# Pod 副本管理～ 實作篇 ### 觀察現況 我們透過 ReplicaSet 物件提交到 K8s API 的方式建立副本管理。\ 先行察看兩個物件，可以得知哪一個 pod 尚未套用 ReplicaSet 物件～\ `kubectl get pods`\ `kubectl get rs` ```bash [user@minikube ~]$ kubectl get pods NAME READY STATUS RESTARTS AGE hello-kitty 1/1 Running 0 6d hello-minikube-7c77b68cff-ncqmn 1/1 Running 1 11d hello-moto-6456dff9f-67jh5 1/1 Running 0 5d [user@minikube ~]$ kubectl get rs NAME DESIRED CURRENT READY AGE hello-minikube-7c77b68cff 1 1 1 11d hello-moto-6456dff9f 1 1 1 5d ``` 另一個方式查看該 pod 是否已有 ReplicaSet 管理\ `kubectl get pods -o yaml`\ 如果顯示的內容中，有出現`kind: ReplicaSet`物件，即代表受管理中 ```bash ... ownerReferences: - apiVersion: extensions/v1beta1 blockOwnerDeletion: true controller: true kind: ReplicaSet name: hello-moto-6456dff9f ``` ### 新增 ReplicaSet 實例 經比對，pod **hello-kitty** 尚未套用 ReplicaSet，可以最為下列範例～ \ 新增 **hello-kitty** 的 ReplicaSet 物件～`rs-hello-kitty.yaml` ```yaml apiVersion: extensions/v1beta1 kind: ReplicaSet metadata: name: hello-kitty spec: replicas: 2 template: metadata: labels: run: hello-kitty spec: containers: - name: hello-kitty image: "k8s.gcr.io/echoserver:1.10" ``` 套用 ReplicaSet 物件之前 Pod 狀態： ```bash [user@minikube ~]$ kubectl get pods hello-kitty -o wide NAME READY STATUS RESTARTS AGE IP NODE hello-kitty 1/1 Running 0 5d 172.17.0.6 minikube ``` 套用 ReplicaSet 物件： ```bash [user@minikube ~]$ kubectl apply -f kk8s/rs-hello-kitty.yaml replicaset.extensions/hello-kitty created ``` 套用 ReplicaSet 物件之後 Pod 狀態： ```bash [user@minikube ~]$ kubectl get pods NAME READY STATUS RESTARTS AGE hello-kitty 1/1 Running 0 6d hello-kitty-lk7lg 1/1 Running 0 1m hello-kitty-x94s2 1/1 Running 0 1m hello-minikube-7c77b68cff-ncqmn 1/1 Running 1 11d hello-moto-6456dff9f-67jh5 1/1 Running 0 5d ``` 檢查 ReplicaSet，可以看見裏頭的 Labels、replicas 資訊： ```bash [user@minikube ~]$ kubectl describe rs hello-kitty Name: hello-kitty Namespace: default Selector: run=hello-kitty Labels: run=hello-kitty Annotations: kubectl.kubernetes.io/last-applied-configuration: {"apiVersion":"extensions/v1beta1","kind":"ReplicaSet","metadata":{"annotations":{},"name":"hello-kitty","namespace":"default"},"spec":{"r... Replicas: 2 current / 2 desired Pods Status: 2 Running / 0 Waiting / 0 Succeeded / 0 Failed Pod Template: Labels: run=hello-kitty Containers: hello-kitty: Image: k8s.gcr.io/echoserver:1.10 Port: Host Port: Environment: Mounts: Volumes: Events: Type Reason Age From Message ---- ------ ---- ---- ------- Normal SuccessfulCreate 84s replicaset-controller Created pod: hello-kitty-lk7lg Normal SuccessfulCreate 84s replicaset-controller Created pod: hello-kitty-x94s2 ``` ### Scale Pod 方式一 進行擴展 Pod`[2 --> 4]`，透過指令`kubectl scale` ```bash [user@minikube ~]$ kubectl scale --replicas=4 rs/hello-kitty replicaset.extensions/hello-kitty scaled [user@minikube ~]$ kubectl get rs hello-kitty NAME DESIRED CURRENT READY AGE hello-kitty 4 4 4 9h [user@minikube ~]$ kubectl get pods NAME READY STATUS RESTARTS AGE hello-kitty 1/1 Running 0 6d hello-kitty-d7tfq 1/1 Running 0 22s hello-kitty-lk7lg 1/1 Running 0 9h hello-kitty-x94s2 1/1 Running 0 9h hello-kitty-zgrb8 1/1 Running 0 22s hello-minikube-7c77b68cff-ncqmn 1/1 Running 1 11d hello-moto-6456dff9f-67jh5 1/1 Running 0 6d ``` {% hint style="info" %} 書中提到上述指令：`kubectl scale --replicas=value，因為 K8s 迭代，透過 kubectl scale -h 得知語法已經變更。` {% endhint %} ### Scale Pod 方式二 透過 kubectl apply 進行宣告式scale`[4 --> 3]`\ 編輯 rs-hello-kitty.yaml 內容中的 replicas: value 進行管理 ```yaml apiVersion: extensions/v1beta1 kind: ReplicaSet metadata: name: hello-kitty spec: replicas: 3 # 修改此行 value ``` ```bash # 套用新 replicas value 設定 [user@minikube ~]$ kubectl apply -f kk8s/rs-hello-kitty.yaml replicaset.extensions/hello-kitty configured # 確認已經生效，從剛剛稍早 pod hello-kitty 4個，變成 3個。 [user@minikube ~]$ kubectl get rs hello-kitty NAME DESIRED CURRENT READY AGE hello-kitty 3 3 3 10h # 觀察 pod 變化，副本 hello-kitty-d7tfq 正在消失中，狀態：Terminating [user@minikube ~]$ kubectl get pods NAME READY STATUS RESTARTS AGE hello-kitty 1/1 Running 0 6d hello-kitty-d7tfq 1/1 Terminating 0 15m hello-kitty-lk7lg 1/1 Running 0 10h hello-kitty-x94s2 1/1 Running 0 10h hello-kitty-zgrb8 1/1 Running 0 15m hello-minikube-7c77b68cff-ncqmn 1/1 Running 1 11d hello-moto-6456dff9f-67jh5 1/1 Running 0 6d # 觀察 pod 變化，確認副本 pod hello-kitty-XXXXX 剩下3個。 [user@minikube ~]$ kubectl get pods NAME READY STATUS RESTARTS AGE hello-kitty 1/1 Running 0 6d hello-kitty-lk7lg 1/1 Running 0 10h hello-kitty-x94s2 1/1 Running 0 10h hello-kitty-zgrb8 1/1 Running 0 16m hello-minikube-7c77b68cff-ncqmn 1/1 Running 1 11d hello-moto-6456dff9f-67jh5 1/1 Running 0 6d ``` ### Scale Pod 方法差異 上述用了兩種方式進行 pod scale，`kubectl scale --replicas`跟`kubectl apply -f`\ \&#xNAN;***scale pod*** 目的與效果是相同的，但其實意義上，是有所不同的～\ `kubectl scale --replicas`指令，適合自己工作上運用，但自己也要清楚當前 replicas valus。\ `kubectl apply -f`是透過宣告式 yaml 檔案進行 replicas valus 管理，非常適合納入版控系統並搭配自動化腳本進行管理。進而更加適合團隊性質的協同合作需求。 你會希望我做過什麼變動，我也必須讓大家知道做過哪些變動，**這是 DevOps 精神**。 {% hint style="info" %} 參考書籍：\ Kubernetes 建置與執行 書中 P.98\~102 內容 {% endhint %}