Add external checks to canary analysis

LICENSE

@@ -186,7 +186,7 @@
       same "printed page" as the copyright notice for easier
       identification within third-party archives.
 
-   Copyright [yyyy] [name of copyright owner]
+   Copyright 2018 Weaveworks. All rights reserved.
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.

README.md

@@ -73,7 +73,7 @@ You can change the canary analysis _max weight_ and the _step weight_ percentage
 For a deployment named _podinfo_, a canary promotion can be defined using Flagger's custom resource:
 
 ```yaml
-apiVersion: flagger.app/v1alpha1
+apiVersion: flagger.app/v1alpha2
 kind: Canary
 metadata:
   name: podinfo
@@ -87,7 +87,7 @@ spec:
   # the maximum time in seconds for the canary deployment
   # to make progress before it is rollback (default 600s)
   progressDeadlineSeconds: 60
-  # hpa reference (optional)
+  # HPA reference (optional)
   autoscalerRef:
     apiVersion: autoscaling/v2beta1
     kind: HorizontalPodAutoscaler
@@ -100,16 +100,17 @@ spec:
     - public-gateway.istio-system.svc.cluster.local
     # Istio virtual service host names (optional)
     hosts:
-    - app.istio.weavedx.com
+    - app.iowa.weavedx.com
   canaryAnalysis:
     # max number of failed metric checks before rollback
-    threshold: 5
+    threshold: 10
     # max traffic percentage routed to canary
     # percentage (0-100)
     maxWeight: 50
     # canary increment step
     # percentage (0-100)
-    stepWeight: 10
+    stepWeight: 5
+    # Istio Prometheus checks
     metrics:
     - name: istio_requests_total
       # minimum req success rate (non 5xx responses)
@@ -121,6 +122,14 @@ spec:
       # milliseconds
       threshold: 500
       interval: 30s
+    # external checks (optional)
+    webhooks:
+      - name: integration-tests
+        url: http://podinfo.test:9898/echo
+        timeout: 1m
+        metadata:
+          test: "all"
+          token: "16688eb5e9f289f1991c"
 ```
 
