-	"strconv"
 	"time"
 )
 
-const istioSuccessRateQuery = `
-sum(rate(
-istio_requests_total{reporter="destination",
-destination_workload_namespace="{{ .Namespace }}",
-destination_workload=~"{{ .Name }}",
-response_code!~"5.*"}
-[{{ .Interval }}])) 
-/ 
-sum(rate(
-istio_requests_total{reporter="destination",
-destination_workload_namespace="{{ .Namespace }}",
-destination_workload=~"{{ .Name }}"}
-[{{ .Interval }}])) 
-* 100
-`
-
-// GetIstioSuccessRate returns the requests success rate (non 5xx) using istio_requests_total metric
-func (c *Observer) GetIstioSuccessRate(name string, namespace string, metric string, interval string) (float64, error) {
-	if c.metricsServer == "fake" {
-		return 100, nil
-	}
-
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		name,
-		namespace,
-		interval,
-	}
-
-	query, err := render(meta, istioSuccessRateQuery)
-	if err != nil {
-		return 0, err
-	}
-
-	var rate *float64
-	querySt := url.QueryEscape(query)
-	result, err := c.queryMetric(querySt)
-	if err != nil {
-		return 0, err
-	}
-
-	for _, v := range result.Data.Result {
-		metricValue := v.Value[1]
-		switch metricValue.(type) {
-		case string:
-			f, err := strconv.ParseFloat(metricValue.(string), 64)
-			if err != nil {
-				return 0, err
-			}
-			rate = &f
-		}
-	}
-	if rate == nil {
-		return 0, fmt.Errorf("no values found for metric %s", metric)
-	}
-	return *rate, nil
+var istioQueries = map[string]string{
+	"request-success-rate": `
+	sum(
+		rate(
+			istio_requests_total{
+				reporter="destination",
+				destination_workload_namespace="{{ .Namespace }}",
+				destination_workload=~"{{ .Name }}",
+				response_code!~"5.*"
+			}[{{ .Interval }}]
+		)
+	) 
+	/ 
+	sum(
+		rate(
+			istio_requests_total{
+				reporter="destination",
+				destination_workload_namespace="{{ .Namespace }}",
+				destination_workload=~"{{ .Name }}"
+			}[{{ .Interval }}]
+		)
+	) 
+	* 100`,
+	"request-duration": `
+	histogram_quantile(
+		0.99,
+		sum(
+			rate(
+				istio_request_duration_seconds_bucket{
+					reporter="destination",
+					destination_workload_namespace="{{ .Namespace }}",
+					destination_workload=~"{{ .Name }}"
+				}[{{ .Interval }}]
+			)
+		) by (le)
+	)`,
 }
 
-const istioRequestDurationQuery = `
-histogram_quantile(0.99, sum(rate(
-istio_request_duration_seconds_bucket{reporter="destination",
-destination_workload_namespace="{{ .Namespace }}",
-destination_workload=~"{{ .Name }}"}
-[{{ .Interval }}])) by (le))
-`
+type IstioObserver struct {
+	client *PrometheusClient
+}
 
-// GetIstioRequestDuration returns the 99P requests delay using istio_request_duration_seconds_bucket metrics
-func (c *Observer) GetIstioRequestDuration(name string, namespace string, metric string, interval string) (time.Duration, error) {
-	if c.metricsServer == "fake" {
-		return 1, nil
-	}
-
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		name,
-		namespace,
-		interval,
-	}
-
-	query, err := render(meta, istioRequestDurationQuery)
+func (ob *IstioObserver) GetRequestSuccessRate(name string, namespace string, interval string) (float64, error) {
+	query, err := ob.client.RenderQuery(name, namespace, interval, istioQueries["request-success-rate"])
 	if err != nil {
 		return 0, err
 	}
 
-	var rate *float64
-	querySt := url.QueryEscape(query)
-	result, err := c.queryMetric(querySt)
+	value, err := ob.client.RunQuery(query)
 	if err != nil {
 		return 0, err
 	}
 
-	for _, v := range result.Data.Result {
-		metricValue := v.Value[1]
-		switch metricValue.(type) {
-		case string:
-			f, err := strconv.ParseFloat(metricValue.(string), 64)
-			if err != nil {
-				return 0, err
-			}
-			rate = &f
-		}
+	return value, nil
+}
+
+func (ob *IstioObserver) GetRequestDuration(name string, namespace string, interval string) (time.Duration, error) {
+	query, err := ob.client.RenderQuery(name, namespace, interval, istioQueries["request-duration"])
+	if err != nil {
+		return 0, err
 	}
-	if rate == nil {
-		return 0, fmt.Errorf("no values found for metric %s", metric)
+
+	value, err := ob.client.RunQuery(query)
+	if err != nil {
+		return 0, err
 	}
-	ms := time.Duration(int64(*rate*1000)) * time.Millisecond
+
+	ms := time.Duration(int64(value*1000)) * time.Millisecond
 	return ms, nil
 }

pkg/metrics/istio_test.go

@@ -1,51 +1,74 @@
 package metrics
 
 import (
+	"net/http"
+	"net/http/httptest"
 	"testing"
+	"time"
 )
 
-func Test_IstioSuccessRateQueryRender(t *testing.T) {
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		"podinfo",
-		"default",
-		"1m",
-	}
+func TestIstioObserver_GetRequestSuccessRate(t *testing.T) {
+	expected := `sum(rate(istio_requests_total{reporter="destination",destination_workload_namespace="default",destination_workload=~"podinfo",response_code!~"5.*"}[1m]))/sum(rate(istio_requests_total{reporter="destination",destination_workload_namespace="default",destination_workload=~"podinfo"}[1m]))*100`
 
-	query, err := render(meta, istioSuccessRateQuery)
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		promql := r.URL.Query()["query"][0]
+		if promql != expected {
+			t.Errorf("\nGot %s \nWanted %s", promql, expected)
+		}
+
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1,"100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
 	if err != nil {
 		t.Fatal(err)
 	}
 
-	expected := `sum(rate(istio_requests_total{reporter="destination",destination_workload_namespace="default",destination_workload=~"podinfo",response_code!~"5.*"}[1m])) / sum(rate(istio_requests_total{reporter="destination",destination_workload_namespace="default",destination_workload=~"podinfo"}[1m])) * 100`
+	observer := &IstioObserver{
+		client: client,
+	}
 
-	if query != expected {
-		t.Errorf("\nGot %s \nWanted %s", query, expected)
+	val, err := observer.GetRequestSuccessRate("podinfo", "default", "1m")
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100 {
+		t.Errorf("Got %v wanted %v", val, 100)
 	}
 }
 
-func Test_IstioRequestDurationQueryRender(t *testing.T) {
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		"podinfo",
-		"default",
-		"1m",
-	}
+func TestIstioObserver_GetRequestDuration(t *testing.T) {
+	expected := `histogram_quantile(0.99,sum(rate(istio_request_duration_seconds_bucket{reporter="destination",destination_workload_namespace="default",destination_workload=~"podinfo"}[1m]))by(le))`
 
-	query, err := render(meta, istioRequestDurationQuery)
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		promql := r.URL.Query()["query"][0]
+		if promql != expected {
+			t.Errorf("\nGot %s \nWanted %s", promql, expected)
+		}
+
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1,"0.100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
 	if err != nil {
 		t.Fatal(err)
 	}
 
