從docker到istio之三 - kubernetes集羣編排

前言

容器化,雲原生越演越烈,新概念很是之多。信息爆炸的同時,帶來層層迷霧。我嘗試從擴容出發理解其脈路,通過實踐探索,整理造成一個入門教程,包括下面四篇文章。html

這是第三篇,kubernetes編排應用。node

kubernetes

Kubernetes是一個開源的,用於管理雲平臺中多個主機上的容器化的應用,Kubernetes的目標是讓部署容器化的應用簡單而且高效(powerful),Kubernetes提供了應用部署,規劃,更新,維護的一種機制。python

Kubernetes在希臘語中意思是船長或領航員,這也剛好與它在容器集羣管理中的做用吻合,即做爲裝載了集裝箱(Container)的衆多貨船的指揮者,負擔着全局調度和運行監控的職責。由於Kubernetes在k和s之間有8個字母,因此又簡稱k8slinux

快速體驗k8s,可使用Docker for mac中集成的k8s。nginx

DockerForMac

啓動k8s後,等待其初始化完成,而後 docker ps 能夠看到k8s啓動了一系列的容器:git

CONTAINER ID        IMAGE                            COMMAND                  CREATED             STATUS              PORTS               NAMES
17a693617137        docker/kube-compose-controller   "/compose-controller…"   3 days ago          Up 3 days                               k8s_compose_compose-74649b4db6-szsqz_docker_4f5997b7-5c47-11e9-95b9-025000000001_0
a9b666b48815        docker/kube-compose-api-server   "/api-server --kubec…"   3 days ago          Up 3 days                               k8s_compose_compose-api-5d754cdd89-ncwrq_docker_131b4d65-04e7-11e9-837c-025000000001_0
f4b05eefc73a        6f7f2dc7fab5                     "/sidecar --v=2 --lo…"   3 days ago          Up 3 days                               k8s_sidecar_kube-dns-86f4d74b45-zh6qc_kube-system_f669bc59-04e6-11e9-837c-025000000001_0
867f8f040258        c2ce1ffb51ed                     "/dnsmasq-nanny -v=2…"   3 days ago          Up 3 days                               k8s_dnsmasq_kube-dns-86f4d74b45-zh6qc_kube-system_f669bc59-04e6-11e9-837c-025000000001_0
17f26a6e91d2        80cc5ea4b547                     "/kube-dns --domain=…"   3 days ago          Up 3 days                               k8s_kubedns_kube-dns-86f4d74b45-zh6qc_kube-system_f669bc59-04e6-11e9-837c-025000000001_0

...

複製代碼

kubectl version 查看集羣版本:github

Client Version: version.Info{Major:"1", Minor:"10", GitVersion:"v1.10.11", GitCommit:"637c7e288581ee40ab4ca210618a89a555b6e7e9", GitTreeState:"clean", BuildDate:"2018-11-26T14:38:32Z", GoVersion:"go1.9.3", Compiler:"gc", Platform:"darwin/amd64"}
Server Version: version.Info{Major:"1", Minor:"10", GitVersion:"v1.10.11", GitCommit:"637c7e288581ee40ab4ca210618a89a555b6e7e9", GitTreeState:"clean", BuildDate:"2018-11-26T14:25:46Z", GoVersion:"go1.9.3", Compiler:"gc", Platform:"linux/amd64"}
複製代碼

kubectl get nodes 查看k8s集羣節點:redis

NAME                 STATUS    ROLES     AGE       VERSION
docker-for-desktop   Ready     master    123d      v1.10.11
複製代碼

kubectl get service查看k8s默認啓動的服務:docker

NAME         TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)   AGE
kubernetes   ClusterIP   10.96.0.1    <none>        443/TCP   123d
複製代碼

