Add AnalysisTemplate & update logic for canary rollout strategy

.github/workflows/test.yaml

@@ -32,5 +32,6 @@ jobs:
           kubectl apply -f https://raw.githubusercontent.com/opsgy/loki-rule-operator/main/chart/crds/logging.opsgy.com_lokirules.yaml
           kubectl apply -f https://raw.githubusercontent.com/istio/istio/master/manifests/charts/base/crds/crd-all.gen.yaml
           kubectl apply -f https://raw.githubusercontent.com/argoproj/argo-rollouts/master/manifests/crds/rollout-crd.yaml
+          kubectl apply -f https://raw.githubusercontent.com/argoproj/argo-rollouts/master/manifests/crds/analysis-template-crd.yaml
       - name: Install charts
         run: ct install --all

charts/generic-service/README.md

@@ -68,7 +68,11 @@ app:
 | `rollout.autoPromotion`                         | `true`                      | Automatically promote rollouts (if `rollout.strategy` is `BlueGreen` and `rollout.flagger` if `false`)   |
 | `rollout.flagger`                               | `false`                     | Use Flagger to control rollouts (`rollout.controller` must be `Deployment` or `StatefulSet`)             |
 | `rollout.analysis`                              | req. for Canary or Flagger  | Flagger or Argo Rollouts analysis for automatic `Canary` or `BlueGreen` promotion                        |
-| `rollout.revisionHistoryLimit`                  | `null`                      | Number of old ReplicaSets to retain (`rollout.controller` must be `Deployment` or `ArgoRollout`)         |
+| `rollout.revisionHistoryLimit`                  | `600`                       | Number of old ReplicaSets to retain (`rollout.controller` must be `Deployment` or `ArgoRollout`)         |
+| `rollout.progressDeadlineSeconds`               | `null`                      | The maximum time in seconds for a deployment to make progress before it is considered to be failed       |
+| `rollout.minReadySeconds`                       | `0`                         | Minimum number of seconds for which a new pod should be ready for it to be considered available          |
+| `rollout.progressDeadlineAbort`                 | `false`                     | Whether to abort the update when `rollout.progressDeadlineSeconds` is exceeded (`rollout.controller` must be `ArgoRollout` and `rollout.strategy` `Canary`) |
+| `rollout.notifications`                         | `{}`                        | Option to subscribe to argo rollout events and send slack notifications                                  |
 | `replicas`                                      | `1`                         | The number of instances of the service to run (set at least `2` for Pod Disruption Budget)               |
 | `autoscaling.enabled`                           | `false`                     | Enables automatic starting of additional instances                                                       |
 | `autoscaling.maxReplicas`                       | `3`                         | The maximum number of instances to run (must be larger than `replicas`)                                  |

charts/generic-service/ci/argo-bluegreen-values.yaml

@@ -0,0 +1,14 @@
+# Argo Rollout controller BlueGreen test
+
+image:
+  repository: jwilder/whoami
+  tag: latest
+
+ingress:
+  enabled: true
+  port: 8000
+
+rollout:
+  controller: ArgoRollout
+  strategy: BlueGreen
+  autoPromotion: true

charts/generic-service/ci/argo-canary-values.yaml

@@ -0,0 +1,42 @@
+# Argo Rollout controller Canary test
+
+image:
+  repository: jwilder/whoami
+  tag: latest
+
+ingress:
+  enabled: true
+  port: 8000
+
+replicas: 2
+
+rollout:
+  controller: ArgoRollout
+  strategy: Canary
+  steps:
+    - setWeight: 20
+    - analysis:
+        templates:
+        - templateName: success-rate
+        args:
+        - name: service-name
+          value: canary-demo
+    - pause:
+        duration: 1h
+    - setWeight: 40
+  analysisTemplates:
+    - name: success-rate
+      args:
+        - name: service-name
+      metrics:
+        - name: success-rate
+          interval: 5m
+          count: 10
+          successCondition: "result[0] >= 0.95"
+          query: |
+            sum(irate(
+              istio_requests_total{reporter="source",destination_service=~"{{args.service-name}}",response_code!~"5.*"}[5m]
+            )) /
+            sum(irate(
+              istio_requests_total{reporter="source",destination_service=~"{{args.service-name}}"}[5m]
+            ))