-	expected := `histogram_quantile(0.99, sum(rate(istio_request_duration_seconds_bucket{reporter="destination",destination_workload_namespace="default",destination_workload=~"podinfo"}[1m])) by (le))`
+	observer := &IstioObserver{
+		client: client,
+	}
 
-	if query != expected {
-		t.Errorf("\nGot %s \nWanted %s", query, expected)
+	val, err := observer.GetRequestDuration("podinfo", "default", "1m")
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100*time.Millisecond {
+		t.Errorf("Got %v wanted %v", val, 100*time.Millisecond)
 	}
 }

pkg/metrics/linkerd.go

@@ -0,0 +1,76 @@
+package metrics
+
+import (
+	"time"
+)
+
+var linkerdQueries = map[string]string{
+	"request-success-rate": `
+	sum(
+		rate(
+			response_total{
+				namespace="{{ .Namespace }}",
+				deployment=~"{{ .Name }}",
+				classification="failure",
+				direction="inbound"
+			}[{{ .Interval }}]
+		)
+	) 
+	/ 
+	sum(
+		rate(
+			response_total{
+				namespace="{{ .Namespace }}",
+				deployment=~"{{ .Name }}",
+				direction="inbound"
+			}[{{ .Interval }}]
+		)
+	) 
+	* 100`,
+	"request-duration": `
+	histogram_quantile(
+		0.99,
+		sum(
+			rate(
+				response_latency_ms_bucket{
+					namespace="{{ .Namespace }}",
+					deployment=~"{{ .Name }}",
+					direction="inbound"
+				}[{{ .Interval }}]
+			)
+		) by (le)
+	)`,
+}
+
+type LinkerdObserver struct {
+	client *PrometheusClient
+}
+
+func (ob *LinkerdObserver) GetRequestSuccessRate(name string, namespace string, interval string) (float64, error) {
+	query, err := ob.client.RenderQuery(name, namespace, interval, linkerdQueries["request-success-rate"])
+	if err != nil {
+		return 0, err
+	}
+
+	value, err := ob.client.RunQuery(query)
+	if err != nil {
+		return 0, err
+	}
+
+	return value, nil
+}
+
+func (ob *LinkerdObserver) GetRequestDuration(name string, namespace string, interval string) (time.Duration, error) {
+	query, err := ob.client.RenderQuery(name, namespace, interval, linkerdQueries["request-duration"])
+	if err != nil {
+		return 0, err
+	}
+
+	value, err := ob.client.RunQuery(query)
+	if err != nil {
+		return 0, err
+	}
+
+	ms := time.Duration(int64(value)) * time.Millisecond
+	return ms, nil
+}

pkg/metrics/linkerd_test.go

@@ -0,0 +1,74 @@
+package metrics
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"testing"
+	"time"
+)
+
+func TestLinkerdObserver_GetRequestSuccessRate(t *testing.T) {
+	expected := `sum(rate(response_total{namespace="default",deployment=~"podinfo",classification="failure",direction="inbound"}[1m]))/sum(rate(response_total{namespace="default",deployment=~"podinfo",direction="inbound"}[1m]))*100`
+
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		promql := r.URL.Query()["query"][0]
+		if promql != expected {
+			t.Errorf("\nGot %s \nWanted %s", promql, expected)
+		}
+
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1,"100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	observer := &LinkerdObserver{
+		client: client,
+	}
+
+	val, err := observer.GetRequestSuccessRate("podinfo", "default", "1m")
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100 {
+		t.Errorf("Got %v wanted %v", val, 100)
+	}
+}
+
+func TestLinkerdObserver_GetRequestDuration(t *testing.T) {
+	expected := `histogram_quantile(0.99,sum(rate(response_latency_ms_bucket{namespace="default",deployment=~"podinfo",direction="inbound"}[1m]))by(le))`
+
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		promql := r.URL.Query()["query"][0]
+		if promql != expected {
+			t.Errorf("\nGot %s \nWanted %s", promql, expected)
+		}
+
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1,"100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	observer := &LinkerdObserver{
+		client: client,
+	}
+
+	val, err := observer.GetRequestDuration("podinfo", "default", "1m")
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100*time.Millisecond {
+		t.Errorf("Got %v wanted %v", val, 100*time.Millisecond)
+	}
+}

pkg/metrics/nginx.go

@@ -1,122 +1,80 @@
 package metrics
 
 import (
-	"fmt"
-	"net/url"
-	"strconv"
 	"time"
 )
 
-const nginxSuccessRateQuery = `
-sum(rate(
-nginx_ingress_controller_requests{namespace="{{ .Namespace }}",
-ingress="{{ .Name }}",
-status!~"5.*"}
-[{{ .Interval }}])) 
-/ 
-sum(rate(
-nginx_ingress_controller_requests{namespace="{{ .Namespace }}",
-ingress="{{ .Name }}"}
-[{{ .Interval }}])) 
-* 100
-`
-
-// GetNginxSuccessRate returns the requests success rate (non 5xx) using nginx_ingress_controller_requests metric
-func (c *Observer) GetNginxSuccessRate(name string, namespace string, metric string, interval string) (float64, error) {
-	if c.metricsServer == "fake" {
-		return 100, nil
-	}
-
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		name,
-		namespace,
-		interval,
-	}
-
-	query, err := render(meta, nginxSuccessRateQuery)
-	if err != nil {
-		return 0, err
-	}
-
-	var rate *float64
-	querySt := url.QueryEscape(query)
-	result, err := c.queryMetric(querySt)
-	if err != nil {
-		return 0, err
-	}
-
-	for _, v := range result.Data.Result {
-		metricValue := v.Value[1]
-		switch metricValue.(type) {
-		case string:
-			f, err := strconv.ParseFloat(metricValue.(string), 64)
-			if err != nil {
-				return 0, err
-			}
-			rate = &f
-		}
-	}
-	if rate == nil {
-		return 0, fmt.Errorf("no values found for metric %s", metric)
-	}
-	return *rate, nil
+var nginxQueries = map[string]string{
+	"request-success-rate": `
+	sum(
+		rate(
+			nginx_ingress_controller_requests{
+				namespace="{{ .Namespace }}",
+				ingress="{{ .Name }}",
+				status!~"5.*"
+			}[{{ .Interval }}]
+		)
+	) 
+	/ 
+	sum(
+		rate(
+			nginx_ingress_controller_requests{
+				namespace="{{ .Namespace }}",
+				ingress="{{ .Name }}"
+			}[{{ .Interval }}]
+		)
+	) 
+	* 100`,
+	"request-duration": `
+	sum(
+		rate(
+			nginx_ingress_controller_ingress_upstream_latency_seconds_sum{
+				namespace="{{ .Namespace }}",
+				ingress="{{ .Name }}"
+			}[{{ .Interval }}]
+		)
+	) 
+	/ 
+	sum(
+		rate(
+			nginx_ingress_controller_ingress_upstream_latency_seconds_count{
+				namespace="{{ .Namespace }}",
+				ingress="{{ .Name }}"
+			}[{{ .Interval }}]
+		)
+	) 
+	* 1000`,
 }
 