部署應用及測試

編寫應用部署文件

1. flaskapp文件k8s/flaskapp.yaml

apiVersion: v1
kind: Service
metadata:
  name: flaskapp
spec:
  ports:
    - port: 5000
  selector:
    name: flaskapp
---
apiVersion: extensions/v1beta1 
kind: Deployment
metadata:
  name: flaskapp
spec:
  replicas: 1
  template:
    metadata:
      labels:
        name: flaskapp
    spec:
      containers:
      - image: flaskapp:0.0.2
        name: flaskapp
        ports:
        - containerPort: 5000
複製代碼

瞭解這個部署文件,須要先大概瞭解一下k8s的運做方式。k8s經過api server提供restful接口,用於集羣交互。每個部署對象,都有apiVersionkind, metadata, spec這幾個關鍵字。flask

  • 定義了Service和Deployment2個類型的對象。Service表示k8s對外提供的服務,Deployment表示某個service的部署方式。
  • Service對象的ports描述了服務端口,這個是集羣內部網絡的端口。
  • Service對象的selector描述了服務如何選擇對於的部署,採用標籤name: flaskapp,這是一種解耦合的依賴關係。
  • Deployment的replicas描述了容器的副本個數,下文會演示如何擴充。
  • Deployment的containers描述了鏡像名稱,服務端口等。

2. redis服務文件k8s/redis.yaml

apiVersion: v1
kind: Service
metadata:
  name: redis
spec:
  ports:
    - port: 6379
  selector:
    name: redis
---
apiVersion: extensions/v1beta1 
kind: Deployment
metadata:
  name: redis
spec:
  replicas: 1
  template:
    metadata:
      labels:
        name: redis
    spec:
      containers:
      - image: redis:4-alpine3.8
        name: redis
        ports:
        - containerPort: 6379
複製代碼

redis的部署文件和flaskapp的部署文件相似。

3. nginx服務文件k8s/nginx.yaml

kind: ConfigMap
apiVersion: v1
metadata:
  name: nginx-config
data:
  default.conf: |
    upstream flaskapp {
        server flaskapp:5000;
    }
    server {
        listen       80;
        server_name  localhost;

        root /usr/share/nginx/html;

        location / {
            proxy_pass http://flaskapp;
            proxy_set_header X-Real-IP $remote_addr;
            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
            proxy_set_header Host $host;
            proxy_redirect off;
        }

    }
---
apiVersion: v1
kind: Service
metadata:
  name: nginx
spec:
  ports:
    - port: 80
  selector:
    name: nginx
  type: NodePort
---
apiVersion: extensions/v1beta1 
kind: Deployment
metadata:
  name: nginx
spec:
  replicas: 1
  template:
    metadata:
      labels:
        name: nginx
    spec:
      containers:
      - image: nginx:1.15.8-alpine
        name: nginx
        ports:
        - containerPort: 80
        volumeMounts:
          - name: nginx-config-volume
            mountPath: /etc/nginx/conf.d/default.conf
            subPath: default.conf
      volumes:
        - name: nginx-config-volume
          configMap:
            name: nginx-config
複製代碼

nginx的部署文件,變化在:

  • 多出了ConfigMap對象,這個對象主要定義了nginx.conf文件,其內容和nginx\default.conf一致。
  • nginx的container中mount了一個configmap對象做爲nginx的配置文件。

部署應用到集羣

使用kubectl apply -f k8s命令將編寫yaml文件提交到k8s集羣,集羣會自動根據yaml文件的聲明,進行部署。

service "flaskapp" created
deployment.extensions "flaskapp" created
configmap "nginx-config" created
service "nginx" created
deployment.extensions "nginx" created
service "redis" created
deployment.extensions "redis" created
複製代碼

這裏的kubectl apply -f k8s 表示將k8s目錄下的文件都提交給k8s集羣。固然,也能夠逐個文件提交 kubectl apply -f k8s/redis.yaml

訪問應用

kubectl get service檢查一下k8s內的服務:

NAME         TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)        AGE
flaskapp     ClusterIP   10.110.202.47    <none>        5000/TCP       31s
kubernetes   ClusterIP   10.96.0.1        <none>        443/TCP        123d
nginx        NodePort    10.100.233.149   <none>        80:30457/TCP   31s
redis        ClusterIP   10.106.55.214    <none>        6379/TCP       31s
複製代碼

注意nginx服務部分的PORTS爲80:30457/TCP,這表示將容器的80端口暴露到本機網絡的30457端口,和咱們以前的docker啓動時候的-p 80:80參數相似。

服務是由Pod提供的,繼續檢查一下pods的情況kubectl get pods:

NAME                        READY     STATUS    RESTARTS   AGE
flaskapp-6c4fccdf99-v6w2v   1/1       Running   0          2m
nginx-85fb469b96-lr982      1/1       Running   0          2m
redis-5b44bb8d97-wwmll      1/1       Running   0          2m
複製代碼

固然,也能夠直接查看docker的容器docker ps:

➜  docker2istio docker ps
CONTAINER ID        IMAGE                            COMMAND                  CREATED             STATUS              PORTS               NAMES
ad7377ae7196        ae70b17240ec                     "docker-entrypoint.s…"   About an hour ago   Up About an hour                        k8s_redis_redis-5b44bb8d97-wwmll_default_2907f4a3-6639-11e9-b8cb-025000000001_0
c01108b49076        1a61773c4c07                     "python flaskapp.py"     About an hour ago   Up About an hour                        k8s_flaskapp_flaskapp-6c4fccdf99-xcmwb_default_28fbe1b1-6639-11e9-b8cb-025000000001_0
11d1fa3f182b        315798907716                     "nginx -g 'daemon of…"   About an hour ago   Up About an hour                        k8s_nginx_nginx-85fb469b96-lr982_default_28fbdeee-6639-11e9-b8cb-025000000001_0
c28032a4b068        k8s.gcr.io/pause-amd64:3.1       "/pause"                 About an hour ago   Up About an hour                        k8s_POD_redis-5b44bb8d97-wwmll_default_2907f4a3-6639-11e9-b8cb-025000000001_0
7091657acfbc        k8s.gcr.io/pause-amd64:3.1       "/pause"                 About an hour ago   Up About an hour                        k8s_POD_flaskapp-6c4fccdf99-xcmwb_default_28fbe1b1-6639-11e9-b8cb-025000000001_0
97007670c247        k8s.gcr.io/pause-amd64:3.1       "/pause"                 About an hour ago   Up About an hour                        k8s_POD_nginx-85fb469b96-lr982_default_28fbdeee-6639-11e9-b8cb-025000000001_0
...
複製代碼

!!!注意: pod並不等同於docker的容器,Pod纔是k8s操做的最小單元。簡單的說,一個Pod可能包含多個容器,從yaml文件中containers:這個關鍵字能夠看出。仔細觀察docker ps的輸出,能夠發現每一個pod除了用戶自定義的容器外,還有鏡像爲k8s.gcr.io/pause-amd64:3.1的系統容器。

最後使用 curl http://127.0.0.1:30457 訪問服務

Hello World by 10.1.0.21 from 192.168.65.3 ! 該頁面已被訪問 1 次。
複製代碼

擴容

k8s集羣下,擴容很是簡單

➜  docker2istio kubectl edit deployment/flaskapp
deployment.extensions "flaskapp" edited
複製代碼

修改其中的** replicas: 3 **。

也能夠修改k8s\flaskapp.yaml中的值,而後kubectl apply -f k8s\flaskapp.yaml

另外,若是鏡像有更新,也是採用修改flaskapp.yaml文件而後apply的方式。

kubectl get pods -o wide 檢查擴容結果, 這裏使用了 -o wide ,能夠顯示更多信息

NAME                        READY     STATUS    RESTARTS   AGE       IP          NODE
flaskapp-6c4fccdf99-9xsjl   1/1       Running   0          3m        10.1.0.23   docker-for-desktop
flaskapp-6c4fccdf99-xcmwb   1/1       Running   0          1h        10.1.0.21   docker-for-desktop
flaskapp-6c4fccdf99-zp8mk   1/1       Running   0          3m        10.1.0.24   docker-for-desktop
nginx-85fb469b96-lr982      1/1       Running   0          1h        10.1.0.19   docker-for-desktop
redis-5b44bb8d97-wwmll      1/1       Running   0          1h        10.1.0.22   docker-for-desktop
複製代碼

