移除nodeSelectorTerms后Pod处于Pending状态，如何让Kubernetes自动调度？

你遇到的Pod Pending问题核心在于Local Persistent Volume（本地PV）的特性限制：
Local PV是节点本地的存储资源，Kubernetes必须明确知道该PV属于哪个节点，才能将使用该PV的Pod调度到对应节点。移除nodeAffinity后，K8s无法识别PV所在节点，导致PVC无法绑定，Pod自然无法被调度。另外你之前配置的ReadWriteMany访问模式也是错误的——Local PV不支持RWX模式，仅支持ReadWriteOnce（单节点读写）或ReadOnlyMany（多节点只读）。

下面提供两种可行方案，推荐使用第一种实现自动调度：

方案1：使用Local PV动态供给（推荐） #

通过创建Local PV的StorageClass，让K8s自动完成PV创建、节点绑定和Pod调度，无需手动指定节点亲和性。

步骤1：创建Local存储类 #

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: local-registry-storage
provisioner: kubernetes.io/no-provisioner
volumeBindingMode: WaitForFirstConsumer

• volumeBindingMode: WaitForFirstConsumer是关键：它会等待Pod创建后，根据Pod的调度节点自动绑定对应节点的Local PV，实现调度和存储的联动。

步骤2：修改PVC配置 #

删除手动创建的PV，修改PVC指定使用上述存储类，并修正访问模式：

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: pv1-claim
  namespace: registry-space
spec:
  accessModes:
    - ReadWriteOnce  # 修正为Local PV支持的模式
  volumeMode: Filesystem
  resources:
    requests:
      storage: 5Gi
  storageClassName: local-registry-storage

步骤3：优化证书管理（可选但推荐） #

替换原来的hostPath证书卷为Secret，避免依赖节点本地目录的一致性：

1. 创建证书Secret：

kubectl create secret tls registry-tls --cert=/opt/certs/registry.crt --key=/opt/certs/registry.key -n registry-space

2. 修改Deployment的卷配置：

spec:
  volumes:
   - name: certs-vol
     secret:
       secretName: registry-tls
   - name: task-pv-storage
     persistentVolumeClaim:
       claimName: pv1-claim
  containers:
    - image: registry:2
      name: private-repository-k8s
      env:
      - name: REGISTRY_HTTP_TLS_CERTIFICATE
        value: "/etc/registry-tls/tls.crt"
      - name: REGISTRY_HTTP_TLS_KEY
        value: "/etc/registry-tls/tls.key"
      volumeMounts:
      - name: certs-vol
        mountPath: /etc/registry-tls
        readOnly: true
      - name: task-pv-storage
        mountPath: /opt/registry

方案2：手动创建多节点Local PV #

如果坚持手动管理PV，需要为每个节点创建单独的Local PV并配置节点亲和性，让K8s自动选择可用PV并调度Pod到对应节点。

为每个节点创建PV示例 #

# 节点kubernetes1的PV
apiVersion: v1
kind: PersistentVolume
metadata:
  name: pv-kubernetes1
  namespace: registry-space
spec:
  capacity:
    storage: 5Gi
  volumeMode: Filesystem
  persistentVolumeReclaimPolicy: Retain
  local:
    path: /opt/registry
  nodeAffinity:
    required:
      nodeSelectorTerms:
      - matchExpressions:
        - key: kubernetes.io/hostname
          operator: In
          values:
          - kubernetes1
  accessModes:
    - ReadWriteOnce

---
# 节点kubernetes2的PV
apiVersion: v1
kind: PersistentVolume
metadata:
  name: pv-kubernetes2
  namespace: registry-space
spec:
  capacity:
    storage: 5Gi
  volumeMode: Filesystem
  persistentVolumeReclaimPolicy: Retain
  local:
    path: /opt/registry
  nodeAffinity:
    required:
      nodeSelectorTerms:
      - matchExpressions:
        - key: kubernetes.io/hostname
          operator: In
          values:
          - kubernetes2
  accessModes:
    - ReadWriteOnce

---
# 节点kubernetes3的PV
apiVersion: v1
kind: PersistentVolume
metadata:
  name: pv-kubernetes3
  namespace: registry-space
spec:
  capacity:
    storage: 5Gi
  volumeMode: Filesystem
  persistentVolumeReclaimPolicy: Retain
  local:
    path: /opt/registry
  nodeAffinity:
    required:
      nodeSelectorTerms:
      - matchExpressions:
        - key: kubernetes.io/hostname
          operator: In
          values:
          - kubernetes3
  accessModes:
    - ReadWriteOnce

然后修改PVC的访问模式为ReadWriteOnce，K8s会自动选择一个可用PV绑定，并将Pod调度到对应节点。

内容的提问来源于stack exchange，提问作者Peter Penzov