-const nginxRequestDurationQuery = `
-sum(rate(
-nginx_ingress_controller_ingress_upstream_latency_seconds_sum{namespace="{{ .Namespace }}",
-ingress="{{ .Name }}"}[{{ .Interval }}])) 
-/
-sum(rate(nginx_ingress_controller_ingress_upstream_latency_seconds_count{namespace="{{ .Namespace }}",
-ingress="{{ .Name }}"}[{{ .Interval }}])) * 1000
-`
+type NginxObserver struct {
+	client *PrometheusClient
+}
 
-// GetNginxRequestDuration returns the avg requests latency using nginx_ingress_controller_ingress_upstream_latency_seconds_sum metric
-func (c *Observer) GetNginxRequestDuration(name string, namespace string, metric string, interval string) (time.Duration, error) {
-	if c.metricsServer == "fake" {
-		return 1, nil
-	}
-
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		name,
-		namespace,
-		interval,
-	}
-
-	query, err := render(meta, nginxRequestDurationQuery)
+func (ob *NginxObserver) GetRequestSuccessRate(name string, namespace string, interval string) (float64, error) {
+	query, err := ob.client.RenderQuery(name, namespace, interval, nginxQueries["request-success-rate"])
 	if err != nil {
 		return 0, err
 	}
 
-	var rate *float64
-	querySt := url.QueryEscape(query)
-	result, err := c.queryMetric(querySt)
+	value, err := ob.client.RunQuery(query)
 	if err != nil {
 		return 0, err
 	}
 
-	for _, v := range result.Data.Result {
-		metricValue := v.Value[1]
-		switch metricValue.(type) {
-		case string:
-			f, err := strconv.ParseFloat(metricValue.(string), 64)
-			if err != nil {
-				return 0, err
-			}
-			rate = &f
-		}
+	return value, nil
+}
+
+func (ob *NginxObserver) GetRequestDuration(name string, namespace string, interval string) (time.Duration, error) {
+	query, err := ob.client.RenderQuery(name, namespace, interval, nginxQueries["request-duration"])
+	if err != nil {
+		return 0, err
 	}
-	if rate == nil {
-		return 0, fmt.Errorf("no values found for metric %s", metric)
+
+	value, err := ob.client.RunQuery(query)
+	if err != nil {
+		return 0, err
 	}
-	ms := time.Duration(int64(*rate)) * time.Millisecond
+
+	ms := time.Duration(int64(value)) * time.Millisecond
 	return ms, nil
 }

pkg/metrics/nginx_test.go

@@ -1,51 +1,74 @@
 package metrics
 
 import (
+	"net/http"
+	"net/http/httptest"
 	"testing"
+	"time"
 )
 
-func Test_NginxSuccessRateQueryRender(t *testing.T) {
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		"podinfo",
-		"nginx",
-		"1m",
-	}
+func TestNginxObserver_GetRequestSuccessRate(t *testing.T) {
+	expected := `sum(rate(nginx_ingress_controller_requests{namespace="nginx",ingress="podinfo",status!~"5.*"}[1m]))/sum(rate(nginx_ingress_controller_requests{namespace="nginx",ingress="podinfo"}[1m]))*100`
 
-	query, err := render(meta, nginxSuccessRateQuery)
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		promql := r.URL.Query()["query"][0]
+		if promql != expected {
+			t.Errorf("\nGot %s \nWanted %s", promql, expected)
+		}
+
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1,"100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
 	if err != nil {
 		t.Fatal(err)
 	}
 
-	expected := `sum(rate(nginx_ingress_controller_requests{namespace="nginx",ingress="podinfo",status!~"5.*"}[1m])) / sum(rate(nginx_ingress_controller_requests{namespace="nginx",ingress="podinfo"}[1m])) * 100`
+	observer := &NginxObserver{
+		client: client,
+	}
 
-	if query != expected {
-		t.Errorf("\nGot %s \nWanted %s", query, expected)
+	val, err := observer.GetRequestSuccessRate("podinfo", "nginx", "1m")
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100 {
+		t.Errorf("Got %v wanted %v", val, 100)
 	}
 }
 
-func Test_NginxRequestDurationQueryRender(t *testing.T) {
-	meta := struct {
-		Name      string
-		Namespace string
-		Interval  string
-	}{
-		"podinfo",
-		"nginx",
-		"1m",
-	}
+func TestNginxObserver_GetRequestDuration(t *testing.T) {
+	expected := `sum(rate(nginx_ingress_controller_ingress_upstream_latency_seconds_sum{namespace="nginx",ingress="podinfo"}[1m]))/sum(rate(nginx_ingress_controller_ingress_upstream_latency_seconds_count{namespace="nginx",ingress="podinfo"}[1m]))*1000`
 
-	query, err := render(meta, nginxRequestDurationQuery)
+	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		promql := r.URL.Query()["query"][0]
+		if promql != expected {
+			t.Errorf("\nGot %s \nWanted %s", promql, expected)
+		}
+
+		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1,"100"]}]}}`
+		w.Write([]byte(json))
+	}))
+	defer ts.Close()
+
+	client, err := NewPrometheusClient(ts.URL, time.Second)
 	if err != nil {
 		t.Fatal(err)
 	}
 
-	expected := `sum(rate(nginx_ingress_controller_ingress_upstream_latency_seconds_sum{namespace="nginx",ingress="podinfo"}[1m])) /sum(rate(nginx_ingress_controller_ingress_upstream_latency_seconds_count{namespace="nginx",ingress="podinfo"}[1m])) * 1000`
+	observer := &NginxObserver{
+		client: client,
+	}
 
-	if query != expected {
-		t.Errorf("\nGot %s \nWanted %s", query, expected)
+	val, err := observer.GetRequestDuration("podinfo", "nginx", "1m")
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if val != 100*time.Millisecond {
+		t.Errorf("Got %v wanted %v", val, 100*time.Millisecond)
 	}
 }

pkg/metrics/observer.go

@@ -1,186 +1,10 @@
 package metrics
 
 import (
-	"bufio"
-	"bytes"
-	"context"
-	"encoding/json"
-	"fmt"
-	"io/ioutil"
-	"net/http"
-	"net/url"
-	"strconv"
-	"strings"
-	"text/template"
 	"time"
 )
 
