kubernetes - 为什么两个具有相同选择器的Kubernetes ReplicaSet不会相互冲突？

Question

副本集1

apiVersion: apps/v1
kind: ReplicaSet
metadata:
  labels:
    app: nginx
  name: rs-1
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
        version: 1.7.1
    spec:
      containers:
      - image: nginx:1.7.1
        name: nginx-1
      restartPolicy: Always

副本集2

apiVersion: apps/v1
kind: ReplicaSet
metadata:
  labels:
    app: nginx
  name: rs-2
spec:
  replicas: 3
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
        version: 1.7.9
    spec:
      containers:
      - image: nginx:1.7.9
        name: nginx-1
      restartPolicy: Always

当我创建这两个ReplicaSet时，一个会忽略另一个创建的Pod。

C02T30K2GTFM:ask erkanerol$ kubectl get pods --show-labels
NAME         READY   STATUS    RESTARTS   AGE     LABELS
rs-1-996cz   1/1     Running   0          5m13s   app=nginx,version=1.7.1
rs-1-ktv9z   1/1     Running   0          5m13s   app=nginx,version=1.7.1
rs-1-w7sbg   1/1     Running   0          5m13s   app=nginx,version=1.7.1
rs-2-2z8rb   1/1     Running   0          4m26s   app=nginx,version=1.7.9
rs-2-5c56s   1/1     Running   0          4m26s   app=nginx,version=1.7.9
rs-2-hls9p   1/1     Running   0          4m26s   app=nginx,version=1.7.9

据我从文档中了解到，如果有足够的Pod与副本集的选择器匹配，则不应创建新Pod。为什么会这样呢？是否使用ownerReferences？

Answer 1

看来他们正在使用ownerReferences。如果是这样，则不适合记录的行为。

PR:https://github.com/kubernetes/kubernetes/pull/27600

代码:https://github.com/kubernetes/kubernetes/blob/0048d2da400b8c48ae83acc6a223a65f3551674a/pkg/controller/controller_ref_manager.go#L69-L72

问题:https://github.com/kubernetes/website/issues/12205