kubernetes是google公司基于docker所做的一个分布式集群,有以下主件组成
etcd: 高可用存储共享配置和服务发现,作为与minion机器上的flannel配套使用,作用是使每台 minion上运行的docker拥有不同的ip段,最终目的是使不同minion上正在运行的docker containner都有一个与别的任意一个containner(别的minion上运行的docker containner)不一样的IP地址。
flannel: 网络结构支持
kube-apiserver: 不论通过kubectl还是使用remote api 直接控制,都要经过apiserver
kube-controller-manager: 对replication controller, endpoints controller, namespace controller, and serviceaccounts controller的循环控制,与kube-apiserver交互,保证这些controller工作
kube-scheduler: Kubernetes scheduler的作用就是根据特定的调度算法将pod调度到指定的工作节点(minion)上,这一过程也叫绑定(bind)
kubelet: Kubelet运行在Kubernetes Minion Node上. 它是container agent的逻辑继任者
kube-proxy: kube-proxy是kubernetes 里运行在minion节点上的一个组件, 它起的作用是一个服务代理的角色
图为GIT+Jenkins+Kubernetes+Docker+Etcd+confd+Nginx+Glusterfs架构:
如下:
环境:
centos7系统机器三台:
192.168.1.165: 用来安装kubernetes master
192.168.1.247: 用作kubernetes minion (minion1)
一、关闭系统运行的防火墙及selinux
1。如果系统开启了防火墙则按如下步骤关闭防火墙(所有机器)
# systemctl stop firewalld # systemctl disable firewalld
2.关闭selinux
1 2 |
#setenforce 0
|
二、MASTER安装配置
1. 安装并配置Kubernetes master(yum 方式)
1 |
# yum -y install etcd kubernetes
|
配置etcd。确保列出的这些项都配置正确并且没有被注释掉,下面的配置都是如此
1 2 3 4 5 6 |
#vim /etc/etcd/etcd.conf
ETCD_LISTEN_CLIENT_URLS="http://0.0.0.0:2379" ETCD_NAME="default" ETCD_ADVERTISE_CLIENT_URLS="http://0.0.0.0:2379"
|
配置kubernetes
vim /etc/kubernetes/apiserver
KUBE_API_ADDRESS="--address=0.0.0.0"
KUBE_API_PORT="--port=8080"
KUBELET_PORT="--kubelet_port=10250"
KUBE_ETCD_SERVERS="--etcd_servers=http://192.168.1.165:2379"
KUBE_SERVICE_ADDRESSES="--service-cluster-ip-range=10.254.0.0/16"
KUBE_ADMISSION_CONTROL="--admission_control=NamespaceLifecycle,NamespaceExists,LimitRanger,SecurityContextDeny,ResourceQuota"
KUBE_API_ARGS=""
2. 启动etcd, kube-apiserver, kube-controller-manager and kube-scheduler服务
1 |
# for SERVICES in etcd kube-apiserver kube-controller-manager kube-scheduler; do systemctl restart $SERVICES ;systemctl enable $SERVICES ;systemctl status $SERVICES ; done;
|
3.设置etcd网络
1 |
报错如下: parse 192.168.1.165:2379: first path segment in URL cannot contain colon 解决方法: etcdctl -C http://127.0.0.1:2379 set /atomic.io/network/config '{"Network":"10.1.0.0/16"}' {"Network":"10.1.0.0/16"}
|
4. 至此master配置完成,运行kubectl get nodes可以查看有多少minion在运行,以及其状态。这里我们的minion还都没有开始安装配置,所以运行之后结果为空
1 |
# kubectl get nodes
|
三、MINION安装配置(每台minion机器都按如下安装配置)
1. 环境安装和配置
1 |
# yum -y install flannel kubernetes
|
配置kubernetes连接的服务端IP
1 2 3 |
#vim /etc/kubernetes/config 修改以下两个配置项
KUBE_MASTER="--master=http://192.168.1.165:8080" KUBE_ETCD_SERVERS="--etcd_servers=http://192.168.1.165:2379" |
配置kubernetes ,(请使用每台minion自己的IP地址比如192.168.1.247:代替下面的$LOCALIP)
1 2 3 4 5 |
#vim /etc/kubernetes/kubelet 红色标记的要调整,具体看后面的步骤,否则创建pod会失败。
KUBELET_ADDRESS="--address=0.0.0.0" KUBELET_HOSTNAME="--hostname-override=192.168.1.247" KUBELET_API_SERVER="--api-servers=http://192.168.1.165:8080" KUBELET_POD_INFRA_CONTAINER="--pod-infra-container-image=registry.access.redhat.com/rhel7/pod-infrastructure:latest" KUBELET_ARGS="" |
2. 准备启动服务(如果本来机器上已经运行过docker的请看过来,没有运行过的请忽略此步骤)
运行ifconfig,查看机器的网络配置情况(有docker0)
1 2 3 4 5 |
# ifconfig docker0
|
warning:在运行过docker的机器上可以看到有docker0,这里在启动服务之前需要删掉docker0配置,在命令行运行:sudo ip link delete docker0
3.配置flannel网络
1 2 3 |
#vim /etc/sysconfig/flanneld
FLANNEL_ETCD_ENDPOINTS="http://192.168.1.165:2379" FLANNEL_ETCD_PREFIX="/atomic.io/network" |
PS:其中atomic.io与上面etcd中的Network对应
4. 启动服务
1 |
# for SERVICES in flanneld kube-proxy kubelet docker; do systemctl restart $SERVICES ;systemctl enable $SERVICES ;systemctl status $SERVICES; done
|
四、配置完成验证安装
确定minion(192.168.1.247)和一台master(192.168.1.165)都已经成功的安装配置并且服务都已经启动了。
切换到master机器上,运行命令kubectl get nodes
1 2 3 4 |
