备考ICA----Istio实验4—使用 Istio 进行金丝雀部署

上一个实验已经通过DestinationRule实现了部分金丝雀部署的功能,这个实验会更完整的模拟展示一个环境由v1慢慢过渡到v2版本的金丝雀发布.

1. 环境清理

kubectl delete gw/helloworld-gateway vs/helloworld dr/helloworld-destination

测试

kubectl get svc,pods
for i in {1..10};do curl $(kubectl get svc helloworld|grep helloworld|awk '{print $3":"$5}'|awk -F"/" '{print $1"/hello"}');sleep .5 ;done
kubectl get gw,vs,dr

恢复到这样就可以通过helloworld的svc将流量随机分配到v1和v2上
如果实验环境有问题,就重新部署hello

kubectl delete -f istio/samples/helloworld/helloworld.yaml
kubectl apple -f istio/samples/helloworld/helloworld.yaml

2. 所有流量转发到v1

这步就模拟只存在1个版本的环境
canary/helloworld-canary-all-v1.yaml

apiVersion: networking.istio.io/v1beta1
kind: Gateway
metadata:
  name: helloworld-gateway
spec:
  selector:
    istio: ingressgateway # use istio default controller
  servers:
  - port:
      number: 80
      name: http
      protocol: HTTP
    hosts:
    - "*"
---
apiVersion: networking.istio.io/v1beta1
kind: DestinationRule
metadata:
  name: helloworld-destination
spec:
  host: helloworld
  subsets:
  - name: v1
    labels:
      version: v1
  - name: v2
    labels:
      version: v2
---
apiVersion: networking.istio.io/v1beta1
kind: VirtualService
metadata:
  name: helloworld
spec:
  hosts:
  - "*"
  gateways:
  - helloworld-gateway
  http:
  - match:
    - uri:
        exact: /hello
    route:
    - destination:
        host: helloworld
        port:
          number: 5000
        subset: v1
      weight: 100

部署gw,vs,dr

kubectl apply -f canary/helloworld-canary-all-v1.yaml

测试效果
此时所有流量都交由v1进行响应

for i in {1..10};do curl http://192.168.126.220/hello;sleep .5;done

3. 90%流量v1,10%流量v2

此时v2版本应用已经上线,将10%流量给v2,其余流量仍由v1进行应答

3.1 配置流量分发比例

canary/helloworld-canary-allin1-10v2.yaml

apiVersion: networking.istio.io/v1beta1
kind: Gateway
metadata:
  name: helloworld-gateway
spec:
  selector:
    istio: ingressgateway # use istio default controller
  servers:
  - port:
      number: 80
      name: http
      protocol: HTTP
    hosts:
    - "*"
---
apiVersion: networking.istio.io/v1beta1
kind: DestinationRule
metadata:
  name: helloworld-destination
spec:
  host: helloworld
  subsets:
  - name: v1
    labels:
      version: v1
  - name: v2
    labels:
      version: v2
---
apiVersion: networking.istio.io/v1beta1
kind: VirtualService
metadata:
  name: helloworld
spec:
  hosts:
  - "*"
  gateways:
  - helloworld-gateway
  http:
  - match:
    - uri:
        exact: /hello
    route:
    - destination:
        host: helloworld
        port:
          number: 5000
        subset: v1
      weight: 90
    - destination:
        host: helloworld
        port:
          number: 5000
        subset: v2
      weight: 10

部署gw,vs,dr

kubectl apply -f canary/helloworld-canary-allin1-10v2.yaml 

测试效果
可以看到10个请求中有1个由v2应答,其他仍由v1进行响应

3.2 加入Hpa

---
apiVersion: autoscaling/v1
kind: HorizontalPodAutoscaler
metadata:
  name: hpa-helloworld-v1
spec:
  maxReplicas: 20
  minReplicas: 1
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: helloworld-v1
  targetCPUUtilizationPercentage: 50
---
apiVersion: autoscaling/v1
kind: HorizontalPodAutoscaler
metadata:
  name: hpa-helloworld-v2
spec:
  maxReplicas: 20
  minReplicas: 1
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: helloworld-v2
  targetCPUUtilizationPercentage: 50

部署hpa

kubectl apply -f canary/hpa.yaml

3.3 压测

 while true;do curl http://192.168.126.220/hello;done

产生大量请求

此时v1,v2因访问量大触发hpa扩容,直到v1到达上线16个pod,v2到达3个

4. 50%流量v1,50%流量v2

4.1 配置流量分发比例

dr和gw部分就不用动了.只要修改vs的weight部分就可以
canary/helloworld-canary-vs-50v2.yaml

---
apiVersion: networking.istio.io/v1beta1
kind: VirtualService
metadata:
  name: helloworld
spec:
  hosts:
  - "*"
  gateways:
  - helloworld-gateway
  http:
  - match:
    - uri:
        exact: /hello
    route:
    - destination:
        host: helloworld
        port:
          number: 5000
        subset: v1
      weight: 50
    - destination:
        host: helloworld
        port:
          number: 5000
        subset: v2
      weight: 50

部署

kubectl apply -f canary/helloworld-canary-vs-50v2.yaml

4.2 压测

此时流量以1:1分发给v1和v2

再观测hpa的情况会发现v2的cpu逐渐升高,v1的cpu逐渐降低,v2开始扩容,v1开始缩容,逐渐扩缩容到10:10

5. 所有流量转发v2

51. 配置流量分发比例

中间的10%,90%其实和前2个版本差不多,直接修改下数值就可以了.我们这里就忽略了,有兴趣的老哥可以进一步的修改模拟.
这里就模拟经过测试v2版本已经没有问题,我们将所有流量打到v2上
canary/helloworld-canary-all-v2.yaml

---
apiVersion: networking.istio.io/v1beta1
kind: VirtualService
metadata:
  name: helloworld
spec:
  hosts:
  - "*"
  gateways:
  - helloworld-gateway
  http:
  - match:
    - uri:
        exact: /hello
    route:
    - destination:
        host: helloworld
        port:
          number: 5000
        subset: v2
      weight: 100

部署

kubectl apply -f canary/helloworld-canary-all-v2.yaml

5.2 压测

while true;do curl http://192.168.126.220/hello;done

至此canary的一个模拟从v1到v2的版本切换就已经完成了

6. 拓展Canary+AB测试

6.1 canary+ab配置

当我们进行canary测试的时候,普通用户是以50%:50%的流量分发到2个版本上,但我们希望测试人员trump同学,每次都是访问到新上线的v2版本上.
canary/canary-ab-vs.yaml

apiVersion: networking.istio.io/v1beta1
kind: VirtualService
metadata:
  name: helloworld
spec:
  hosts:
  - "*"
  gateways:
  - helloworld-gateway
  http:
  - match:
    - headers:
        user:
          exact: trump
      uri:
        exact: /hello
    route:
    - destination:
        host: helloworld
        port:
          number: 5000
        subset: v2
      weight: 100
  - route:
    - destination:
        host: helloworld
        port:
          number: 5000
        subset: v1
      weight: 50
    - destination:
        host: helloworld
        port:
          number: 5000
        subset: v2
      weight: 50

部署应用

kubectl apply -f canary/canary-ab-vs.yaml

6.2 测试

6.2.1 普通用户测试

这部分用户进准的按1:1流量访问v1和v2

for i in {1..20};do curl http://192.168.126.220/hello;done

6.2.2 测试人员访问

当测试人员trump访问时,匹配header中的用户名为trump,流量就被100%的打到v2版本上

for i in {1..20};do curl -H "user:trump" http://192.168.126.220/hello;done

至此整个金丝雀部署完成