-// Observer is used to query Prometheus
-type Observer struct {
-	metricsServer string
-}
-
-type vectorQueryResponse struct {
-	Data struct {
-		Result []struct {
-			Metric struct {
-				Code string `json:"response_code"`
-				Name string `json:"destination_workload"`
-			}
-			Value []interface{} `json:"value"`
-		}
-	}
-}
-
-// NewObserver creates a new observer
-func NewObserver(metricsServer string) Observer {
-	return Observer{
-		metricsServer: metricsServer,
-	}
-}
-
-// GetMetricsServer returns the Prometheus URL
-func (c *Observer) GetMetricsServer() string {
-	return c.metricsServer
-}
-
-func (c *Observer) queryMetric(query string) (*vectorQueryResponse, error) {
-	promURL, err := url.Parse(c.metricsServer)
-	if err != nil {
-		return nil, err
-	}
-
-	u, err := url.Parse(fmt.Sprintf("./api/v1/query?query=%s", query))
-	if err != nil {
-		return nil, err
-	}
-
-	u = promURL.ResolveReference(u)
-
-	req, err := http.NewRequest("GET", u.String(), nil)
-	if err != nil {
-		return nil, err
-	}
-
-	ctx, cancel := context.WithTimeout(req.Context(), 5*time.Second)
-	defer cancel()
-
-	r, err := http.DefaultClient.Do(req.WithContext(ctx))
-	if err != nil {
-		return nil, err
-	}
-	defer r.Body.Close()
-
-	b, err := ioutil.ReadAll(r.Body)
-	if err != nil {
-		return nil, fmt.Errorf("error reading body: %s", err.Error())
-	}
-
-	if 400 <= r.StatusCode {
-		return nil, fmt.Errorf("error response: %s", string(b))
-	}
-
-	var values vectorQueryResponse
-	err = json.Unmarshal(b, &values)
-	if err != nil {
-		return nil, fmt.Errorf("error unmarshaling result: %s, '%s'", err.Error(), string(b))
-	}
-
-	return &values, nil
-}
-
-// GetScalar runs the promql query and returns the first value found
-func (c *Observer) GetScalar(query string) (float64, error) {
-	if c.metricsServer == "fake" {
-		return 100, nil
-	}
-
-	query = strings.Replace(query, "\n", "", -1)
-	query = strings.Replace(query, " ", "", -1)
-
-	var value *float64
-
-	querySt := url.QueryEscape(query)
-	result, err := c.queryMetric(querySt)
-	if err != nil {
-		return 0, err
-	}
-
-	for _, v := range result.Data.Result {
-		metricValue := v.Value[1]
-		switch metricValue.(type) {
-		case string:
-			f, err := strconv.ParseFloat(metricValue.(string), 64)
-			if err != nil {
-				return 0, err
-			}
-			value = &f
-		}
-	}
-	if value == nil {
-		return 0, fmt.Errorf("no values found for query %s", query)
-	}
-	return *value, nil
-}
-
-// CheckMetricsServer call Prometheus status endpoint and returns an error if
-// the API is unreachable
-func CheckMetricsServer(address string) (bool, error) {
-	promURL, err := url.Parse(address)
-	if err != nil {
-		return false, err
-	}
-
-	u, err := url.Parse("./api/v1/status/flags")
-	if err != nil {
-		return false, err
-	}
-
-	u = promURL.ResolveReference(u)
-
-	req, err := http.NewRequest("GET", u.String(), nil)
-	if err != nil {
-		return false, err
-	}
-
-	ctx, cancel := context.WithTimeout(req.Context(), 5*time.Second)
-	defer cancel()
-
-	r, err := http.DefaultClient.Do(req.WithContext(ctx))
-	if err != nil {
-		return false, err
-	}
-	defer r.Body.Close()
-
-	b, err := ioutil.ReadAll(r.Body)
-	if err != nil {
-		return false, fmt.Errorf("error reading body: %s", err.Error())
-	}
-
-	if 400 <= r.StatusCode {
-		return false, fmt.Errorf("error response: %s", string(b))
-	}
-
-	return true, nil
-}
-
-func render(meta interface{}, tmpl string) (string, error) {
-	t, err := template.New("tmpl").Parse(tmpl)
-	if err != nil {
-		return "", err
-	}
-	var data bytes.Buffer
-	b := bufio.NewWriter(&data)
-
-	if err := t.Execute(b, meta); err != nil {
-		return "", err
-	}
-	err = b.Flush()
-	if err != nil {
-		return "", err
-	}
-
-	res := strings.ReplaceAll(data.String(), "\n", "")
-
-	return res, nil
+type Interface interface {
+	GetRequestSuccessRate(name string, namespace string, interval string) (float64, error)
+	GetRequestDuration(name string, namespace string, interval string) (time.Duration, error)
 }

pkg/metrics/observer_test.go

@@ -1,119 +0,0 @@
-package metrics
-
-import (
-	"net/http"
-	"net/http/httptest"
-	"testing"
-	"time"
-)
-
-func TestCanaryObserver_GetEnvoySuccessRate(t *testing.T) {
-	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1545905245.458,"100"]}]}}`
-		w.Write([]byte(json))
-	}))
-	defer ts.Close()
-
-	observer := NewObserver(ts.URL)
-
-	val, err := observer.GetEnvoySuccessRate("podinfo", "default", "envoy_cluster_upstream_rq", "1m")
-	if err != nil {
-		t.Fatal(err.Error())
-	}
-
-	if val != 100 {
-		t.Errorf("Got %v wanted %v", val, 100)
-	}
-
-}
-
-func TestCanaryObserver_GetEnvoyRequestDuration(t *testing.T) {
-	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1545905245.596,"200"]}]}}`
-		w.Write([]byte(json))
-	}))
-	defer ts.Close()
-
-	observer := NewObserver(ts.URL)
-
-	val, err := observer.GetEnvoyRequestDuration("podinfo", "default", "envoy_cluster_upstream_rq_time_bucket", "1m")
-	if err != nil {
-		t.Fatal(err.Error())
-	}
-
-	if val != 200*time.Millisecond {
-		t.Errorf("Got %v wanted %v", val, 200*time.Millisecond)
-	}
-}
-
-func TestCanaryObserver_GetIstioSuccessRate(t *testing.T) {
-	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1545905245.458,"100"]}]}}`
-		w.Write([]byte(json))
-	}))
-	defer ts.Close()
-
-	observer := NewObserver(ts.URL)
-
-	val, err := observer.GetIstioSuccessRate("podinfo", "default", "istio_requests_total", "1m")
-	if err != nil {
-		t.Fatal(err.Error())
-	}
-
-	if val != 100 {
-		t.Errorf("Got %v wanted %v", val, 100)
-	}
-
-}
-
-func TestCanaryObserver_GetIstioRequestDuration(t *testing.T) {
-	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		json := `{"status":"success","data":{"resultType":"vector","result":[{"metric":{},"value":[1545905245.596,"0.2"]}]}}`
-		w.Write([]byte(json))
-	}))
-	defer ts.Close()
-
-	observer := NewObserver(ts.URL)
-
-	val, err := observer.GetIstioRequestDuration("podinfo", "default", "istio_request_duration_seconds_bucket", "1m")
-	if err != nil {
-		t.Fatal(err.Error())
-	}
-
-	if val != 200*time.Millisecond {
-		t.Errorf("Got %v wanted %v", val, 200*time.Millisecond)
-	}
-}
-
-func TestCheckMetricsServer(t *testing.T) {
-	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		json := `{"status":"success","data":{"config.file":"/etc/prometheus/prometheus.yml"}}`
-		w.Write([]byte(json))
-	}))
-	defer ts.Close()
-
-	ok, err := CheckMetricsServer(ts.URL)
-	if err != nil {
-		t.Fatal(err.Error())
-	}
-
-	if !ok {
-		t.Errorf("Got %v wanted %v", ok, true)
-	}
-}
-
-func TestCheckMetricsServer_Offline(t *testing.T) {
-	ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		w.WriteHeader(http.StatusBadGateway)
-	}))
-	defer ts.Close()
-
-	ok, err := CheckMetricsServer(ts.URL)
-	if err == nil {
-		t.Errorf("Got no error wanted %v", http.StatusBadGateway)
-	}
-
-	if ok {
-		t.Errorf("Got %v wanted %v", ok, false)
-	}
-}

