appset: Error informer is not a kubernetes informer when controller generate applications

[dacofr]

Checklist:

• I've searched in the docs and FAQ for my answer: https://bit.ly/argocd-faq.
• I've included steps to reproduce the bug.
• I've pasted the output of argocd version.

Describe the bug

I think that a regression as been introduced since 2.11.3. Now, when I run locally the appset controller, each time an application is created/updated I got this message : informer is not a kubernetes informer coming from updateCache

func (r *ApplicationSetReconciler) updateCache(ctx context.Context, obj client.Object, logger *log.Entry) {
	informer, err := r.Cache.GetInformer(ctx, obj)
	if err != nil {
		logger.Errorf("failed to get informer: %v", err)
		return
	}
	// The controller runtime abstract away informers creation
	// so unfortunately could not find any other way to access informer store.
	k8sInformer, ok := informer.(k8scache.SharedInformer)
	if !ok {
		logger.Error("informer is not a kubernetes informer")
		return
	}
	if err := k8sInformer.GetStore().Update(obj); err != nil {
		logger.Errorf("failed to update cache: %v", err)
		return
	}
}

I haven't knowledge with controller-runtime, but when I run locally on master branch vs commit in 2.11.2, I see that something change in cache implementation : multiNamespaceCache vs informerCache. Maybe a clue #17491 .

To Reproduce

• Create an AppSet, example https://github.com/dacofr/hello-appset/blob/main/appset.yaml
• Watch logs

If you continue by updating appset spec, each time an application is created or updated, we have an error.

Expected behavior

No error log !

Version

Running locally based on this commit : b96be4c

Logs

INFO[0029] generated 3 applications                      applicationset=argocd/hello-appset generator="{&ListGenerator{Elements:[]JSON{{[123 34 110 97 109 101 115 112 97 99 101 34 58 34 100 101 118 34 125]},{[123 34 110 97 109 101 115 112 97 99 101 34 58 34 115 116 103 34 125]},{[123 34 110 97 109 101 115 112 97 99 101 34 58 34 112 114 100 34 125]},},Template:ApplicationSetTemplate{ApplicationSetTemplateMeta:ApplicationSetTemplateMeta{Name:,Namespace:,Labels:map[string]string{},Annotations:map[string]string{},Finalizers:[],},Spec:ApplicationSpec{Source:nil,Destination:ApplicationDestination{Server:,Namespace:,Name:,},Project:,SyncPolicy:nil,IgnoreDifferences:[]ResourceIgnoreDifferences{},Info:[]Info{},RevisionHistoryLimit:nil,Sources:[]ApplicationSource{},},},ElementsYaml:,} nil nil nil nil nil nil nil nil nil}"
ERRO[0029] informer is not a kubernetes informer         app=argocd/dev applicationset=argocd/hello-appset
INFO[0029] created Application                           app=argocd/dev applicationset=argocd/hello-appset
ERRO[0029] informer is not a kubernetes informer         app=argocd/stg applicationset=argocd/hello-appset
INFO[0029] created Application                           app=argocd/stg applicationset=argocd/hello-appset
ERRO[0029] informer is not a kubernetes informer         app=argocd/prd applicationset=argocd/hello-appset
INFO[0029] created Application                           app=argocd/prd applicationset=argocd/hello-appset
INFO[0029] end reconcile                                 applicationset=argocd/hello-appset requeueAfter=0s
INFO[0029] generated 3 applications                      applicationset=argocd/hello-appset generator="{&ListGenerator{Elements:[]JSON{{[123 34 110 97 109 

[uthark]

I'm observing this error after upgrading to 2.12.0 from 2.11.7.

[tbronchain]

I can confirm the issue happens on 2.12.0, but resolved after downgrading to 2.11.7, on multiple argo deployments/clusters/apps.

[Supporterino]

Same here 2.12.0 introduces the error and reverting to 2.11.7 removes the error message and the thus related high load of the container

[bhvishal9]

Observing the same issue as soon as I upgraded to 2.12.0.

[Alwinius]

Same for us, but we cannot downgrade because 2.12.0 fixes some major issues we had in Kubernetes 1.30

[crenshaw-dev]

Thanks for the confirmations, all! If you're affected, please thumbs-up the issue. If you have additional details about the issue, please comment.

[nobbs]

Still happening in v2.12.1.