5 个 Istio 访问外部服务的流量控制常用例子,强烈建议收藏起来,以备不时之需。
环境准备
部署 sleep
服务,作为发送请求的测试源:
kubectl apply -f samples/sleep/sleep.yaml
在 Istio 外部,使用 Nginx 搭建 duckling
服务的v1和v2两个版本,访问时显示简单的文本:
> curl -s http://192.168.1.118/
This is the v1 version of duckling.
> curl -s http://192.168.1.119/
This is the v2 version of duckling.
访问外部服务
执行如下命名访问外部服务 httpbin.org :
export SLEEP_POD=$(kubectl get pods -l app=sleep -o 'jsonpath={.items[0].metadata.name}')
kubectl exec "$SLEEP_POD" -c sleep -- curl -s http://httpbin.org/headers
返回结果如下:
{"headers": {"Accept": "*/*", "Host": "httpbin.org", "User-Agent": "curl/7.81.0-DEV", "X-Amzn-Trace-Id": "Root=1-62bbfa10-3237e3b9662c65ae005148ab", "X-B3-Sampled": "0", "X-B3-Spanid": "9e650093bf7ae862", "X-B3-Traceid": "1da46d7fafa5d71c9e650093bf7ae862", "X-Envoy-Attempt-Count": "1", "X-Envoy-Peer-Metadata": "......", "X-Envoy-Peer-Metadata-Id": "sidecar~......"}
}
此时的方法,没有通过Service Entry,没有 Istio 的流量监控和控制特性。创建 Service Entry :
kubectl apply -f - <<EOF
apiVersion: networking.istio.io/v1alpha3
kind: ServiceEntry
metadata:name: httpbin-ext
spec:hosts:- httpbin.orgports:- number: 80name: httpprotocol: HTTPresolution: DNSlocation: MESH_EXTERNAL
EOF
再此次访问,返回结果如下:
{"headers": {"Accept": "*/*", "Host": "httpbin.org", "User-Agent": "curl/7.81.0-DEV", "X-Amzn-Trace-Id": "Root=1-62bbfbd6-254b05344b3cde2c0c41b3b8", "X-B3-Sampled": "0", "X-B3-Spanid": "307c0b106c8b262e", "X-B3-Traceid": "f684a50776c088ac307c0b106c8b262e", "X-Envoy-Attempt-Count": "1", "X-Envoy-Decorator-Operation": "httpbin.org:80/*", "X-Envoy-Peer-Metadata": "......", "X-Envoy-Peer-Metadata-Id": "sidecar~......"}
}
可以发现由 Istio 边车添加的请求头:X-Envoy-Decorator-Operation
。
设置请求超时
向外部服务 httpbin.org 的 /delay 发出请求:
export SLEEP_POD=$(kubectl get pods -l app=sleep -o 'jsonpath={.items[0].metadata.name}')
kubectl exec "$SLEEP_POD" -c sleep -- time curl -o /dev/null -sS -w "%{http_code}\n" http://httpbin.org/delay/5
返回结果如下:
200
real 0m 5.69s
user 0m 0.00s
sys 0m 0.00s
请求大约在 5 秒后返回 200 (OK)。
创建虚拟服务,访问外部服务 httpbin.org 时, 请求超时设置为 3 秒:
kubectl apply -f - <<EOF
apiVersion: networking.istio.io/v1alpha3
kind: VirtualService
metadata:name: httpbin-ext
spec:hosts:- httpbin.orghttp:- timeout: 3sroute:- destination:host: httpbin.orgweight: 100
EOF
再此次访问,返回结果如下:
504
real 0m 3.01s
user 0m 0.00s
sys 0m 0.00s
可以看出,在 3 秒后出现了 504 (Gateway Timeout)。 Istio 在 3 秒后切断了响应时间为 5 秒的 httpbin.org 服务。
注入 HTTP 延迟故障
向外部服务 httpbin.org 的 /get 发出请求:
export SLEEP_POD=$(kubectl get pods -l app=sleep -o 'jsonpath={.items[0].metadata.name}')
kubectl exec "$SLEEP_POD" -c sleep -- time curl -o /dev/null -sS -w "%{http_code}\n" http://httpbin.org/get
返回结果如下:
200
real 0m 0.45s
user 0m 0.00s
sys 0m 0.00s
请求不到 1 秒就返回 200 (OK)。
创建虚拟服务,访问外部服务 httpbin.org 时, 注入一个 3 秒的延迟:
kubectl apply -f - <<EOF
apiVersion: networking.istio.io/v1alpha3
kind: VirtualService
metadata:name: httpbin-ext
spec:hosts:- httpbin.orghttp:- fault:delay:fixedDelay: 3spercentage:value: 100route:- destination:host: httpbin.org
EOF
再此次访问 httpbin.org 的 /get ,返回结果如下:
