stage is stuck forever if targeted 0 clusters

[riuvshyn]

stage is stuck forever if targeted 0 clusters

[maruina]

Fixed in #143 and a new test covering this at

applicationset-progressive-sync/controllers/progressivesync_controller_test.go

Line 247 in 1685ca7

t.Run("complete when a stage is targeting zero clusters", func(t *testing.T) {

[riuvshyn]

I've recently tested 0.6.0 and confirm that this issues is gone now 👍🏽

[riuvshyn]

hello @maruina I had to re-open this ticket as I believe the issue is still there. I was able to reproduce it with following configuration:
we have only 1 cluster registered in argocd labeled as environment: production

progressive sync object looks like this:

apiVersion: argoproj.skyscanner.net/v1alpha1
kind: ProgressiveSync
metadata:
  annotations:
    argocd.argoproj.io/sync-options: SkipDryRunOnMissingResource=true
  labels:
    argocd.argoproj.io/instance: argocd-apps
  name: aws-load-balancer-controller
  namespace: argocd
spec:
  appSetRef:
    name: aws-load-balancer-controller
  stages:
  - maxParallel: 1
    maxTargets: 1
    name: canary rollout staging
    targets:
      clusters:
        selector:
          matchLabels:
            environment: staging
  - maxParallel: 2
    maxTargets: 99999
    name: finish rollout staging
    targets:
      clusters:
        selector:
          matchLabels:
            environment: staging
  - maxParallel: 1
    maxTargets: 1
    name: canary rollout production
    targets:
      clusters:
        selector:
          matchLabels:
            environment: production
  - maxParallel: 2
    maxTargets: 99999
    name: finish rollout production
    targets:
      clusters:
        selector:
          matchLabels:
            environment: production
status:
  conditions:
  - lastTransitionTime: "2022-02-03T12:14:52Z"
    message: Reconciliation in progress
    reason: Progressing
    status: Unknown
    type: Ready
  lastSyncedStage: canary rollout staging
  lastSyncedStageStatus: StageProgressing
  observedGeneration: 1

progressive sync logs:

2022-02-03T12:14:52.886Z	INFO	controller-runtime.manager.controller.progressivesync	progressive sync started	{"reconciler group": "argoproj.skyscanner.net", "reconciler kind": "ProgressiveSync", "name": "aws-load-balancer-controller", "namespace": "argocd"}
2022-02-03T12:14:52.886Z	INFO	controller-runtime.manager.controller.progressivesync	stage reconciled	{"reconciler group": "argoproj.skyscanner.net", "reconciler kind": "ProgressiveSync", "name": "aws-load-balancer-controller", "namespace": "argocd", "stage": "canary rollout staging", "status": "StageProgressing"}
2022-02-03T12:14:52.967Z	INFO	controller-runtime.manager.controller.progressivesync	progressive sync started	{"reconciler group": "argoproj.skyscanner.net", "reconciler kind": "ProgressiveSync", "name": "aws-load-balancer-controller", "namespace": "argocd"}
2022-02-03T12:14:52.968Z	INFO	controller-runtime.manager.controller.progressivesync	stage reconciled	{"reconciler group": "argoproj.skyscanner.net", "reconciler kind": "ProgressiveSync", "name": "aws-load-balancer-controller", "namespace": "argocd", "stage": "canary rollout staging", "status": "StageProgressing"}

so it is stuck on first stage name: canary rollout staging with status:

status:
  conditions:
  - lastTransitionTime: "2022-02-03T12:14:52Z"
    message: Reconciliation in progress
    reason: Progressing
    status: Unknown
    type: Ready
  lastSyncedStage: canary rollout staging
  lastSyncedStageStatus: StageProgressing
  observedGeneration: 1

there are no clusters yet known by argo tagged with environment: staging and our expectation is that controller should skip stage if no clusters were targeted for that stage.

[maruina]

Hey, thank you very much for all the details. Hopefully, I will be able to reproduce the issue in a test and then a fix.

I should start working on this in the next couple of days.

[riuvshyn]

@maruina I've proposed PR #159
not sure if that can be considered as legit fix or workaround but that worked for me.

[maruina]

Fixed in #159