charts/generic-service/templates/analysis-template.yaml

@@ -0,0 +1,55 @@
+{{- if eq .Values.rollout.controller "ArgoRollout" }}
+{{- range .Values.rollout.analysisTemplates }}
+---
+apiVersion: argoproj.io/v1alpha1
+kind: AnalysisTemplate
+metadata:
+  name: {{ include "generic-service.fullname" $ }}-{{ .name }}
+spec:
+  {{- if .args }}
+  args:
+    {{- toYaml .args | nindent 4 }}
+  {{- end }}
+  {{- if .dryRun }}
+  dryRun:
+    {{- toYaml .dryRun | nindent 4 }}
+  {{- end }}
+  {{- if .measurementRetention }}
+  measurementRetention:
+    {{- toYaml .measurementRetention | nindent 4 }}
+  {{- end }}
+  metrics:
+    {{- range .metrics }}
+    - name: {{ .name }}
+      {{- if .consecutiveErrorLimit }}
+      consecutiveErrorLimit: {{ .consecutiveErrorLimit }}
+      {{- end }}
+      {{- if .count }}
+      count: {{ .count }}
+      {{- end }}
+      {{- if .failureCondition }}
+      failureCondition: {{ .failureCondition }}
+      {{- end }}
+      {{- if .successCondition }}
+      successCondition: {{ .successCondition }}
+      {{- end }}
+      {{- if .failureLimit }}
+      failureLimit: {{ .failureLimit }}
+      {{- end }}
+      {{- if .inconclusiveLimit }}
+      inconclusiveLimit: {{ .inconclusiveLimit }}
+      {{- end }}
+      {{- if .initialDelay }}
+      initialDelay: {{ .initialDelay }}
+      {{- end }}
+      {{- if .interval }}
+      interval: {{ .interval }}
+      {{- end }}
+      provider:
+        prometheus:
+          address: {{ .prometheusAddress | default "http://prometheus-prometheus.monitoring:9090" }}
+          query: |
+            {{ .query | nindent 12 }}
+    {{- end }}
+{{- end }}
+{{- end }}

charts/generic-service/templates/controller.yaml

@@ -32,6 +32,13 @@ spec:
   selector:
     matchLabels: {{- include "generic-service.selector-labels" . | nindent 6 }}
 
+  progressDeadlineSeconds: {{ .Values.rollout.progressDeadlineSeconds| int }}
+  minReadySeconds: {{ .Values.rollout.minReadySeconds | int }}
+
+  {{- if and .Values.rollout.progressDeadlineAbort (eq .Values.rollout.controller "ArgoRollout") }}
+  progressDeadlineAbort: {{ .Values.rollout.progressDeadlineAbort }}
+  {{- end }}
+
   strategy:
   {{- if eq .Values.rollout.strategy "RollingUpdate" }}
     {{- if eq .Values.rollout.controller "ArgoRollout" }}
@@ -51,59 +58,56 @@ spec:
     {{- end }}
     type: OnDelete
   {{- else if eq .Values.rollout.strategy "Canary" }}
-    {{- if not .Values.ingress.enabled }}
-      {{ fail "ingress.enabled must be true if rollout.strategy is Canary" }}
-    {{- end }}
-    {{- if not .Values.ingress.domains }}
-      {{ fail "ingress.domains must not be empty if rollout.strategy is Canary" }}
-    {{- end }}
     {{- if eq .Values.rollout.controller "ArgoRollout" }}
     canary:
-      stableService: {{ include "generic-service.fullname" . }}
-      stableMetadata:
-        labels:
-          role: stable
-      canaryService: {{ include "generic-service.fullname" . }}-preview
-      canaryMetadata:
-        labels:
-          role: preview
-      analysis: {{- .Values.rollout.analysis | required "rollout.analysis is required if rollout.strategy is Canary" | toYaml |  nindent 8 }}
-      trafficRouting:
-        {{- if .Values.ingress.istio.enabled }}
-        istio:
-          virtualService:
-            name: {{ include "generic-service.fullname" . }}
-        {{- else if contains "nginx" .Values.ingress.class }}
-        nginx:
-          stableIngress: {{ include "generic-service.fullname" . }}
-        {{- else }}
-          {{ fail "Ingress must use Istio or nginx if rollout.strategy is Canary" }}
+      {{- with .Values.rollout.analysis -}}
+        analysis: {{- . | toYaml | nindent 8 }}
+      {{- end }}
+      {{- if .Values.rollout.useTrafficRouting -}}
+        {{- if not eq .Values.rollout.strategy "Canary" }}
+          {{ fail "Traffic routing is only available if rollout.strategy is Canary" }}
+        {{- end }}
+        stableService: {{ include "generic-service.fullname" . }}
+        stableMetadata:
+          labels:
+            role: stable
+        canaryService: {{ include "generic-service.fullname" . }}-preview
+        canaryMetadata:
+          labels:
+            role: preview
+        trafficRouting:
+          {{- if .Values.ingress.istio.enabled }}
+            istio:
+              virtualService:
+                name: {{ include "generic-service.fullname" . }}
+          {{- else if eq .Values.ingress.class "nginx" }}
+            nginx:
+              stableIngress: {{ include "generic-service.fullname" . }}
+          {{- else }}
+            {{ fail "Ingress must use Istio or nginx if rollout.strategy is Canary" }}
+          {{- end }}
+        {{- end }}
+        {{- with .Values.rollout.steps }}
+        steps: {{- . | toYaml | nindent 8 }}
         {{- end }}
-    {{- else if not .Values.rollout.flagger }}
-      {{ fail "rollout.flagger must be true or rollout.controller must be ArgoRollout if rollout.strategy is Canary" }}
     {{- end }}
   {{- else if eq .Values.rollout.strategy "BlueGreen" }}
-    {{- if not .Values.ingress.enabled }}
-      {{ fail "ingress.enabled must be true if rollout.strategy is BlueGreen" }}
-    {{- end }}
     {{- if eq .Values.rollout.controller "ArgoRollout" }}
-    blueGreen:
-      activeService: {{ include "generic-service.fullname" . }}
-      activeMetadata:
-        labels:
-          role: stable
-      previewService: {{ include "generic-service.fullname" . }}-preview
-      previewMetadata:
-        labels:
-          role: preview
-      autoPromotionEnabled: {{ .Values.rollout.autoPromotion }}
-      {{- if .Values.rollout.analysis }}
-      prePromotionAnalysis: {{- .Values.rollout.analysis | toYaml | nindent 8 }}
-      {{- end }}
-    {{- else if not .Values.rollout.flagger }}
-      {{ fail "rollout.flagger must be true or rollout.controller must be ArgoRollout if rollout.strategy is BlueGreen" }}
+      blueGreen:
+        activeService: {{ include "generic-service.fullname" . }}
+        activeMetadata:
+          labels:
+            role: stable
+        previewService: {{ include "generic-service.fullname" . }}-preview
+        previewMetadata:
+          labels:
+            role: preview
+        autoPromotionEnabled: {{ .Values.rollout.autoPromotion }}
+        {{- if .Values.rollout.analysis }}
+          prePromotionAnalysis: {{- .Values.rollout.analysis | toYaml | nindent 8 }}
+        {{- end }}
     {{- end }}
-  {{ else }}
+  {{- else }}
     {{ fail "Unknown rollout.strategy" }}
   {{- end }}
 