 The canary analysis is using the following promql queries:
@@ -166,6 +175,22 @@ histogram_quantile(0.99,
 )
 ```
 
+The canary analysis can be extended with webhooks. 
+Flagger would call a URL (HTTP POST) and determine from the response status code (HTTP 2xx) if the canary is failing or not.
+
+Webhook payload:
+
+```json
+{
+    "name": "podinfo",
+    "namespace": "test", 
+    "metadata": {
+        "test":  "all",
+        "token":  "16688eb5e9f289f1991c"
+    }
+}
+```
+
 ### Automated canary analysis, promotions and rollbacks
 
 Create a test namespace with Istio sidecar injection enabled:

artifacts/canaries/canary.yaml

@@ -1,4 +1,4 @@
-apiVersion: flagger.app/v1alpha1
+apiVersion: flagger.app/v1alpha2
 kind: Canary
 metadata:
   name: podinfo
@@ -35,6 +35,7 @@ spec:
     # canary increment step
     # percentage (0-100)
     stepWeight: 5
+    # Istio Prometheus checks
     metrics:
     - name: istio_requests_total
       # minimum req success rate (non 5xx responses)
@@ -46,3 +47,11 @@ spec:
       # milliseconds
       threshold: 500
       interval: 30s
+    # external checks (optional)
+    webhooks:
+      - name: integration-tests
+        url: http://podinfo.test:9898/echo
+        timeout: 1m
+        metadata:
+          test: "all"
+          token: "16688eb5e9f289f1991c"

artifacts/flagger/crd.yaml

@@ -4,11 +4,14 @@ metadata:
   name: canaries.flagger.app
 spec:
   group: flagger.app
-  version: v1alpha1
+  version: v1alpha2
   versions:
-    - name: v1alpha1
+    - name: v1alpha2
       served: true
       storage: true
+    - name: v1alpha1
+      served: true
+      storage: false
   names:
     plural: canaries
     singular: canary
@@ -26,6 +29,8 @@ spec:
             progressDeadlineSeconds:
               type: number
             targetRef:
+              type: object
+              required: ['apiVersion', 'kind', 'name']
               properties:
                 apiVersion:
                   type: string
@@ -34,6 +39,8 @@ spec:
                 name:
                   type: string
             autoscalerRef:
+              type: object
+              required: ['apiVersion', 'kind', 'name']
               properties:
                 apiVersion:
                   type: string
@@ -42,6 +49,8 @@ spec:
                 name:
                   type: string
             service:
+              type: object
+              required: ['port']
               properties:
                 port:
                   type: number
@@ -58,6 +67,7 @@ spec:
                   properties:
                     items:
                       type: object
+                      required: ['name', 'interval', 'threshold']
                       properties:
                         name:
                           type: string
@@ -66,3 +76,18 @@ spec:
                           pattern: "^[0-9]+(m)"
                         threshold:
                           type: number
+                webhooks:
+                  type: array
+                  properties:
+                    items:
+                      type: object
+                      required: ['name', 'url', 'timeout']
+                      properties:
+                        name:
+                          type: string
+                        url:
+                          type: string
+                          format: url
+                        timeout:
+                          type: string
+                          pattern: "^[0-9]+(s)"

artifacts/flagger/deployment.yaml

@@ -22,7 +22,7 @@ spec:
       serviceAccountName: flagger
       containers:
       - name: flagger
-        image: quay.io/stefanprodan/flagger:0.1.2
+        image: quay.io/stefanprodan/flagger:0.2.0-alpha.1
         imagePullPolicy: Always
         ports:
         - name: http

charts/flagger/Chart.yaml

@@ -1,7 +1,7 @@
 apiVersion: v1
 name: flagger
 version: 0.1.2
-appVersion: 0.1.2
+appVersion: 0.2.0-alpha.1
 kubeVersion: ">=1.9.0-0"
 engine: gotpl
 description: Flagger is a Kubernetes operator that automates the promotion of canary deployments using Istio routing for traffic shifting and Prometheus metrics for canary analysis.

charts/flagger/templates/crd.yaml

@@ -3,15 +3,16 @@ apiVersion: apiextensions.k8s.io/v1beta1
 kind: CustomResourceDefinition
 metadata:
   name: canaries.flagger.app
-  annotations:
-    "helm.sh/resource-policy": keep
 spec:
   group: flagger.app
-  version: v1alpha1
+  version: v1alpha2
   versions:
-    - name: v1alpha1
+    - name: v1alpha2
       served: true
       storage: true
+    - name: v1alpha1
+      served: true
+      storage: false
   names:
     plural: canaries
     singular: canary
@@ -22,13 +23,15 @@ spec:
       properties:
         spec:
           required:
-          - targetRef
-          - service
-          - canaryAnalysis
+            - targetRef
+            - service
+            - canaryAnalysis
           properties:
             progressDeadlineSeconds:
               type: number
             targetRef:
+              type: object
+              required: ['apiVersion', 'kind', 'name']
               properties:
                 apiVersion:
                   type: string
@@ -37,6 +40,8 @@ spec:
                 name:
                   type: string
             autoscalerRef:
+              type: object
+              required: ['apiVersion', 'kind', 'name']
               properties:
                 apiVersion:
                   type: string
@@ -45,6 +50,8 @@ spec:
                 name:
                   type: string
             service:
+              type: object
+              required: ['port']
               properties:
                 port:
                   type: number
@@ -61,6 +68,7 @@ spec:
                   properties:
                     items:
                       type: object
+                      required: ['name', 'interval', 'threshold']
                       properties:
                         name:
                           type: string
@@ -69,4 +77,20 @@ spec:
                           pattern: "^[0-9]+(m)"
                         threshold:
                           type: number
+                webhooks:
+                  type: array
+                  properties:
+                    items:
+                      type: object
+                      required: ['name', 'url', 'timeout']
+                      properties:
+                        name:
+                          type: string
+                        url:
+                          type: string
+                          format: url
+                        timeout:
+                          type: string
+                          pattern: "^[0-9]+(s)"
+
 {{- end }}

charts/flagger/values.yaml

@@ -2,7 +2,7 @@
 
 image:
   repository: quay.io/stefanprodan/flagger
-  tag: 0.1.2
+  tag: 0.2.0-alpha.1
   pullPolicy: IfNotPresent
 
 controlLoopInterval: "10s"

cmd/flagger/main.go

@@ -77,7 +77,7 @@ func main() {
 	}
 
 	flaggerInformerFactory := informers.NewSharedInformerFactory(flaggerClient, time.Second*30)
-	canaryInformer := flaggerInformerFactory.Flagger().V1alpha1().Canaries()
+	canaryInformer := flaggerInformerFactory.Flagger().V1alpha2().Canaries()
 
 	logger.Infof("Starting flagger version %s revision %s", version.VERSION, version.REVISION)