kubernetes Metrics-server 安裝

一、Metrics-server安裝：

由於使用自動擴容，得安裝heapster或metrics-server（由於heapster停止更新故用metrics-server）：

1、安裝之前需要為kubernetes增加配置項：

1、為/etc/kubernetes/controller-manager增加啟動項：
  ***
    --horizontal-pod-autoscaler-use-rest-clients=true
  ***
 systemctl restart kube-controller-manager

2、為/usr/lib/systemd/system/kube-apiserver.service增加啟動項（也可以修改/etc/kubernetes/apiserver文件增加）：
  ***

--requestheader-client-ca-file ./cert/ca.pem \
--requestheader-allowed-names= \
--requestheader-extra-headers-prefix=X-Remote-Extra- \
--requestheader-group-headers=X-Remote-Group \
--requestheader-username-headers=X-Remote-User \
--proxy-client-cert-file ./cert/kube-proxy-client.pem \
--proxy-client-key-file ./cert/kube-proxy-client-key.pem \
--enable-aggregator-routing=true 

  ***
 systemctl daemon-reload
 systemctl restart kube-apiserver

2、在kubenetes上安裝Metrics-server：

1、創建：
kubectl create -f metrics-server.yml 

2、刪除：
kubectl delete ServiceAccount metrics-server -n kube-system
kubectl delete ClusterRoleBinding metrics-server:system:auth-delegator -n kube-system
kubectl delete RoleBinding metrics-server-auth-reader -n kube-system
kubectl delete ClusterRole system:metrics-server -n kube-system
kubectl delete  ClusterRoleBinding system:metrics-server -n kube-system
kubectl delete  APIService v1beta1.metrics.k8s.io -n kube-system
kubectl delete Service metrics-server -n kube-system
kubectl delete Deployment metrics-server -n kube-system

注：要修改metrics-server的鏡

 3、驗證：

1、創建測試資源：
kubectl create -f nginx.yml

2、創建自動伸縮規則：
kubectl autoscale deployment web --cpu-percent=75 --min=2 --max=10
kubectl get hpa

NAME           REFERENCE                 TARGETS    MINPODS   MAXPODS   REPLICAS   AGE
bill-service   Deployment/bill-service   0% / 80%   5         10        5          2h
注：TARGETS出現 <unknown> / 80% 說明沒有成功就得看metrics-server的日志和hpa的describe來查找原因了。

3、測試是否生效（使用腳本測）：
python pressure-web.py
kubectl get hpa

apiVersion: v1
kind: Service
metadata:
  name: nginx
  labels:
    app: nginx
spec:
  ports:
  - port: 80
    name: web
  selector:
    app: nginx
---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: web
spec:
  selector:
    matchLabels:
      app: nginx 
  replicas: 3 
  template:
    metadata:
      labels:
        app: nginx 
    spec:
      containers:
      - name: nginx
        image: nginx
        resources:
          limits:
            cpu: 100m
            memory: 100Mi
          requests:
            cpu: 50m
            memory: 50Mi
        ports:
        - containerPort: 80
          name: web

import requests,time,threading

url = "https://www.baidu.com"
requests_timeout = 2


def get_response_time(url,count,ring):
    start_time = time.time()
    try:
        req_obj = requests.get(url,timeout=requests_timeout)
    except requests.exceptions.ReadTimeout as e:
        status = False
    else:
        status = True
    end_time = time.time()
    requests_time = end_time-start_time
    if status and (req_obj.status_code == requests.codes.ok) :
        requests_status = "OK"
    else:
        requests_status = "NO"
    print("第%s輪，第%s連接，請求用時（s）：%s，返回狀態：%s"%(ring,count,requests_time,requests_status))


if __name__ == "__main__":
    '''在規定超時時間內，11秒內處理不了100個請求，請求就會累積，一直拖垮'''
    ring = 1
    while True:
        for count in range(100):
            t = threading.Thread(target=get_response_time,args=(url,count,ring))
            time.sleep(0.01)
            t.start()
        time.sleep(10)
        ring += 1
        print("第%s輪開始"%ring)