pkg/router/factory.go

@@ -56,12 +56,27 @@ func (factory *Factory) MeshRouter(provider string) Interface {
 			kubeClient:    factory.kubeClient,
 			appmeshClient: factory.meshClient,
 		}
+	case strings.HasPrefix(provider, "smi:"):
+		mesh := strings.TrimPrefix(provider, "smi:")
+		return &SmiRouter{
+			logger:        factory.logger,
+			flaggerClient: factory.flaggerClient,
+			kubeClient:    factory.kubeClient,
+			smiClient:     factory.meshClient,
+			targetMesh:    mesh,
+		}
 	case strings.HasPrefix(provider, "supergloo"):
 		supergloo, err := NewSuperglooRouter(context.TODO(), provider, factory.flaggerClient, factory.logger, factory.kubeConfig)
 		if err != nil {
 			panic("failed creating supergloo client")
 		}
 		return supergloo
+	case strings.HasPrefix(provider, "gloo"):
+		gloo, err := NewGlooRouter(context.TODO(), provider, factory.flaggerClient, factory.logger, factory.kubeConfig)
+		if err != nil {
+			panic("failed creating gloo client")
+		}
+		return gloo
 	default:
 		return &IstioRouter{
 			logger:        factory.logger,

pkg/router/gloo.go

@@ -0,0 +1,187 @@
+package router
+
+import (
+	"context"
+	"fmt"
+	"strings"
+
+	solokitclients "github.com/solo-io/solo-kit/pkg/api/v1/clients"
+	"github.com/solo-io/solo-kit/pkg/api/v1/clients/factory"
+	"github.com/solo-io/solo-kit/pkg/api/v1/clients/kube"
+	crdv1 "github.com/solo-io/solo-kit/pkg/api/v1/clients/kube/crd/solo.io/v1"
+	solokitcore "github.com/solo-io/solo-kit/pkg/api/v1/resources/core"
+	solokiterror "github.com/solo-io/solo-kit/pkg/errors"
+
+	gloov1 "github.com/solo-io/gloo/projects/gloo/pkg/api/v1"
+	flaggerv1 "github.com/weaveworks/flagger/pkg/apis/flagger/v1alpha3"
+	clientset "github.com/weaveworks/flagger/pkg/client/clientset/versioned"
+	"go.uber.org/zap"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/runtime/schema"
+	"k8s.io/client-go/rest"
+)
+
+// GlooRouter is managing Istio virtual services
+type GlooRouter struct {
+	ugClient            gloov1.UpstreamGroupClient
+	logger              *zap.SugaredLogger
+	upstreamDiscoveryNs string
+}
+
+func NewGlooRouter(ctx context.Context, provider string, flaggerClient clientset.Interface, logger *zap.SugaredLogger, cfg *rest.Config) (*GlooRouter, error) {
+	// TODO if cfg is nil use memory client instead?
+	sharedCache := kube.NewKubeCache(ctx)
+	upstreamGroupClient, err := gloov1.NewUpstreamGroupClient(&factory.KubeResourceClientFactory{
+		Crd:             gloov1.UpstreamGroupCrd,
+		Cfg:             cfg,
+		SharedCache:     sharedCache,
+		SkipCrdCreation: true,
+	})
+	if err != nil {
+		// this should never happen.
+		return nil, fmt.Errorf("creating UpstreamGroup client %v", err)
+	}
+	if err := upstreamGroupClient.Register(); err != nil {
+		return nil, err
+	}
+	upstreamDiscoveryNs := ""
+	if strings.HasPrefix(provider, "gloo:") {
+		upstreamDiscoveryNs = strings.TrimPrefix(provider, "gloo:")
+	}
+
+	return NewGlooRouterWithClient(ctx, upstreamGroupClient, upstreamDiscoveryNs, logger), nil
+}
+
+func NewGlooRouterWithClient(ctx context.Context, routingRuleClient gloov1.UpstreamGroupClient, upstreamDiscoveryNs string, logger *zap.SugaredLogger) *GlooRouter {
+
+	if upstreamDiscoveryNs == "" {
+		upstreamDiscoveryNs = "gloo-system"
+	}
+	return &GlooRouter{ugClient: routingRuleClient, logger: logger, upstreamDiscoveryNs: upstreamDiscoveryNs}
+}
+
+// Reconcile creates or updates the Istio virtual service
+func (gr *GlooRouter) Reconcile(canary *flaggerv1.Canary) error {
+	// do we have routes already?
+	if _, _, err := gr.GetRoutes(canary); err == nil {
+		// we have routes, no need to do anything else
+		return nil
+	} else if solokiterror.IsNotExist(err) {
+		return gr.SetRoutes(canary, 100, 0)
+	} else {
+		return err
+	}
+}
+
+// GetRoutes returns the destinations weight for primary and canary
+func (gr *GlooRouter) GetRoutes(canary *flaggerv1.Canary) (
+	primaryWeight int,
+	canaryWeight int,
+	err error,
+) {
+	targetName := canary.Spec.TargetRef.Name
+	var ug *gloov1.UpstreamGroup
+	ug, err = gr.ugClient.Read(canary.Namespace, targetName, solokitclients.ReadOpts{})
+	if err != nil {
+		return
+	}
+
+	dests := ug.GetDestinations()
+	for _, dest := range dests {
+		if dest.GetDestination().GetUpstream().Name == upstreamName(canary.Namespace, fmt.Sprintf("%s-primary", targetName), canary.Spec.Service.Port) {
+			primaryWeight = int(dest.Weight)
+		}
+		if dest.GetDestination().GetUpstream().Name == upstreamName(canary.Namespace, fmt.Sprintf("%s-canary", targetName), canary.Spec.Service.Port) {
+			canaryWeight = int(dest.Weight)
+		}
+	}
+
+	if primaryWeight == 0 && canaryWeight == 0 {
+		err = fmt.Errorf("RoutingRule %s.%s does not contain routes for %s-primary and %s-canary",
+			targetName, canary.Namespace, targetName, targetName)
+	}
+
+	return
+}
+
+// SetRoutes updates the destinations weight for primary and canary
+func (gr *GlooRouter) SetRoutes(
+	canary *flaggerv1.Canary,
+	primaryWeight int,
+	canaryWeight int,
+) error {
+	targetName := canary.Spec.TargetRef.Name
+
+	if primaryWeight == 0 && canaryWeight == 0 {
+		return fmt.Errorf("RoutingRule %s.%s update failed: no valid weights", targetName, canary.Namespace)
+	}
+
+	destinations := []*gloov1.WeightedDestination{}
+	destinations = append(destinations, &gloov1.WeightedDestination{
+		Destination: &gloov1.Destination{
+			Upstream: solokitcore.ResourceRef{
+				Name:      upstreamName(canary.Namespace, fmt.Sprintf("%s-primary", targetName), canary.Spec.Service.Port),
+				Namespace: gr.upstreamDiscoveryNs,
+			},
+		},
+		Weight: uint32(primaryWeight),
+	})
+
+	destinations = append(destinations, &gloov1.WeightedDestination{
+		Destination: &gloov1.Destination{
+			Upstream: solokitcore.ResourceRef{
+				Name:      upstreamName(canary.Namespace, fmt.Sprintf("%s-canary", targetName), canary.Spec.Service.Port),
+				Namespace: gr.upstreamDiscoveryNs,
+			},
+		},
+		Weight: uint32(canaryWeight),
+	})
+
+	upstreamGroup := &gloov1.UpstreamGroup{
+		Metadata: solokitcore.Metadata{
+			Name:      canary.Spec.TargetRef.Name,
+			Namespace: canary.Namespace,
+		},
+		Destinations: destinations,
+	}
+
+	return gr.writeUpstreamGroupRuleForCanary(canary, upstreamGroup)
+}
+
+func (gr *GlooRouter) writeUpstreamGroupRuleForCanary(canary *flaggerv1.Canary, ug *gloov1.UpstreamGroup) error {
+	targetName := canary.Spec.TargetRef.Name
+
+	if oldUg, err := gr.ugClient.Read(ug.Metadata.Namespace, ug.Metadata.Name, solokitclients.ReadOpts{}); err != nil {
+		if solokiterror.IsNotExist(err) {
+			gr.logger.With("canary", fmt.Sprintf("%s.%s", canary.Name, canary.Namespace)).
+				Infof("UpstreamGroup %s created", ug.Metadata.Name)
+		} else {
+			return fmt.Errorf("RoutingRule %s.%s read failed: %v", targetName, canary.Namespace, err)
+		}
+	} else {
+		ug.Metadata.ResourceVersion = oldUg.Metadata.ResourceVersion
+		// if the old and the new one are equal, no need to do anything.
+		oldUg.Status = solokitcore.Status{}
+		if oldUg.Equal(ug) {
+			return nil
+		}
+	}
+
+	kubeWriteOpts := &kube.KubeWriteOpts{
+		PreWriteCallback: func(r *crdv1.Resource) {
+			r.ObjectMeta.OwnerReferences = []metav1.OwnerReference{
+				*metav1.NewControllerRef(canary, schema.GroupVersionKind{
+					Group:   flaggerv1.SchemeGroupVersion.Group,
+					Version: flaggerv1.SchemeGroupVersion.Version,
+					Kind:    flaggerv1.CanaryKind,
+				}),
+			}
+		},
+	}
+	writeOpts := solokitclients.WriteOpts{OverwriteExisting: true, StorageWriteOpts: kubeWriteOpts}
+	_, err := gr.ugClient.Write(ug, writeOpts)
+	if err != nil {
+		return fmt.Errorf("UpstreamGroup %s.%s update failed: %v", targetName, canary.Namespace, err)
+	}
+	return nil
+}