屢次訪問服務:

➜  docker2istio curl http://127.0.0.1:30457
Hello World by 10.1.0.21 from 192.168.65.3 ! 該頁面已被訪問 2 次。
➜  docker2istio curl http://127.0.0.1:30457
Hello World by 10.1.0.23 from 192.168.65.3 ! 該頁面已被訪問 3 次。
➜  docker2istio curl http://127.0.0.1:30457
Hello World by 10.1.0.24 from 192.168.65.3 ! 該頁面已被訪問 4 次。
➜  docker2istio curl http://127.0.0.1:30457
複製代碼

結合前面看到的flaskapp的IP,能夠比較清晰的看到請求會自動負載到不一樣的Pod。

清理

k8s下的容器清理也很是簡單, 使用 kubectl delete -f k8s:

service "flaskapp" deleted
deployment.extensions "flaskapp" deleted
configmap "nginx-config" deleted
service "nginx" deleted
deployment.extensions "nginx" deleted
service "redis" deleted
deployment.extensions "redis" deleted
複製代碼

容器編排

實際上,k8s集羣在多集羣狀況下,會自動將Pod調度到合適的節點,這就是容器編排的概念。這種能力,主要有2個方式。

節點標籤

咱們的k8s演示集羣節點狀況以下:

[tyhall51@192-168-10-21 k8s]$ kubectl get nodes
NAME            STATUS   ROLES    AGE    VERSION
192-168-10-14   Ready    <none>   13d    v1.14.0
192-168-10-18   Ready    <none>   130d   v1.14.0
192-168-10-21   Ready    master   131d   v1.14.0
複製代碼

部署示例應用到k8s演示集羣:

[tyhall51@192-168-10-21 docker2istio]$ kubectl apply -f k8s -n docker2istio
service/flaskapp created
deployment.extensions/flaskapp created
configmap/nginx-config created
service/nginx created
deployment.extensions/nginx created
service/redis created
deployment.extensions/redis created
複製代碼

!!!注意 爲了避免和別的服務發生名稱衝突,這裏部署時候使用了 -n docker2istio 參數,建立了一個獨立的名稱空間。名稱空間可使用 kubectl create namespace docker2istio 命令建立。

查看名稱空間下的服務:

[tyhall51@192-168-10-21 docker2istio]$ kubectl get service -n docker2istio
NAME       TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)        AGE
flaskapp   ClusterIP   10.101.127.107   <none>        5000/TCP       47s
nginx      NodePort    10.103.147.187   <none>        80:30387/TCP   46s
redis      ClusterIP   10.106.162.13    <none>        6379/TCP       46s
複製代碼

查看名稱空間下的pod:

[tyhall51@192-168-10-21 docker2istio]$ kubectl get pods -o wide -n docker2istio
NAME                        READY   STATUS    RESTARTS   AGE   IP             NODE            NOMINATED NODE   READINESS GATES
flaskapp-589c4cdf86-sftr9   1/1     Running   0          81s   10.244.2.30    192-168-10-14   <none>           <none>
nginx-55b87f44ff-b4x88      1/1     Running   0          81s   10.244.2.31    192-168-10-14   <none>           <none>
redis-7fc7fc64fb-2nzjq      1/1     Running   0          81s   10.244.1.195   192-168-10-18   <none>           <none>
複製代碼

參考前文,修改副本數量參數replicas,對flaskapp進行擴容:

[tyhall51@192-168-10-21 docker2istio]$ kubectl get pods -o wide -n docker2istio
NAME                        READY   STATUS    RESTARTS   AGE     IP             NODE            NOMINATED NODE   READINESS GATES
flaskapp-589c4cdf86-8jzwx   1/1     Running   0          4s      10.244.1.197   192-168-10-18   <none>           <none>
flaskapp-589c4cdf86-sftr9   1/1     Running   0          3m10s   10.244.2.30    192-168-10-14   <none>           <none>
flaskapp-589c4cdf86-tz98x   1/1     Running   0          4s      10.244.1.196   192-168-10-18   <none>           <none>
nginx-55b87f44ff-b4x88      1/1     Running   0          3m10s   10.244.2.31    192-168-10-14   <none>           <none>
redis-7fc7fc64fb-2nzjq      1/1     Running   0          3m10s   10.244.1.195   192-168-10-18   <none>           <none>
複製代碼

這裏就能夠看到,擴容完成後,flaskapp的3個pod會自動調度到192-168-10-18192-168-10-182個業務節點。

192-168-10-14節點的磁盤使用的是高速ssd,io性能會更好一些,咱們但願redis可以調度到該節點。

首先,給192-168-10-14節點打上storage=ssd的標籤:

[tyhall51@192-168-10-21 docker2istio]$ kubectl label nodes 192-168-10-14 storage=ssd
node/192-168-10-14 labeled
複製代碼

檢查標籤是否正常標記:

[tyhall51@192-168-10-21 docker2istio]$ kubectl get nodes --show-labels | grep ssd
192-168-10-14   Ready    <none>   13d    v1.14.0   beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,kubernetes.io/arch=amd64,kubernetes.io/hostname=192-168-10-14,kubernetes.io/os=linux,storage=ssd
複製代碼

而後修改k8s/redis.yaml,增長nodeSelector數值,其值爲storage: ssd, 修改完成的deployment以下:

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: redis
spec:
  replicas: 1
  template:
    metadata:
      labels:
        name: redis
    spec:
      containers:
      - image: redis:4-alpine3.8
        name: redis
        ports:
        - containerPort: 6379
      nodeSelector:
        storage: ssd
複製代碼

使用kubectl apply -f k8s/redis.yaml -n docker2istio應用修改。查看docker2istio的pod分佈狀況:

[tyhall51@192-168-10-21 docker2istio]$ kubectl get pods -o wide -n docker2istio
NAME                        READY   STATUS    RESTARTS   AGE   IP             NODE            NOMINATED NODE   READINESS GATES
flaskapp-589c4cdf86-8jzwx   1/1     Running   0          11m   10.244.1.197   192-168-10-18   <none>           <none>
flaskapp-589c4cdf86-sftr9   1/1     Running   0          14m   10.244.2.30    192-168-10-14   <none>           <none>
flaskapp-589c4cdf86-tz98x   1/1     Running   0          11m   10.244.1.196   192-168-10-18   <none>           <none>
nginx-55b87f44ff-b4x88      1/1     Running   0          14m   10.244.2.31    192-168-10-14   <none>           <none>
redis-66f66896b6-7666t      1/1     Running   0          4s    10.244.2.35    192-168-10-14   <none>           <none>
複製代碼

可見redis節點從新被調度到192-168-10-14節點,表現出了節點標籤的親和力。

節點污點

在k8s演示集羣中192-168-10-21是master節點,默認不會調度業務pod,這種能力是採用節點污點實現的。 取消192-168-10-21調度污點:

kubectl taint node 192-168-10-21 node-role.kubernetes.io/master:NoSchedule-
複製代碼

而後擴容flaskapp的副本數到6個,觀察pod分佈狀況:

