Istio開啓mtls請求503問題分析

背景

爲測試Istio流量管理，將兩個服務sleep、flaskapp的兩個版本v1、v2(部署文件見參考鏈接)部署到Istio環境中，通過sleep-v1向flaskapp發起調用http://flaskapp/env/version，正常結果會交替打印出結果v1和v2，然而在調用過程中報錯503 reset reason: connection failure，故將問題的步驟、現象、分析、驗證整理於此。

步驟

部署sleep、flaskapp應用，同時Istio平臺開啓mTls，命名空間kangxzh開啓自動注入，部署如下圖所示：

kubectl apply -f sleep.istio.yaml -n kangxzh
kubectl apply -f flask.isito.yaml -n kangxzh

#查看pod創建情況
kubectl -n kangxzh get pod -w

flaskapp-v1-775dbb9b79-z54fj   2/2     Running           0          13s
flaskapp-v2-d454cdd47-mdb8s    2/2     Running           0          14s
sleep-v1-7f45c6cf94-zgdsf      2/2     Running           0          19h
sleep-v2-58dff94b49-fz6sj      2/2     Running           0          19h 

現象

在sleep應用中發起http請求，調用flaskapp，curl http://flaskapp/env/version，如下所示：

#
export SOURCE_POD=$(kubectl get pod -l app=sleep,version=v1 -o jsonpath={.items..metadata.name})
# 進入sleep發起http請求
kubectl -n kangxzh exec -it -c sleep $SOURCE_POD bash
bash-4.4# curl http://flaskapp/env/version
# 響應
upstream connect error or disconnect/reset before headers. reset reason: connection failure 

背景

1.檢測flaskapp tls 配置，如下：

[root@kubernetes-master flaskapp]# istioctl authn tls-check flaskapp-v1-775dbb9b79-z54fj flaskapp.kangxzh.svc.cluster.local
HOST:PORT                                 STATUS     SERVER     CLIENT     AUTHN POLICY     DESTINATION RULE
flaskapp.kangxzh.svc.cluster.local:80     OK         mTLS       mTLS       default/         default/istio-system 

STATUS OK 證明flaskapp tls配置正確。

進入sleep istio-proxy向flaskapp發起http請求：

kubectl -n kangxzh exec -it -c istio-proxy $SOURCE_POD bash
# 發起請求
curl http://flaskapp/env/version
# 響應
v1

2.發現通過istio-proxy可以得到相應，因爲開啓了mtls，通過istio-proxy直接請求是需要添加istio相關證書的，此時沒有加入證書也可請求，所以想到檢查flaskapp iptables配置，如下所示：

# 獲取進程號
PID=$(docker inspect --format {{.State.Pid}} $(docker ps | grep flaskapp-v1 | awk '{print $1}' | head -n 1))
# 查看iptables 規則
nsenter -t ${PID} -n iptables -t nat -L -n -v
# 輸出
Chain PREROUTING (policy ACCEPT 477 packets, 28620 bytes)
 pkts bytes target     prot opt in     out     source               destination
  487 29220 ISTIO_INBOUND  tcp  --  *      *       0.0.0.0/0            0.0.0.0/0

Chain INPUT (policy ACCEPT 487 packets, 29220 bytes)
 pkts bytes target     prot opt in     out     source               destination

Chain OUTPUT (policy ACCEPT 220 packets, 20367 bytes)
 pkts bytes target     prot opt in     out     source               destination
    0   480 ISTIO_OUTPUT  tcp  --  *      *       0.0.0.0/0            0.0.0.0/0

Chain POSTROUTING (policy ACCEPT 220 packets, 20367 bytes)
 pkts bytes target     prot opt in     out     source               destination

Chain ISTIO_INBOUND (1 references)
 pkts bytes target     prot opt in     out     source               destination
   10   600 ISTIO_IN_REDIRECT  tcp  --  *      *       0.0.0.0/0            0.0.0.0/0            tcp dpt:80    #是沒有的，修改後增加

Chain ISTIO_IN_REDIRECT (1 references)
 pkts bytes target     prot opt in     out     source               destination
   10   600 REDIRECT   tcp  --  *      *       0.0.0.0/0            0.0.0.0/0            redir ports 15001