pkg/router/gloo_test.go

@@ -0,0 +1,142 @@
+package router
+
+import (
+	"context"
+	"fmt"
+	"testing"
+
+	gloov1 "github.com/solo-io/gloo/projects/gloo/pkg/api/v1"
+	solokitclients "github.com/solo-io/solo-kit/pkg/api/v1/clients"
+	"github.com/solo-io/solo-kit/pkg/api/v1/clients/factory"
+	solokitmemory "github.com/solo-io/solo-kit/pkg/api/v1/clients/memory"
+)
+
+func TestGlooRouter_Sync(t *testing.T) {
+	mocks := setupfakeClients()
+
+	upstreamGroupClient, err := gloov1.NewUpstreamGroupClient(&factory.MemoryResourceClientFactory{
+		Cache: solokitmemory.NewInMemoryResourceCache(),
+	})
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+	if err := upstreamGroupClient.Register(); err != nil {
+		t.Fatal(err.Error())
+	}
+	router := NewGlooRouterWithClient(context.TODO(), upstreamGroupClient, "gloo-system", mocks.logger)
+	err = router.Reconcile(mocks.canary)
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	// test insert
+	ug, err := upstreamGroupClient.Read("default", "podinfo", solokitclients.ReadOpts{})
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+	dests := ug.GetDestinations()
+	if len(dests) != 2 {
+		t.Errorf("Got Destinations %v wanted %v", len(dests), 2)
+	}
+
+	if dests[0].Weight != 100 {
+		t.Errorf("Primary weight should is %v wanted 100", dests[0].Weight)
+	}
+	if dests[1].Weight != 0 {
+		t.Errorf("Canary weight should is %v wanted 0", dests[0].Weight)
+	}
+
+}
+
+func TestGlooRouter_SetRoutes(t *testing.T) {
+
+	mocks := setupfakeClients()
+
+	upstreamGroupClient, err := gloov1.NewUpstreamGroupClient(&factory.MemoryResourceClientFactory{
+		Cache: solokitmemory.NewInMemoryResourceCache(),
+	})
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+	if err := upstreamGroupClient.Register(); err != nil {
+		t.Fatal(err.Error())
+	}
+	router := NewGlooRouterWithClient(context.TODO(), upstreamGroupClient, "gloo-system", mocks.logger)
+
+	err = router.Reconcile(mocks.canary)
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	p, c, err := router.GetRoutes(mocks.canary)
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	p = 50
+	c = 50
+
+	err = router.SetRoutes(mocks.canary, p, c)
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	ug, err := upstreamGroupClient.Read("default", "podinfo", solokitclients.ReadOpts{})
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	var pRoute *gloov1.WeightedDestination
+	var cRoute *gloov1.WeightedDestination
+	targetName := mocks.canary.Spec.TargetRef.Name
+
+	for _, dest := range ug.GetDestinations() {
+		if dest.GetDestination().GetUpstream().Name == upstreamName(mocks.canary.Namespace, fmt.Sprintf("%s-primary", targetName), mocks.canary.Spec.Service.Port) {
+			pRoute = dest
+		}
+		if dest.GetDestination().GetUpstream().Name == upstreamName(mocks.canary.Namespace, fmt.Sprintf("%s-canary", targetName), mocks.canary.Spec.Service.Port) {
+			cRoute = dest
+		}
+	}
+
+	if pRoute.Weight != uint32(p) {
+		t.Errorf("Got primary weight %v wanted %v", pRoute.Weight, p)
+	}
+
+	if cRoute.Weight != uint32(c) {
+		t.Errorf("Got canary weight %v wanted %v", cRoute.Weight, c)
+	}
+
+}
+
+func TestGlooRouter_GetRoutes(t *testing.T) {
+	mocks := setupfakeClients()
+
+	upstreamGroupClient, err := gloov1.NewUpstreamGroupClient(&factory.MemoryResourceClientFactory{
+		Cache: solokitmemory.NewInMemoryResourceCache(),
+	})
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+	if err := upstreamGroupClient.Register(); err != nil {
+		t.Fatal(err.Error())
+	}
+	router := NewGlooRouterWithClient(context.TODO(), upstreamGroupClient, "gloo-system", mocks.logger)
+	err = router.Reconcile(mocks.canary)
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	p, c, err := router.GetRoutes(mocks.canary)
+	if err != nil {
+		t.Fatal(err.Error())
+	}
+
+	if p != 100 {
+		t.Errorf("Got primary weight %v wanted %v", p, 100)
+	}
+
+	if c != 0 {
+		t.Errorf("Got canary weight %v wanted %v", c, 0)
+	}
+}