@@ -463,4 +467,3 @@ spec:
       {{- end }}
       {{- end }}
       {{- end }}
-

charts/generic-service/templates/service.yaml

@@ -93,7 +93,7 @@ spec:
     {{- end }}
 {{- end }}
 
-{{- if eq .Values.rollout.controller "ArgoRollout" }}
+{{- if and (eq .Values.rollout.controller "ArgoRollout") (or (eq .Values.rollout.strategy "BlueGreen") (eq .Values.rollout.useTrafficRouting true)) }}
 ---
 apiVersion: v1
 kind: Service

charts/generic-service/values.schema.json

@@ -301,11 +301,38 @@
           "description": "Use Flagger to control rollouts (rollout.controller must be Deployment or StatefulSet)"
         },
         "analysis": {
-          "description": "Flagger or Argo Rollouts analysis for automatic Canary or BlueGreen promotion"
+          "type": "object",
+          "description": "Flagger or Argo Rollouts analysis for automatic Canary promotion"
+        },
+        "steps": {
+          "type": "array",
+          "description": "Argo Rollouts list of steps to perform during a canary rollout"
+        },
+        "analysisTemplates": {
+          "type": "array",
+          "description": "List of analysis templates to create (only applicable if rollout.controller is ArgoRollout)"
+        },
+        "useTrafficRouting": {
+          "type": "boolean",
+          "default": false,
+          "description": "Whether to use Argo Rollouts traffig routing feature"
         },
         "revisionHistoryLimit": {
           "type": ["integer", "null"],
           "description": "Number of old ReplicaSets to retain (rollout.controller must be Deployment or ArgoRollout)"
+        },
+        "progressDeadlineSeconds": {
+          "type": ["integer"],
+          "description": "The maximum time in seconds for a deployment to make progress before it is considered to be failed"
+        },
+        "minReadySeconds": {
+          "type": ["integer"],
+          "description": "Minimum number of seconds for which a new pod should be ready for it to be considered available"
+        },
+        "progressDeadlineAbort": {
+          "type": ["boolean"],
+          "default": false,
+          "description": "Whether to abort the update when progressDeadlineSeconds is exceeded (rollout.controller must be ArgoRollout and rollout.strategy Canary)"
         }
       },
       "additionalProperties": false

charts/generic-service/values.yaml

@@ -63,10 +63,16 @@ resources:
 rollout:
   controller: Deployment
   strategy: RollingUpdate
+  useTrafficRouting: false
   autoPromotion: true
   flagger: false
-  analysis: null
+  analysis: {}
+  steps: []
+  analysisTemplates: []
   revisionHistoryLimit: null
+  progressDeadlineSeconds: 600
+  minReadySeconds: 0
+  progressDeadlineAbort: false
 
 replicas: 1