200
real 0m 3.43s
user 0m 0.00s
sys 0m 0.00s
可以看出,在 3 秒后出现了 200 (OK)。
流量转移
访问duckling
服务时,所有流量都路由到v1版本,具体配置如下:
kubectl apply -f - <<EOF
apiVersion: networking.istio.io/v1alpha3
kind: ServiceEntry
metadata:name: duckling
spec:hosts:- duckling.comports:- number: 80name: httpprotocol: HTTPlocation: MESH_EXTERNALresolution: STATICendpoints:- address: 172.24.29.118ports:http: 80labels:version: v1- address: 172.24.29.119ports:http: 80labels:version: v2
---
apiVersion: networking.istio.io/v1alpha3
kind: VirtualService
metadata:name: duckling
spec:hosts:- duckling.comhttp:- route:- destination:host: duckling.comsubset: v1
---
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:name: duckling
spec:host: duckling.comsubsets:- labels:version: v1name: v1- labels:version: v2name: v2
EOF
执行如下命名访问外部服务 duckling.com :
export SLEEP_POD=$(kubectl get pods -l app=sleep -o 'jsonpath={.items[0].metadata.name}')
kubectl exec "$SLEEP_POD" -c sleep -- curl -s http://duckling.com/
多次访问后,返回结果一直是:This is the v1 version of duckling.
访问duckling
服务时,把50%流量转移到v2版本,具体配置如下:
kubectl apply -f - <<EOF
apiVersion: networking.istio.io/v1alpha3
kind: VirtualService
metadata:name: duckling
spec:hosts:- duckling.comhttp:- route:- destination:host: duckling.comsubset: v1weight: 50- destination:host: duckling.comsubset: v2weight: 50
EOF
多次访问外部服务 duckling.com ,有时返回This is the v1 version of duckling.
,有时返回This is the v2 version of duckling.
。
访问duckling
服务时,所有流量都路由到v2版本,具体配置如下:
kubectl apply -f - <<EOF
apiVersion: networking.istio.io/v1alpha3
kind: VirtualService
metadata:name: duckling
spec:hosts:- duckling.comhttp:- route:- destination:host: duckling.comsubset: v2
EOF
多次访问外部服务 duckling.com ,一直返回This is the v2 version of duckling.
。
基于请求头的路由
请求头end-user
为OneMore的所有流量都路由到v2版本,其他流量都路由到v1版本,具体配置如下:
kubectl apply -f - <<EOF
apiVersion: networking.istio.io/v1alpha3
kind: ServiceEntry
metadata:name: duckling
spec:hosts:- duckling.comports:- number: 80name: httpprotocol: HTTPlocation: MESH_EXTERNALresolution: STATICendpoints:- address: 172.24.29.118ports:http: 80labels:version: v1- address: 172.24.29.119ports:http: 80labels:version: v2
---
apiVersion: networking.istio.io/v1alpha3
kind: VirtualService
metadata:name: duckling
spec:hosts:- duckling.comhttp:- match:- headers:end-user:exact: OneMoreroute:- destination:host: duckling.comsubset: v2- route:- destination:host: duckling.comsubset: v1
---
apiVersion: networking.istio.io/v1alpha3
kind: DestinationRule
metadata:name: duckling
spec:host: duckling.comsubsets:- labels:version: v1name: v1- labels:version: v2name: v2
EOF
执行如下命名访问外部服务 duckling.com :
export SLEEP_POD=$(kubectl get pods -l app=sleep -o 'jsonpath={.items[0].metadata.name}')
kubectl exec "$SLEEP_POD" -c sleep -- curl -s http://duckling.com/
多次访问的返回结果一直是:This is the v1 version of duckling.
设置请求头end-user
为OneMore,访问外部服务 duckling.com :
kubectl exec "$SLEEP_POD" -c sleep -- curl -H "end-user:OneMore" -s http://duckling.com/
多次访问的返回结果一直是:This is the v2 version of duckling.