Kubernetes 无法删除pod实例的排查过程

桃扇骨 2022-01-29 10:09 471阅读 0赞

原本由两台Kubernetes组成的小集群，但是今天只开启了一台机器，也就是只有一个节点，造成了无法删除pod实例的原因。

先查看一下现在的容器的运行状态：

[root@k8s ~]# kubectl get pods
    
    NAME                      READY     STATUS             RESTARTS   AGE 
    nginx-controller-lv8md    1/1       Unknown            0          16h 
    nginx-controller-sb3fx    1/1       Unknown            2          16h 
    nginx2-1216651254-4b2dw   0/1       ImagePullBackOff   0          8m 
    nginx2-1216651254-dbtms   0/1       ImagePullBackOff   0          8m 
    nginx2-1216651254-fhb4r   0/1       ImagePullBackOff   0          8m

查看有哪些replicationcontroller \[简写rc\]

[root@k8s ~]# kubectl get rc
    No resources found.

查看有哪些services

[root@k8s ~]# kubectl get svc
    NAME         CLUSTER-IP   EXTERNAL-IP   PORT(S)   AGE
    kubernetes   10.254.0.1   <none>        443/TCP   2d

看到上面没有rc，也没有services，那尝试这样删除所有的pods：

[root@k8s ~]# kubectl delete pods --all
    pod "nginx-controller-lv8md" deleted
    pod "nginx-controller-sb3fx" deleted
    pod "nginx2-1216651254-4b2dw" deleted
    pod "nginx2-1216651254-dbtms" deleted
    pod "nginx2-1216651254-fhb4r" deleted

但是还是无法删除，查看已经部署的容器；

[root@k8s ~]# kubectl get deployment
    NAME      DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
    nginx2    3         3         3            0           16h

[root@k8s ~]# kubectl delete deployment nginx2
    deployment "nginx2" deleted

为什么这三个Pod实例没有rc或者services呢，因为创建它的时候是使用run来实现的；

但是剩下的两个实例怎么删除呢？

[root@k8s ~]# kubectl get pods 
    
    NAME                     READY     STATUS    RESTARTS   AGE 
    
    nginx-controller-lv8md   1/1       Unknown   0          20h 
    nginx-controller-sb3fx   1/1       Unknown   2          20h

查看一下实例的信息：

[root@k8s ~]# kubectl describe pod nginx-controller-lv8md  
    
    Name:               nginx-controller-lv8md 
    Namespace:          default 
    Node:               k8s-node/10.0.10.11           #这里是重点，因为这两个容器是分配到了k8s-node节点上，而这个节点现在宕机了。 
    Start Time:         Tue, 13 Jun 2017 02:01:45 +0800 
    Labels:             app=nginx 
    Status:             Terminating (expires Mon, 12 Jun 2017 21:46:21 +0800) 
    Termination Grace Period:   30s 
    Reason:             NodeLost 
    Message:            Node k8s-node which was running pod nginx-controller-lv8md is unresponsive 
    IP:             172.21.42.3 
    Controllers:            ReplicationController/nginx-controller 
    Containers: 
      nginx: 
        Container ID:       docker://03fa59f9efc06e43ed8c9acc7d4c7533983d5733223dbb2efa5f65928d965b5b 
        Image:          reg.docker.lc/share/nginx:latest 
        Image ID:           docker-pullable://reg.docker.lc/share/nginx@sha256:e5c82328a509aeb7c18c1d7fb36633dc638fcf433f651bdcda59c1cc04d3ee55 
        Port:           80/TCP 
        State:          Running 
          Started:          Tue, 13 Jun 2017 02:01:47 +0800 
        Ready:          True 
        Restart Count:      0 
        Volume Mounts:      <none> 
        Environment Variables:  <none> 
    Conditions: 
      Type      Status 
      Initialized   True  
      Ready     False  
      PodScheduled  True  
    No volumes. 
    QoS Class:  BestEffort 
    Tolerations:    <none> 
    No events.