# kubectl get nodes
NAME STATUS AGE 192.168.1.247 Ready 6h
|
可以看到配置的minion已经在master的node列表中了。如果想要更多的node,只需要按照minion的配置,配置更多的机器就可以了。
三、Kubernetes之深入了解Pod
2、Pod基本用法:
在使用docker时,我们可以使用docker run命令创建并启动一个容器,而在Kubernetes系统中对长时间运行的容器要求是:其主程序需要一直在前台运行。如果我们创建的docker镜像的启动命令是后台执行程序,例如Linux脚本:
nohup ./startup.sh &
则kubelet创建包含这个容器的pod后运行完该命令,即认为Pod执行结束,之后根据RC中定义的pod的replicas副本数量生产一个新的pod,而一旦创建出新的pod,将在执行完命令后陷入无限循环的过程中,这就是Kubernetes需要我们创建的docker镜像以一个前台命令作为启动命令的原因。
对于无法改造为前台执行的应用,也可以使用开源工具supervisor辅助进行前台运行的功能。
****Pod可以由一个或多个容器组合而成
例如:两个容器应用的前端frontend和redis为紧耦合的关系,应该组合成一个整体对外提供服务,则应该将这两个打包为一个pod.
配置文件frontend-localredis-pod.yaml如下,刚开始的时候没有直接复制,yaml对格式严格要求,所以可以通过kubectl create -f frontend-localredis-pod.yaml --dry-run --validate=true检查配置文件哪里有问题。
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 |
apiVersion: v1 kind: Pod metadata: name: redis-php labels: name: redis-php spec: containers: - name: frontend image: kubeguide/guestbook-php-frontend:localredis ports: - containerPort: 80 - name: redis-php image: kubeguide/redis-master ports: - containerPort: 637
|
kubernetes启动pod,pod状态一直不正常,查看pod状态显示
image pull failed for registry.access.redhat.com/rhel7/pod-infrastructure:latest, this may be because there are no credentials on this request. details: (open /etc/docker/certs.d/registry.access.redhat.com/redhat-ca.crt: no such file or directory)
按照网上所说的解决方案:yum install *rhsm* -y
Failed to create pod infra container: ImagePullBackOff; Skipping pod "redis-master-jj6jw_default(fec25a87-cdbe-11e7-ba32-525400cae48b)": Back-off pulling image "registry.access.redhat.com/rhel7/pod-infrastructure:latest
解决方法:试试通过手动下载
docker pull registry.access.redhat.com/rhel7/pod-infrastructure:latest
docker pull 是还是报错
open /etc/docker/certs.d/registry.access.redhat.com/redhat-ca.crt: no such file or directory
查看下redhat-ca.crt确实不存在,registry.access.redhat.com/rhel7/pod-infrastructure:latest默认是https下载。
最终解决方案:
1.docker search pod-infrastructure
docker search pod-infrastructure
INDEX NAME DESCRIPTION STARS OFFICIAL AUTOMATED
docker.io docker.io/openshift/origin-pod The pod infrastructure image for OpenShift 3 8
docker.io docker.io/davinkevin/podcast-server Container around the Podcast-Server Applic... 5
docker.io docker.io/infrastructureascode/aws-cli Containerized AWS CLI on alpine to avoid r... 4 [OK]
docker.io docker.io/newrelic/infrastructure Public image for New Relic Infrastructure. 4
docker.io docker.io/infrastructureascode/uwsgi uWSGI application server 2 [OK]
docker.io docker.io/infrastructureascode/serf A tiny Docker image with HashiCorp Serf us... 1 [OK]
docker.io docker.io/mosquitood/k8s-rhel7-pod-infrastructure 1
docker.io docker.io/podigg/podigg-lc-hobbit A HOBBIT dataset generator wrapper for PoDiGG 1 [OK]
docker.io docker.io/stefanprodan/podinfo Kubernetes multi-arch pod info 1
docker.io docker.io/tianyebj/pod-infrastructure registry.access.redhat.com/rhel7/pod-infra... 1
docker.io docker.io/w564791/pod-infrastructure latest 1
docker.io docker.io/infrastructureascode/hello-world A tiny "Hello World" web server with a hea... 0 [OK]
docker.io docker.io/jqka/pod-infrastructure redhat pod 0 [OK]
docker.io docker.io/ocpqe/hello-pod Copy form docker.io/deshuai/hello-pod:latest 0
docker.io docker.io/oudi/pod-infrastructure pod-infrastructure 0 [OK]
docker.io docker.io/sebastianhutter/podcaster python script to download podcasts https:/... 0 [OK]
docker.io docker.io/shadowalker911/pod-infrastructure 0
docker.io docker.io/statemood/pod-infrastructure Automated build from registry.access.redha... 0 [OK]
docker.io docker.io/tfgco/podium Podium is a blazing-fast player ranking se... 0
docker.io docker.io/trancong/pod2consul register pod with consul 0
