最簡單的 K8S 部署檔案編寫姿勢，沒有之一！

kevwan發表於2020-12-12

K8S

1. 頭疼編寫 K8S 部署檔案？

K8S yaml 引數很多，需要邊寫邊查？
保留回滾版本數怎麼設？
如何探測啟動成功，如何探活？
如何分配和限制資源？
如何設定時區？否則列印日誌是 GMT 標準時間
如何暴露服務供其它服務呼叫？
如何根據 CPU 和記憶體使用率來配置水平伸縮？

首先，你需要知道有這些知識點，其次要把這些知識點都搞明白也不容易，再次，每次編寫依然容易出錯！

2. 建立服務映象

前一篇文章講解了如何快速建立自己的服務映象，不過為了演示，這篇文章我們以 redis:6-alpine 映象為例。

3. 完整 K8S 部署檔案編寫過程

首先安裝 goctl 工具

GO111MODULE=on GOPROXY=https://goproxy.cn/,direct go get -u github.com/tal-tech/go-zero/tools/goctl

一鍵生成 K8S 部署檔案

goctl kube deploy -name redis -namespace adhoc -image redis:6-alpine -o redis.yaml -port 6379

生成的 yaml 檔案如下：

apiVersion: apps/v1
kind: Deployment
metadata:
  name: redis
  namespace: adhoc
  labels:
    app: redis
spec:
  replicas: 3
  revisionHistoryLimit: 5
  selector:
    matchLabels:
      app: redis
  template:
    metadata:
      labels:
        app: redis
    spec:
      containers:
      - name: redis
        image: redis:6-alpine
        lifecycle:
          preStop:
            exec:
              command: ["sh","-c","sleep 5"]
        ports:
        - containerPort: 6379
        readinessProbe:
          tcpSocket:
            port: 6379
          initialDelaySeconds: 5
          periodSeconds: 10
        livenessProbe:
          tcpSocket:
            port: 6379
          initialDelaySeconds: 15
          periodSeconds: 20
        resources:
          requests:
            cpu: 500m
            memory: 512Mi
          limits:
            cpu: 1000m
            memory: 1024Mi
        volumeMounts:
        - name: timezone
          mountPath: /etc/localtime
      volumes:
        - name: timezone
          hostPath:
            path: /usr/share/zoneinfo/Asia/Shanghai

---

apiVersion: v1
kind: Service
metadata:
  name: redis-svc
  namespace: adhoc
spec:
  ports:
    - port: 6379
  selector:
    app: redis

---

apiVersion: autoscaling/v2beta1
kind: HorizontalPodAutoscaler
metadata:
  name: redis-hpa-c
  namespace: adhoc
  labels:
    app: redis-hpa-c
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: redis
  minReplicas: 3
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      targetAverageUtilization: 80

---

apiVersion: autoscaling/v2beta1
kind: HorizontalPodAutoscaler
metadata:
  name: redis-hpa-m
  namespace: adhoc
  labels:
    app: redis-hpa-m
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: redis
  minReplicas: 3
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: memory
      targetAverageUtilization: 80

部署服務，如果 adhoc namespace 不存在的話，請先通過 kubectl create namespace adhoc 建立

$ kubectl apply -f redis.yaml
deployment.apps/redis created
service/redis-svc created
horizontalpodautoscaler.autoscaling/redis-hpa-c created
horizontalpodautoscaler.autoscaling/redis-hpa-m created

檢視服務允許狀態

$ kubectl get all -n adhoc
NAME                         READY   STATUS    RESTARTS   AGE
pod/redis-585bc66876-5ph26   1/1     Running   0          6m5s
pod/redis-585bc66876-bfqxz   1/1     Running   0          6m5s
pod/redis-585bc66876-vvfc9   1/1     Running   0          6m5s

NAME                TYPE        CLUSTER-IP    EXTERNAL-IP   PORT(S)    AGE
service/redis-svc   ClusterIP   172.24.15.8   <none>        6379/TCP   6m5s

NAME                    READY   UP-TO-DATE   AVAILABLE   AGE
deployment.apps/redis   3/3     3            3           6m6s

NAME                               DESIRED   CURRENT   READY   AGE
replicaset.apps/redis-585bc66876   3         3         3       6m6s

NAME                                              REFERENCE          TARGETS   MINPODS   MAXPODS   REPLICAS   AGE
horizontalpodautoscaler.autoscaling/redis-hpa-c   Deployment/redis   0%/80%    3         10        3          6m6s
horizontalpodautoscaler.autoscaling/redis-hpa-m   Deployment/redis   0%/80%    3         10        3          6m6s

測試服務

$ kubectl run -i --tty --rm cli --image=redis:6-alpine -n adhoc -- sh
/data # redis-cli -h redis-svc
redis-svc:6379> set go-zero great
OK
redis-svc:6379> get go-zero
"great"