[tyhall51@192-168-10-21 docker2istio]$ kubectl get pods -o wide -n docker2istio
NAME                        READY   STATUS         RESTARTS   AGE     IP             NODE            NOMINATED NODE   READINESS GATES
flaskapp-589c4cdf86-8jzwx   1/1     Running        0          20m     10.244.1.197   192-168-10-18   <none>           <none>
flaskapp-589c4cdf86-92rm5   1/1     Running        0          5s      10.244.2.36    192-168-10-14   <none>           <none>
flaskapp-589c4cdf86-bfhs8   1/1     Running        0          5s      10.244.0.26    192-168-10-21   <none>           <none>
flaskapp-589c4cdf86-sftr9   1/1     Running        0          23m     10.244.2.30    192-168-10-14   <none>           <none>
flaskapp-589c4cdf86-srv25   1/1     Running        0          5s      10.244.0.25    192-168-10-21   <none>           <none>
flaskapp-589c4cdf86-tz98x   1/1     Running        0          20m     10.244.1.196   192-168-10-18   <none>           <none>
nginx-55b87f44ff-b4x88      1/1     Running        0          23m     10.244.2.31    192-168-10-14   <none>           <none>
redis-66f66896b6-7666t      1/1     Running        0          9m30s   10.244.2.35    192-168-10-14   <none>           <none>
複製代碼

這裏能夠看到,有2個pod被調到到192-168-10-21節點了。

從新設置污點:

[tyhall51@192-168-10-21 docker2istio]$ kubectl taint node 192-168-10-21 node-role.kubernetes.io/master=:NoSchedule
node/192-168-10-21 tainted
複製代碼

刪除在192-168-10-21上的2個pod:

kubectl delete pod/flaskapp-589c4cdf86-bfhs8 -n docker2istio
kubectl delete pod/flaskapp-589c4cdf86-srv25 -n docker2istio
複製代碼

觀察pod分佈狀況:

[tyhall51@192-168-10-21 docker2istio]$ kubectl get pods -o wide -n docker2istio
NAME                        READY   STATUS    RESTARTS   AGE     IP             NODE            NOMINATED NODE   READINESS GATES
flaskapp-589c4cdf86-8jzwx   1/1     Running   0          25m     10.244.1.197   192-168-10-18   <none>           <none>
flaskapp-589c4cdf86-92rm5   1/1     Running   0          4m40s   10.244.2.36    192-168-10-14   <none>           <none>
flaskapp-589c4cdf86-fp5w4   1/1     Running   0          73s     10.244.2.37    192-168-10-14   <none>           <none>
flaskapp-589c4cdf86-lv2ch   1/1     Running   0          73s     10.244.1.199   192-168-10-18   <none>           <none>
flaskapp-589c4cdf86-p9kb6   1/1     Running   0          7s      10.244.2.38    192-168-10-14   <none>           <none>
flaskapp-589c4cdf86-sftr9   1/1     Running   0          28m     10.244.2.30    192-168-10-14   <none>           <none>
nginx-55b87f44ff-b4x88      1/1     Running   0          28m     10.244.2.31    192-168-10-14   <none>           <none>
redis-66f66896b6-7666t      1/1     Running   0          14m     10.244.2.35    192-168-10-14   <none>           <none>
複製代碼

能夠看到刪除後的pod,在192-168-10-18和192-168-10-14這2個業務節點上重建了。

總結

k8s相對於compose:

  1. 管理規模擴大,由單機到集羣。
  2. 擴容更方便了,能夠無縫擴容。
  3. 部署策略更完善,能夠對容器進行編排

相關組件

  1. Etcd

etcd 是一個分佈式鍵值對存儲,設計用來可靠而快速的保存關鍵數據並提供訪問。經過分佈式鎖,leader選舉和寫屏障(write barriers)來實現可靠的分佈式協做。etcd集羣是爲高可用,持久性數據存儲和檢索而準備。k8s中使用etcd做爲集羣信息存儲。

  1. Efk

EFK (Elasticsearch + Fluentd + Kibana) 是kubernetes官方推薦的日誌收集方案

  1. helm

Helm helps you manage Kubernetes applications — Helm Charts help you define, install, and upgrade even the most complex Kubernetes application.

  1. Rock

File, Block, and Object Storage Services for your Cloud-Native Environments

...

...

相關文章
相關標籤/搜索