pkg/router/smi.go

@@ -0,0 +1,190 @@
+package router
+
+import (
+	"encoding/json"
+	"fmt"
+
+	"github.com/google/go-cmp/cmp"
+	"github.com/google/go-cmp/cmp/cmpopts"
+	flaggerv1 "github.com/weaveworks/flagger/pkg/apis/flagger/v1alpha3"
+	smiv1 "github.com/weaveworks/flagger/pkg/apis/smi/v1alpha1"
+	clientset "github.com/weaveworks/flagger/pkg/client/clientset/versioned"
+	"go.uber.org/zap"
+	"k8s.io/apimachinery/pkg/api/errors"
+	"k8s.io/apimachinery/pkg/api/resource"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/runtime/schema"
+	"k8s.io/client-go/kubernetes"
+)
+
+type SmiRouter struct {
+	kubeClient    kubernetes.Interface
+	flaggerClient clientset.Interface
+	smiClient     clientset.Interface
+	logger        *zap.SugaredLogger
+	targetMesh    string
+}
+
+// Reconcile creates or updates the SMI traffic split
+func (sr *SmiRouter) Reconcile(canary *flaggerv1.Canary) error {
+	targetName := canary.Spec.TargetRef.Name
+	canaryName := fmt.Sprintf("%s-canary", targetName)
+	primaryName := fmt.Sprintf("%s-primary", targetName)
+
+	var host string
+	if len(canary.Spec.Service.Hosts) > 0 {
+		host = canary.Spec.Service.Hosts[0]
+	} else {
+		host = targetName
+	}
+
+	tsSpec := smiv1.TrafficSplitSpec{
+		Service: host,
+		Backends: []smiv1.TrafficSplitBackend{
+			{
+				Service: canaryName,
+				Weight:  resource.NewQuantity(0, resource.DecimalExponent),
+			},
+			{
+				Service: primaryName,
+				Weight:  resource.NewQuantity(100, resource.DecimalExponent),
+			},
+		},
+	}
+
+	ts, err := sr.smiClient.SplitV1alpha1().TrafficSplits(canary.Namespace).Get(targetName, metav1.GetOptions{})
+	// create traffic split
+	if errors.IsNotFound(err) {
+		t := &smiv1.TrafficSplit{
+			ObjectMeta: metav1.ObjectMeta{
+				Name:      targetName,
+				Namespace: canary.Namespace,
+				OwnerReferences: []metav1.OwnerReference{
+					*metav1.NewControllerRef(canary, schema.GroupVersionKind{
+						Group:   flaggerv1.SchemeGroupVersion.Group,
+						Version: flaggerv1.SchemeGroupVersion.Version,
+						Kind:    flaggerv1.CanaryKind,
+					}),
+				},
+				Annotations: sr.makeAnnotations(canary.Spec.Service.Gateways),
+			},
+			Spec: tsSpec,
+		}
+
+		_, err := sr.smiClient.SplitV1alpha1().TrafficSplits(canary.Namespace).Create(t)
+		if err != nil {
+			return err
+		}
+
+		sr.logger.With("canary", fmt.Sprintf("%s.%s", canary.Name, canary.Namespace)).
+			Infof("TrafficSplit %s.%s created", t.GetName(), canary.Namespace)
+		return nil
+	}
+
+	if err != nil {
+		return fmt.Errorf("traffic split %s query error %v", targetName, err)
+	}
+
+	// update traffic split
+	if diff := cmp.Diff(tsSpec, ts.Spec, cmpopts.IgnoreTypes(resource.Quantity{})); diff != "" {
+		tsClone := ts.DeepCopy()
+		tsClone.Spec = tsSpec
+
+		_, err := sr.smiClient.SplitV1alpha1().TrafficSplits(canary.Namespace).Update(tsClone)
+		if err != nil {
+			return fmt.Errorf("TrafficSplit %s update error %v", targetName, err)
+		}
+
+		sr.logger.With("canary", fmt.Sprintf("%s.%s", canary.Name, canary.Namespace)).
+			Infof("TrafficSplit %s.%s updated", targetName, canary.Namespace)
+		return nil
+	}
+
+	return nil
+}
+
+// GetRoutes returns the destinations weight for primary and canary
+func (sr *SmiRouter) GetRoutes(canary *flaggerv1.Canary) (
+	primaryWeight int,
+	canaryWeight int,
+	err error,
+) {
+	targetName := canary.Spec.TargetRef.Name
+	canaryName := fmt.Sprintf("%s-canary", targetName)
+	primaryName := fmt.Sprintf("%s-primary", targetName)
+	ts, err := sr.smiClient.SplitV1alpha1().TrafficSplits(canary.Namespace).Get(targetName, metav1.GetOptions{})
+	if err != nil {
+		if errors.IsNotFound(err) {
+			err = fmt.Errorf("TrafficSplit %s.%s not found", targetName, canary.Namespace)
+			return
+		}
+		err = fmt.Errorf("TrafficSplit %s.%s query error %v", targetName, canary.Namespace, err)
+		return
+	}
+
+	for _, r := range ts.Spec.Backends {
+		w, _ := r.Weight.AsInt64()
+		if r.Service == primaryName {
+			primaryWeight = int(w)
+		}
+		if r.Service == canaryName {
+			canaryWeight = int(w)
+		}
+	}
+
+	if primaryWeight == 0 && canaryWeight == 0 {
+		err = fmt.Errorf("TrafficSplit %s.%s does not contain routes for %s and %s",
+			targetName, canary.Namespace, primaryName, canaryName)
+	}
+
+	return
+}
+
+// SetRoutes updates the destinations weight for primary and canary
+func (sr *SmiRouter) SetRoutes(
+	canary *flaggerv1.Canary,
+	primaryWeight int,
+	canaryWeight int,
+) error {
+	targetName := canary.Spec.TargetRef.Name
+	canaryName := fmt.Sprintf("%s-canary", targetName)
+	primaryName := fmt.Sprintf("%s-primary", targetName)
+	ts, err := sr.smiClient.SplitV1alpha1().TrafficSplits(canary.Namespace).Get(targetName, metav1.GetOptions{})
+	if err != nil {
+		if errors.IsNotFound(err) {
+			return fmt.Errorf("TrafficSplit %s.%s not found", targetName, canary.Namespace)
+
+		}
+		return fmt.Errorf("TrafficSplit %s.%s query error %v", targetName, canary.Namespace, err)
+	}
+
+	backends := []smiv1.TrafficSplitBackend{
+		{
+			Service: canaryName,
+			Weight:  resource.NewQuantity(int64(canaryWeight), resource.DecimalExponent),
+		},
+		{
+			Service: primaryName,
+			Weight:  resource.NewQuantity(int64(primaryWeight), resource.DecimalExponent),
+		},
+	}
+
+	tsClone := ts.DeepCopy()
+	tsClone.Spec.Backends = backends
+
+	_, err = sr.smiClient.SplitV1alpha1().TrafficSplits(canary.Namespace).Update(tsClone)
+	if err != nil {
+		return fmt.Errorf("TrafficSplit %s update error %v", targetName, err)
+	}
+
+	return nil
+}
+
+func (sr *SmiRouter) makeAnnotations(gateways []string) map[string]string {
+	res := make(map[string]string)
+	if sr.targetMesh == "istio" && len(gateways) > 0 {
+		g, _ := json.Marshal(gateways)
+		res["VirtualService.v1alpha3.networking.istio.io/spec.gateways"] = string(g)
+	}
+	return res
+}