docker.io docker.io/tundradotcom/podyn dockerized Podyn 0
docker.io docker.io/vistalba/podget Podget Docker with rename included. 0 [OK]
docker.io docker.io/wedeploy/infrastructure 0
docker.io docker.io/xplenty/rhel7-pod-infrastructure registry.access.redhat.com/rhel7/pod-infra... 0
docker.io docker.io/zengshaoyong/pod-infrastructure pod-infrastructure 0 [OK]
2.vi /etc/kubernetes/kubelet
替换为上面第一个节点(Deven:后面改回去了也没有报错了,所以这个步骤可能不是必须的。) KUBELET_POD_INFRA_CONTAINER="--pod-infra-container-image=docker.io/openshift/origin-pod"
3.重启
systemctl restart kube-apiserver
systemctl restart kube-controller-manager
systemctl restart kube-scheduler
systemctl restart kubelet
systemctl restart kube-proxy
4. kubectl get pods 查看之前的pods已经正常状态Running
属于一个Pod的多个容器应用之间相互访问只需要通过localhost就可以通信,这一组容器被绑定在一个环境中。
使用kubectl create创建该Pod后,get pod信息可以看到如下图:
1 2 3 |
#kubectl get pods NAME READY STATUS RESTARTS AGE myweb-1rr24 1/1 Running 1 27m redis-php 2/2 Running 0 16s
|
查看pod的详细信息,可以看到两个容器的定义和创建过程。
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 |
#kubectl describe pods redis-php Name: redis-php Namespace: default Node: 192.168.1.247/192.168.1.247 Start Time: Sat, 04 Aug 2018 17:37:53 +0800 Labels: name=redis-php Status: Running IP: 10.1.49.3 Controllers: <none> Containers: frontend: Container ID: docker://1c109acce5c81f57f7c02619c489855ae67ece114fdfa104189521e1f2fc052b Image: kubeguide/guestbook-php-frontend:localredis Image ID: docker-pullable://docker.io/kubeguide/[email protected]:37c2c1dcfcf0a51bf9531430fe057bcb1d4b94c64048be40ff091f01e384f81e Port: 80/TCP State: Running Started: Sat, 04 Aug 2018 17:37:54 +0800 Ready: True Restart Count: 0 Volume Mounts: <none> Environment Variables: <none> redis-php: Container ID: docker://6edf7724a548f178975eb9abcbee675788720a5d804867124f1dc454e7e3b058 Image: kubeguide/redis-master Image ID: docker-pullable://docker.io/kubeguide/[email protected]:e11eae36476b02a195693689f88a325b30540f5c15adbf531caaecceb65f5b4d Port: 6379/TCP State: Running Started: Sat, 04 Aug 2018 17:38:00 +0800 Ready: True Restart Count: 0 Volume Mounts: <none> Environment Variables: <none> Conditions: Type Status Initialized True Ready True PodScheduled True No volumes. QoS Class: BestEffort Tolerations: <none> Events: FirstSeen LastSeen Count From SubObjectPath Type Reason Message --------- -------- ----- ---- ------------- -------- ------ ------- 2m 2m 1 {default-scheduler } Normal Scheduled Successfully assigned redis-php to 192.168.1.247 2m 2m 1 {kubelet 192.168.1.247} spec.containers{frontend} Normal Pulled Container image "kubeguide/guestbook-php-frontend:localredis" already present on machine 2m 2m 1 {kubelet 192.168.1.247} spec.containers{frontend} Normal Created Created container with docker id 1c109acce5c8; Security:[seccomp=unconfined] 2m 2m 1 {kubelet 192.168.1.247} spec.containers{frontend} Normal Started Started container with docker id 1c109acce5c8 2m 2m 1 {kubelet 192.168.1.247} spec.containers{redis-php} Normal Pulling pulling image "kubeguide/redis-master" 2m 2m 3 {kubelet 192.168.1.247} Warning MissingClusterDNS kubelet does not have ClusterDNS IP configured and cannot create Pod using "ClusterFirst" policy. Falling back to DNSDefault policy. 2m 2m 1 {kubelet 192.168.1.247} spec.containers{redis-php} Normal Pulled Successfully pulled image "kubeguide/redis-master" 2m 2m 1 {kubelet 192.168.1.247} spec.containers{redis-php} Normal Created Created container with docker id 6edf7724a548; Security:[seccomp=unconfined] 2m 2m 1 {kubelet 192.168.1.247} spec.containers{redis-php} Normal Started Started container with docker id 6edf7724a548 |
原文地址:http://blog.51cto.com/wwdhks/2154711