Kubernetes - pod 具有未绑定的即时 PersistentVolumeClaims

Question

我正在使用 mysql Kubernetes statefulset，我将 PV 映射到主机目录（CentOS 8 VM）但得到“pod has unbound直接 PersistentVolumeClaims”

apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: mysql-container
spec:
  serviceName: mysql
  replicas: 1 
  selector:
    matchLabels:
      app: mysql-container
  template:
    metadata:
      labels:
        app: mysql-container
    spec:
      containers:
      - name: mysql-container
        image: mysql:dev
        imagePullPolicy: "IfNotPresent"
        envFrom:
          - secretRef:
             name: prod-secrets
        ports:
        - containerPort: 3306
         # container (pod) path
        volumeMounts:
          - name: mysql-persistent-storage
            mountPath: /var/lib/mysql


      volumes:
        - name: mysql-persistent-storage
          persistentVolumeClaim:
           claimName: mysql-pvc
 volumeClaimTemplates:
    - metadata:
        name: data
      spec:

        storageClassName: localstorage
        accessModes: ["ReadWriteOnce"]
        resources:
          requests:
            storage: 3Gi
        selector:
           matchLabels:
            type: local

存储类是默认的，PV 中没有事件

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:

  name: localstorage

provisioner: kubernetes.io/no-provisioner
volumeBindingMode: Immediate
reclaimPolicy: Delete
allowVolumeExpansion: True


kind: PersistentVolume
apiVersion: v1
metadata:
  name: mysql-01
  labels:
    type: local
spec:
  storageClassName: localstorage
  capacity:
    storage: 10Gi
  accessModes:
    - ReadWriteOnce
  hostPath:
    path: "/mnt/mysql01"

---
kind: PersistentVolume
apiVersion: v1
metadata:
  name: mysql-02
  labels:
    type: local
spec:
  storageClassName: localstorage
  capacity:
    storage: 10Gi
  accessModes:
    - ReadWriteOnce
  hostPath:
    path: "/mnt/mysql02"

存储类是默认的一个

get sc
NAME                     PROVISIONER                    RECLAIMPOLICY   VOLUMEBINDINGMODE   ALLOWVOLUMEEXPANSION   AGE
localstorage (default)   kubernetes.io/no-provisioner   Delete          Immediate           true                   35m

PVC 也没有显示任何事件：

    Name:          data-mysql-0
    Namespace:     default
    StorageClass:  localstorage
    Status:        Pending
    Volume:        mysql-storage
    Labels:        app=mysql
    Annotations:   <none>
    Finalizers:    [kubernetes.io/pvc-protection]
    Capacity:      0
    Access Modes:
    VolumeMode:    Filesystem
    Mounted By:    mysql-0
    Events:        <none>


Name:            mysql-01
Labels:          type=local
Annotations:     kubectl.kubernetes.io/last-applied-configuration:
                   {"apiVersion":"v1","kind":"PersistentVolume","metadata":{"annotations":{},"labels":{"type":"local"},"name":"mysql-01"},"spec":{"accessMode...
Finalizers:      [kubernetes.io/pv-protection]
StorageClass:    localstorage
Status:          Available
Claim:
Reclaim Policy:  Retain
Access Modes:    RWO
VolumeMode:      Filesystem
Capacity:        10Gi
Node Affinity:   <none>
Message:
Source:
    Type:          HostPath (bare host directory volume)
    Path:          /mnt/mysql01
    HostPathType:
Events:            <none>


Name:            mysql-02
Labels:          type=local
Annotations:     kubectl.kubernetes.io/last-applied-configuration:
                   {"apiVersion":"v1","kind":"PersistentVolume","metadata":{"annotations":{},"labels":{"type":"local"},"name":"mysql-02"},"spec":{"accessMode...
Finalizers:      [kubernetes.io/pv-protection]
StorageClass:    localstorage
Status:          Available
Claim:
Reclaim Policy:  Retain
Access Modes:    RWO
VolumeMode:      Filesystem
Capacity:        10Gi
Node Affinity:   <none>
Message:
Source:
    Type:          HostPath (bare host directory volume)
    Path:          /mnt/mysql02
    HostPathType:
Events:            <none>

Pod 处于挂起状态：

>       Events:
>          Type     Reason            Age                From               Message
>          ----     ------            ----               ----               -------
>          Warning  FailedScheduling  27s (x2 over 27s)  default-scheduler  error while running >"VolumeBinding" filter plugin for pod "mysql-0": pod has unbound immediate PersistentVolumeClaims

有人可以指出这里还应该做什么吗，谢谢

Answer 1

PersistentVolumeClaims如果PersistentVolume不存在匹配项，则将无限期地保持未绑定状态。在PersistentVolume与匹配accessModes和capacity。在这种情况下capacity，PV 为 ，10Gi而 PVCcapacity为3Gi。

该capacity光伏需要一样的要求，即在3Gi修复unbound immediate PersistentVolumeClaims问题。

Answer 2

提到的错误可能有多种原因 - 以下是我遇到的几个选项。

示例 1

将persistentvolume-controller未能找到一种PV具有容量大小是相等或更高，然后将该在指定的值PVC。

所以如果我们拿这个例子：

  # PVC
  resources:
     requests:
       storage: 3Gi


  # PV
  capacity:
    storage: 10Gi 

所以：
如果 PV capacity >= PVC capacity那么PVC应该绑定到PV。

如果不是，那么我们将unbound immediate PersistentVolumeClaims在 pod 级别和no volume plugin matched name描述 PVC 时得到错误。

示例 2

PVC的数量高于PV。

例如，如果只创建了一个 PV（或删除了其他 PV）：

$ kubectl get pv
NAME             CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS   CLAIM                                      STORAGECLASS    REASON   AGE
mongo-local-pv   50Gi       RWO            Retain           Bound    default/mongo-persistent-storage-mongo-0   local-storage            106m

我们可以看到一些工作负载（Pods 或 Stateful 集）将卡在待处理状态：

$ kubectl get pods
NAME      READY   STATUS    RESTARTS   AGE
mongo-0   2/2     Running   0          3m38s
mongo-1   0/2     Pending   0          3m23s

$ kubectl get pvc
NAME                               STATUS    VOLUME           CAPACITY   ACCESS MODES   STORAGECLASS    AGE
mongo-persistent-storage-mongo-0   Bound     mongo-local-pv   50Gi       RWO            local-storage   80m
mongo-persistent-storage-mongo-1   Pending                                              local-storage   45m

我们将在挂起的资源上收到上述错误。

示例 3

如果调度程序未能将节点与 PV 匹配。

使用本地卷时，需要nodeAffinityPV的值，并且应该是集群中现有节点的值：

apiVersion: v1
kind: PersistentVolume
metadata:
  name: local-mongo-pv
  .
  .
  nodeAffinity:
    required:
      nodeSelectorTerms:
      - matchExpressions:
        - key: kubernetes.io/hostname
          operator: In
          values:
          - node-which-doesnt-exists  # <----- Will lead to the error

示例 4

PV集群上已经存在同名不同配置的旧s，新PVC的根据它们创建。

使用本地卷时，管理员必须每次手动清理并重新设置本地卷以供重用。

(*)创建此本地静态供应商是为了帮助处理 PV 生命周期。