pkg/version/version.go

@@ -1,4 +1,4 @@
 package version
 
-var VERSION = "0.13.2"
+var VERSION = "0.14.0"
 var REVISION = "unknown"

test/e2e-gloo-build.sh

@@ -0,0 +1,27 @@
+#!/usr/bin/env bash
+
+set -o errexit
+
+REPO_ROOT=$(git rev-parse --show-toplevel)
+export KUBECONFIG="$(kind get kubeconfig-path --name="kind")"
+
+echo '>>> Building Flagger'
+cd ${REPO_ROOT} && docker build -t test/flagger:latest . -f Dockerfile
+
+echo '>>> Installing Flagger'
+kind load docker-image test/flagger:latest
+
+echo '>>> Installing Flagger'
+helm upgrade -i flagger ${REPO_ROOT}/charts/flagger \
+--wait \
+--namespace gloo-system \
+--set prometheus.install=true \
+--set meshProvider=gloo
+
+# Give flagger permissions for gloo objects
+kubectl create clusterrolebinding flagger-gloo  --clusterrole=gloo-role-gateway --serviceaccount=gloo-system:flagger
+
+kubectl -n gloo-system set image deployment/flagger flagger=test/flagger:latest
+
+kubectl -n gloo-system rollout status deployment/flagger
+kubectl -n gloo-system rollout status deployment/flagger-prometheus

test/e2e-gloo-tests.sh

@@ -0,0 +1,97 @@
+#!/usr/bin/env bash
+
+# This script runs e2e tests for Canary initialization, analysis and promotion
+# Prerequisites: Kubernetes Kind, Helm and NGINX ingress controller
+
+set -o errexit
+
+REPO_ROOT=$(git rev-parse --show-toplevel)
+export KUBECONFIG="$(kind get kubeconfig-path --name="kind")"
+
+echo '>>> Creating test namespace'
+kubectl create namespace test
+
+echo ">>> Downloading Gloo CLI"
+curl -SsL https://github.com/solo-io/gloo/releases/download/v0.13.29/glooctl-linux-amd64 > glooctl
+chmod +x glooctl
+
+echo '>>> Installing load tester'
+kubectl -n test apply -f ${REPO_ROOT}/artifacts/loadtester/
+kubectl -n test rollout status deployment/flagger-loadtester
+
+echo '>>> Initialising canary'
+kubectl apply -f ${REPO_ROOT}/test/e2e-workload.yaml
+./glooctl add route --path-prefix / --upstream-group-name podinfo --upstream-group-namespace test
+
+cat <<EOF | kubectl apply -f -
+apiVersion: flagger.app/v1alpha3
+kind: Canary
+metadata:
+  name: podinfo
+  namespace: test
+spec:
+  targetRef:
+    apiVersion: apps/v1
+    kind: Deployment
+    name: podinfo
+  progressDeadlineSeconds: 60
+  service:
+    port: 9898
+  canaryAnalysis:
+    interval: 15s
+    threshold: 15
+    maxWeight: 30
+    stepWeight: 10
+    metrics:
+    - name: request-success-rate
+      threshold: 99
+      interval: 1m
+    webhooks:
+      - name: load-test
+        url: http://flagger-loadtester.test/
+        timeout: 5s
+        metadata:
+          type: cmd
+          cmd: "hey -z 10m -q 10 -c 2 -host app.example.com http://gateway-proxy.gloo-system"
+          logCmdOutput: "true"
+EOF
+
+echo '>>> Waiting for primary to be ready'
+retries=50
+count=0
+ok=false
+until ${ok}; do
+    kubectl -n test get canary/podinfo | grep 'Initialized' && ok=true || ok=false
+    sleep 5
+    count=$(($count + 1))
+    if [[ ${count} -eq ${retries} ]]; then
+        kubectl -n gloo-system logs deployment/flagger
+        echo "No more retries left"
+        exit 1
+    fi
+done
+
+echo '✔ Canary initialization test passed'
+
+echo '>>> Triggering canary deployment'
+kubectl -n test set image deployment/podinfo podinfod=quay.io/stefanprodan/podinfo:1.4.1
+
+echo '>>> Waiting for canary promotion'
+retries=50
+count=0
+ok=false
+until ${ok}; do
+    kubectl -n test describe deployment/podinfo-primary | grep '1.4.1' && ok=true || ok=false
+    sleep 10
+    kubectl -n gloo-system logs deployment/flagger --tail 1
+    count=$(($count + 1))
+    if [[ ${count} -eq ${retries} ]]; then
+        kubectl -n test describe deployment/podinfo
+        kubectl -n test describe deployment/podinfo-primary
+        kubectl -n gloo-system logs deployment/flagger
+        echo "No more retries left"
+        exit 1
+    fi
+done
+
+echo '✔ Canary promotion test passed'

test/e2e-gloo.sh

@@ -0,0 +1,25 @@
+#!/usr/bin/env bash
+
+set -o errexit
+
+REPO_ROOT=$(git rev-parse --show-toplevel)
+export KUBECONFIG="$(kind get kubeconfig-path --name="kind")"
+
+echo ">>> Installing Helm"
+curl https://raw.githubusercontent.com/kubernetes/helm/master/scripts/get | bash
+
+echo '>>> Installing Tiller'
+kubectl --namespace kube-system create sa tiller
+kubectl create clusterrolebinding tiller-cluster-rule --clusterrole=cluster-admin --serviceaccount=kube-system:tiller
+helm init --service-account tiller --upgrade --wait
+helm repo add gloo https://storage.googleapis.com/solo-public-helm
+
+echo '>>> Installing Gloo'
+helm upgrade -i gloo gloo/gloo --version 0.13.29 \
+--wait \
+--namespace gloo-system \
+--set gatewayProxies.gateway-proxy.service.type=NodePort
+
+kubectl -n gloo-system rollout status deployment/gloo
+kubectl -n gloo-system rollout status deployment/gateway-proxy
+kubectl -n gloo-system get all

test/e2e-smi-istio-build.sh

@@ -0,0 +1,26 @@
+#!/usr/bin/env bash
+
+set -o errexit
+
+REPO_ROOT=$(git rev-parse --show-toplevel)
+export KUBECONFIG="$(kind get kubeconfig-path --name="kind")"
+
+echo '>>> Building Flagger'
+cd ${REPO_ROOT} && docker build -t test/flagger:latest . -f Dockerfile
+
+kind load docker-image test/flagger:latest
+
+echo '>>> Installing Flagger'
+helm upgrade -i flagger ${REPO_ROOT}/charts/flagger \
+--wait \
+--namespace istio-system \
+--set meshProvider=smi:istio
+
+kubectl -n istio-system set image deployment/flagger flagger=test/flagger:latest
+
+kubectl -n istio-system rollout status deployment/flagger
+
+echo '>>> Installing the SMI Istio adapter'
+kubectl apply -f ${REPO_ROOT}/artifacts/smi/istio-adapter.yaml
+
+kubectl -n istio-system rollout status deployment/smi-adapter-istio