Chain ISTIO_OUTPUT (1 references)
 pkts bytes target     prot opt in     out     source               destination
    0     0 ISTIO_REDIRECT  all  --  *      lo      0.0.0.0/0           !127.0.0.1
    8   480 RETURN     all  --  *      *       0.0.0.0/0            0.0.0.0/0            owner UID match 1337
    0     0 RETURN     all  --  *      *       0.0.0.0/0            0.0.0.0/0            owner GID match 1337
    0     0 RETURN     all  --  *      *       0.0.0.0/0            127.0.0.1
    0     0 ISTIO_REDIRECT  all  --  *      *       0.0.0.0/0            0.0.0.0/0

Chain ISTIO_REDIRECT (2 references)
 pkts bytes target     prot opt in     out     source               destination
    0     0 REDIRECT   tcp  --  *      *       0.0.0.0/0            0.0.0.0/0            redir ports 15001 

證明envoy沒有劫持到flaskapp 80的流量，也就是說上述第2步是sleep-istio-proxy直接請求flaskapp，沒有經過flaskapp-istio-proxy 轉發。

3. 此時才檢查flaskapp deployment，如下所示：

...

---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: flaskapp-v1
spec:
  replicas: 1
  template:
    metadata:
      labels:
        app: flaskapp
        version: v1
    spec:
      containers:
      - name: flaskapp
        image: dustise/flaskapp
        imagePullPolicy: IfNotPresent
        ports:
          - name: http
            containerPort: 80  #缺少containerPort
        env:
        - name: version
          value: v1
... 

官網說明https://istio.io/docs/setup/kubernetes/additional-setup/requirements/中：

Pod ports: Pods must include an explicit list of the ports each container listens on. Use a containerPort configuration in the container specification for each port. Any unlisted ports bypass the Istio proxy.
# 未列出來來的端口都會繞過istio proxy 

同時describe flaskapp pod 如下所示：

kubectl describe pod flaskapp-v1-6df8d69fb8-fb5mr -n kangxzh
#
istio-proxy:
    ... 省略若干
    Args:
      ...
      --concurrency
      2
      --controlPlaneAuthPolicy
      MUTUAL_TLS
      --statusPort
      15020
      --applicationPorts
      "" #爲空 

在deployment中增加containerPort: 80後，如下所示：

istio-proxy:
    ... 省略若干
    Args:
      ...
      --concurrency
      2
      --controlPlaneAuthPolicy
      MUTUAL_TLS
      --statusPort
      15020
      --applicationPorts
      80 

注意：在Istio1.2版本以後也可通過設置Pod annotation 中 traffic.sidecar.istio.io/includeInboundPorts來達到同樣的目的，缺省值爲Pod的containerPorts列表，逗號分隔的監聽端口列表，這些流量會被重定向到 Sidecar，* 會重定向所有端口，具體詳情參見官網1.2新特性(見參考鏈接)

驗證

sleep 發起請求：

[root@kubernetes-master flaskapp]# kubectl -n kangxzh exec -it -c sleep $SOURCE_POD bash
bash-4.4# curl http://flaskapp/env/version
#響應
v1 

sleep-istio-proxy 未攜帶證書，發起請求：

kubectl -n kangxzh exec -it -c istio-proxy $SOURCE_POD bash
istio-proxy@sleep-v1-7f45c6cf94-zgdsf:/$ curl http://flaskapp/env/version
#響應
curl: (56) Recv failure: Connection reset by peer 
sleep-istio-proxy 攜帶證書，發起請求
istio-proxy@sleep-v1-7f45c6cf94-zgdsf:/$ curl https://flaskapp:80/env/version --key /etc/certs/key.pem --cert /etc/certs/cert-chain.pem --cacert /etc/certs/root-cert.pem -k
#響應
v1 

參考鏈接

https://github.com/fleeto/sleep

https://github.com/fleeto/flaskapp

https://istio.io/docs/setup/kubernetes/additional-setup/requirements/

https://preliminary.istio.io/about/notes/1.2/