troubles with previewReplicaCount in BlueGreen rollout strategy

[vranystepan]

Hi folks,

When I'm experimenting with previewReplicaCount, I'm getting to the situation when it's creating and killing replicas in the preview RS:

⟳ rollout-bluegreen                            Rollout     ॥ Paused       35s  
├──# revision:2                                                                
│  └──⧉ rollout-bluegreen-5f49884f5c           ReplicaSet  ◌ Progressing  8s   preview
│     ├──□ rollout-bluegreen-5f49884f5c-qpvxn  Pod         ✔ Running      8s   ready:1/1
│     ├──□ rollout-bluegreen-5f49884f5c-z2snd  Pod         ✔ Running      8s   ready:1/1
│     ├──□ rollout-bluegreen-5f49884f5c-4b6vt  Pod         ◌ Terminating  0s   ready:0/1
│     ├──□ rollout-bluegreen-5f49884f5c-69zzs  Pod         ◌ Terminating  0s   ready:0/1
│     ├──□ rollout-bluegreen-5f49884f5c-6k69q  Pod         ◌ Terminating  0s   ready:0/1
│     ├──□ rollout-bluegreen-5f49884f5c-cdhmz  Pod         ◌ Terminating  0s   ready:0/1
│     ├──□ rollout-bluegreen-5f49884f5c-dk2kk  Pod         ◌ Terminating  0s   ready:0/1
│     ├──□ rollout-bluegreen-5f49884f5c-rxt9k  Pod         ◌ Terminating  0s   ready:0/1
│     ├──□ rollout-bluegreen-5f49884f5c-wvgqr  Pod         ◌ Terminating  0s   ready:0/1
│     └──□ rollout-bluegreen-5f49884f5c-x256k  Pod         ◌ Terminating  0s   ready:0/1
└──# revision:1                                                                
   └──⧉ rollout-bluegreen-7d6b6cb796           ReplicaSet  ✔ Healthy      34s  active
      ├──□ rollout-bluegreen-7d6b6cb796-2wkjr  Pod         ✔ Running      34s  ready:1/1
      ├──□ rollout-bluegreen-7d6b6cb796-5svlv  Pod         ✔ Running      34s  ready:1/1
      ├──□ rollout-bluegreen-7d6b6cb796-7lmcx  Pod         ✔ Running      34s  ready:1/1
      ├──□ rollout-bluegreen-7d6b6cb796-hpqj5  Pod         ✔ Running      34s  ready:1/1
      ├──□ rollout-bluegreen-7d6b6cb796-llm7r  Pod         ✔ Running      34s  ready:1/1
      ├──□ rollout-bluegreen-7d6b6cb796-mnzk4  Pod         ✔ Running      34s  ready:1/1
      ├──□ rollout-bluegreen-7d6b6cb796-pss79  Pod         ✔ Running      34s  ready:1/1
      ├──□ rollout-bluegreen-7d6b6cb796-qjjjf  Pod         ✔ Running      34s  ready:1/1
      ├──□ rollout-bluegreen-7d6b6cb796-rbwf9  Pod         ✔ Running      34s  ready:1/1
      └──□ rollout-bluegreen-7d6b6cb796-w2s99  Pod         ✔ Running      34s  ready:1/1

This is the manifest I'm using:

apiVersion: argoproj.io/v1alpha1
kind: Rollout
metadata:
  name: rollout-bluegreen
spec:
  replicas: 10
  revisionHistoryLimit: 2
  selector:
    matchLabels:
      app: rollout-bluegreen
  template:
    metadata:
      labels:
        app: rollout-bluegreen
    spec:
      containers:
      - name: rollouts-demo
        image: argoproj/rollouts-demo:blue
        imagePullPolicy: Always
        ports:
        - containerPort: 8080
  strategy:
    blueGreen: 
      activeService: rollout-bluegreen-active
      previewService: rollout-bluegreen-preview
      autoPromotionEnabled: false
      previewReplicaCount: 2

---
kind: Service
apiVersion: v1
metadata:
  name: rollout-bluegreen-active
spec:
  selector:
    app: rollout-bluegreen
  ports:
  - protocol: TCP
    port: 80
    targetPort: 8080

---
kind: Service
apiVersion: v1
metadata:
  name: rollout-bluegreen-preview
spec:
  selector:
    app: rollout-bluegreen
  ports:
  - protocol: TCP
    port: 80
    targetPort: 8080

Rollouts version:

image: argoproj/argo-rollouts:v0.6.0

Please, do you have any idea what's happening there? Am I doing something wrong?
Many thanks in advance!

[dthomson25]

Hi @vranystepan, it looks like the controller is getting stuck in a loop of scaling the preview ReplicaSet up and then down instead of just honoring the previewReplicaCount value. I can reproduce the issue locally and I will start working on a fix. Thanks for pointing this out!

[vranystepan]

Great! Thank you!

[dthomson25]

Fixed by #308