因为这两个容器的rc，services都已经删除了，但是还保持这个Unknown状态是由于目标主机无法响应并返回信息导致；既然目标主机都已经宕机了，那就直接移除节点；

[root@k8s ~]# kubectl delete node k8s-node  
    node "k8s-node" deleted

[root@k8s ~]# kubectl get node  
    
    NAME      STATUS    AGE 
    k8s       Ready     2d

因为节点都不存在了，那就没有了容器的状态信息；

[root@k8s ~]# kubectl get pods 
    
    No resources found.

上面这样虽然是删除了那些已经也错的实例，但却删除了节点。 那么如果节点现在恢复了，是否会自动加入到集群呢？

主节点上的日志很快就发现了k8s-node节点了；

> Jun 13 15:06:13 k8s kube-controller-manager\[34050\]: E0613 15:06:13.917313 34050 actual\_state\_of\_world.go:475\] Failed to set statusUpdateNeeded to needed true because nodeName=”k8s-node” does not exist
> 
> Jun 13 15:06:14 k8s kube-controller-manager\[34050\]: I0613 15:06:14.618864 34050 event.go:217\] Event(api.ObjectReference\{Kind:”Node”, Namespace:””, Name:”k8s-node”, UID:”c9864434-5006-11e7-ab16-000c29e9277a”, APIVersion:””, ResourceVersion:””, FieldPath:””\}): type: ‘Normal’ reason: ‘RegisteredNode’ Node k8s-node event: Registered Node k8s-node in NodeController

节点自动又回来了，因为在配置文件中已经配置好了，只要节点一上线就会加入到集群中。

[root@k8s ~]# kubectl get nodes 
    
    NAME       STATUS    AGE 
    k8s        Ready     2d 
    k8s-node   Ready     1m

创建一个Nginx实例：

[root@k8s ~]# kubectl create -f Nginx.yaml  
    
    replicationcontroller "nginx-controller" created 
    service "nginx-service" created

Kubernetes已经把此实例分配到刚启动的k8s-node节点上了；

[root@k8s ~]# kubectl get pods 
    NAME                     READY     STATUS    RESTARTS   AGE 
    nginx-controller-zsx2q   1/1       Running   0          6s

[root@k8s ~]# kubectl describe pod nginx-controller-zsx2q |grep "Node" 
    Node:       k8s-node/10.0.10.11

动态扩展，添加一个容器数量；

[root@k8s ~]# kubectl scale replicationcontroller --replicas=2 nginx-controller  
    replicationcontroller "nginx-controller" scaled

查看集群现在的状态信息；

[root@k8s ~]# kubectl describe svc nginx-service  
    Name:           nginx-service 
    Namespace:      default 
    Labels:         <none> 
    Selector:       app=nginx 
    Type:           ClusterIP 
    IP:         10.254.132.82 
    External IPs:       10.0.10.10 
    Port:           <unset> 8000/TCP 
    Endpoints:      172.21.42.2:80,172.21.93.2:80 
    Session Affinity:   None 
    No events.

此时再删除k8s-node节点又会如何呢？

[root@k8s ~]# kubectl delete node k8s-node  
    node "k8s-node" deleted

已经删除了一个节点了：

[root@k8s ~]# kubectl get nodes 
    NAME      STATUS    AGE 
    k8s       Ready     2d

Pod的数量没有减少还是两个：

[root@k8s ~]# kubectl get pods 
    
    NAME                     READY     STATUS    RESTARTS   AGE 
    
    nginx-controller-43qpx   1/1       Running   0          14m 
    nginx-controller-zsx2q   1/1       Running   0          19m

查看集群中的分配信息时发现，两个容器的Ip段都是172.21.93的说明来自同一节点；

[root@k8s ~]# kubectl describe svc nginx-service  
    
    Name:           nginx-service 
    Namespace:      default 
    Labels:         <none> 
    Selector:       app=nginx 
    Type:           ClusterIP 
    IP:         10.254.132.82 
    External IPs:       10.0.10.10 
    Port:           <unset> 8000/TCP 
    Endpoints:      172.21.93.2:80,172.21.93.3:80 
    Session Affinity:   None 
    No events.

重新加入k8s-node 节点，需要在k8s-node上重启一下服务：

systemctl restart kubelet