Target Allocator Generate duplicated Target/Metrics when Service is Dual Stack

[XuhuiSun95]

Component(s)

No response

What happened?

Description

When the Service is configured with Prefer/Required Dual Stack and both IPv4 and IPv6 addresses are provided, the Target Allocator generates two targets because there are two distinct endpoint slices—even though they reference the same service.
This behavior results in duplicated metrics and unnecessary resource consumption by the collector.

Steps to Reproduce

Use the following values.yaml configuration for the opentelemetry-kube-stack chart:

collectors:
  daemon:
    crape_configs_file: ""
    targetAllocator:
      enabled: true
      image: ghcr.io/open-telemetry/opentelemetry-operator/target-allocator:main
      allocationStrategy: per-node
      prometheusCR:
        enabled: true
        podMonitorSelector: {}
        scrapeInterval: "30s"
        serviceMonitorSelector: {}
    presets:
      logsCollection:
        enabled: true
      kubeletMetrics:
        enabled: true 
      hostMetrics:
        enabled: false
      kubernetesAttributes:
        enabled: true 
      kubernetesEvents:
        enabled: false
      clusterMetrics:
        enabled: false
    config:
      receivers:
        prometheus: {}
      exporters:
        otlp/tempo:
          endpoint: lgtm-tempo-gateway.monitoring.svc.cluster.local:4317
          tls:
            insecure: true
        otlphttp/mimir:
          endpoint: http://lgtm-mimir-gateway.monitoring.svc.cluster.local/otlp
          headers:
            X-Scope-OrgID: homelab-k8s
          tls:
            insecure: true
        otlphttp/loki:
          endpoint: http://lgtm-loki-gateway.monitoring.svc/otlp
          headers:
            X-Scope-OrgID: homelab-k8s
          tls:
            insecure: true
      service:
        pipelines:
          traces:
            exporters:
              - otlp/tempo
          metrics:
            receivers:
              - otlp
              - prometheus
            exporters:
              - otlphttp/mimir
          logs:
            exporters:
              - otlphttp/loki

kubeStateMetrics:
  enabled: true 

kube-state-metrics:
  service:
    ipDualStack:
      enabled: true
  affinity:
    nodeAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
        nodeSelectorTerms:
          - matchExpressions:
            - key: karpenter.sh/nodepool
              operator: In
              values: ["system"]
  tolerations:
    - key: node-role.kubernetes.io/system
      operator: Exists
      effect: NoSchedule
  resources:
    requests:
      cpu: 10m
      memory: 100Mi
    limits:
      memory: 100Mi
  priorityClassName: platform-critical

Expected Result

The job
Job: serviceMonitor/opentelemetry-operator-system/opentelemetry-kube-stack-kube-state-metrics/0
should show only one target for the service, even when both IPv4 and IPv6 endpoint slices exist.

Actual Result

Kubernetes Version

v1.34.0

Operator version

0.136.0

Collector version

0.136.0

Environment information

Environment

OS: 6.12.52-talos
opentelemetry-kube-stack chart version: 0.11.1

Log output

opentelemetry-kube-stack-kube-state-metrics               ClusterIP   fd85:ee78:d8a6:8607::1eee   <none>        8080/TCP            5h52m



opentelemetry-kube-stack-kube-state-metrics-tl2q7               IPv6          8080        2600:xxxx:xxxx:b307:48::8a7b                                                                       5h52m
opentelemetry-kube-stack-kube-state-metrics-tzk4w               IPv4          8080        10.233.12.197                                                                                      5h52m



Name:                     opentelemetry-kube-stack-kube-state-metrics
Namespace:                opentelemetry-operator-system
Labels:                   app.kubernetes.io/component=metrics
                          app.kubernetes.io/instance=opentelemetry-kube-stack
                          app.kubernetes.io/managed-by=Helm
                          app.kubernetes.io/name=kube-state-metrics
                          app.kubernetes.io/part-of=kube-state-metrics
                          app.kubernetes.io/version=2.17.0
                          helm.sh/chart=kube-state-metrics-6.3.0
                          release=opentelemetry-kube-stack
Annotations:              argocd.argoproj.io/tracking-id: opentelemetry-kube-stack:/Service:opentelemetry-operator-system/opentelemetry-kube-stack-kube-state-metrics
                          prometheus.io/scrape: true
Selector:                 app.kubernetes.io/instance=opentelemetry-kube-stack,app.kubernetes.io/name=kube-state-metrics
Type:                     ClusterIP
IP Family Policy:         PreferDualStack
IP Families:              IPv6,IPv4
IP:                       fd85:ee78:d8a6:8607::1eee
IPs:                      fd85:ee78:d8a6:8607::1eee,10.233.118.252
Port:                     http  8080/TCP
TargetPort:               http/TCP
Endpoints:                [2600:xxxx:xxxx:b307:48::8a7b]:8080,10.233.12.197:8080
Session Affinity:         None
Internal Traffic Policy:  Cluster
Events:                   <none>

Additional context

No response

Tip

_{React with 👍 to help prioritize this issue. Please use comments to provide useful context, avoiding +1 or me too, to help us triage it. Learn more here.}

[Horiodino]

/assign