Kubernetes学习之路（五）之Flannel网络二进制部署和测试

一、K8S的ip地址

Node IP：节点设备的IP，如物理机，虚拟机等容器宿主的实际IP。

Pod IP：Pod的IP地址，是根据docker0网络IP段进行分配的。

Cluster IP：Service的IP，是一个虚拟IP，仅作用于service对象，由K8S管理和分配，需要结合service port才能使用，单独的IP没有通信功能，集群外访问需要一些修改。

在K8S集群内部，node ip、pod ip、clustere ip的通信机制是由k8s指定的路由规则，不是IP路由。

[root@linux-node1 ~]# kubectl get service
NAME         TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)   AGE
kubernetes   ClusterIP   10.1.0.1     <none>        /TCP   3h

二、Flannel网络部署

（1）为Flannel生成证书

[root@linux-node1 ssl]# vim flanneld-csr.json
{
  "CN": "flanneld",
  "hosts": [],
  "key": {
    "algo": "rsa",
    "size":
  },
  "names": [
    {
      "C": "CN",
      "ST": "BeiJing",
      "L": "BeiJing",
      "O": "k8s",
      "OU": "System"
    }
  ]
}

（2）生成证书

[root@linux-node1 ssl]# cfssl gencert -ca=/opt/kubernetes/ssl/ca.pem \
>    -ca-key=/opt/kubernetes/ssl/ca-key.pem \
>    -config=/opt/kubernetes/ssl/ca-config.json \
>    -profile=kubernetes flanneld-csr.json | cfssljson -bare flanneld
[root@linux-node1 ssl]# ll flannel*
-rw-r--r--  root root   May  : flanneld.csr
-rw-r--r--  root root   May  : flanneld-csr.json
-rw-------  root root  May  : flanneld-key.pem
-rw-r--r--  root root  May  : flanneld.pem

（3）分发证书

[root@linux-node1 ssl]# cp flanneld*.pem /opt/kubernetes/ssl/
[root@linux-node1 ssl]# scp flanneld*.pem 192.168.56.120:/opt/kubernetes/ssl/
flanneld-key.pem                                           %    .2KB/s   :
flanneld.pem                                               %    .3KB/s   :
[root@linux-node1 ssl]# scp flanneld*.pem 192.168.56.130:/opt/kubernetes/ssl/
flanneld-key.pem                                           %    .1KB/s   :
flanneld.pem                                               %     .4KB/s   :    

（4）下载Flannel软件包

[root@linux-node1 ~]# cd /usr/local/src
# wget
 https://github.com/coreos/flannel/releases/download/v0.10.0/flannel-v0.10.0-linux-amd64.tar.gz
[root@linux-node1 src]# tar zxf flannel-v0.10.0-linux-amd64.tar.gz
[root@linux-node1 src]# cp flanneld mk-docker-opts.sh /opt/kubernetes/bin/
复制到linux-node2和linux-node3节点
[root@linux-node1 src]# scp flanneld mk-docker-opts.sh 192.168.56.120:/opt/kubernetes/bin/
[root@linux-node1 src]# scp flanneld mk-docker-opts.sh 192.168.56.130:/opt/kubernetes/bin/

复制对应脚本到/opt/kubernetes/bin目录下。
[root@linux-node1 ~]# cd /usr/local/src/kubernetes/cluster/centos/node/bin/
[root@linux-node1 bin]# cp remove-docker0.sh /opt/kubernetes/bin/
[root@linux-node1 bin]# scp remove-docker0.sh 192.168.56.120:/opt/kubernetes/bin/
[root@linux-node1 bin]# scp remove-docker0.sh 192.168.56.130:/opt/kubernetes/bin/

（5）配置Flannel

[root@linux-node1 ~]# vim /opt/kubernetes/cfg/flannel
FLANNEL_ETCD="-etcd-endpoints=https://192.168.56.110:2379,https://192.168.56.120:2379,https://192.168.56.130:2379"
FLANNEL_ETCD_KEY="-etcd-prefix=/kubernetes/network"
FLANNEL_ETCD_CAFILE="--etcd-cafile=/opt/kubernetes/ssl/ca.pem"
FLANNEL_ETCD_CERTFILE="--etcd-certfile=/opt/kubernetes/ssl/flanneld.pem"
FLANNEL_ETCD_KEYFILE="--etcd-keyfile=/opt/kubernetes/ssl/flanneld-key.pem"
复制配置到其它节点上
[root@linux-node1 ~]# scp /opt/kubernetes/cfg/flannel 192.168.56.120:/opt/kubernetes/cfg/
[root@linux-node1 ~]# scp /opt/kubernetes/cfg/flannel 192.168.56.130:/opt/kubernetes/cfg/

（6）设置Flannel系统服务

[root@linux-node1 ~]# vim /usr/lib/systemd/system/flannel.service
[Unit]
Description=Flanneld overlay address etcd agent
After=network.target
Before=docker.service

[Service]
EnvironmentFile=-/opt/kubernetes/cfg/flannel
ExecStartPre=/opt/kubernetes/bin/remove-docker0.sh
ExecStart=/opt/kubernetes/bin/flanneld ${FLANNEL_ETCD} ${FLANNEL_ETCD_KEY} ${FLANNEL_ETCD_CAFILE} ${FLANNEL_ETCD_CERTFILE} ${FLANNEL_ETCD_KEYFILE}
ExecStartPost=/opt/kubernetes/bin/mk-docker-opts.sh -d /run/flannel/docker

Type=notify

[Install]
WantedBy=multi-user.target
RequiredBy=docker.service
复制系统服务脚本到其它节点上
# scp /usr/lib/systemd/system/flannel.service 192.168.56.120:/usr/lib/systemd/system/
# scp /usr/lib/systemd/system/flannel.service 192.168.56.130:/usr/lib/systemd/system/

三、Flannel CNI集成

（1）下载CNI插件

https://github.com/containernetworking/plugins/releases
wget https://github.com/containernetworking/plugins/releases/download/v0.7.1/cni-plugins-amd64-v0.7.1.tgz
[root@linux-node1 ~]# mkdir /opt/kubernetes/bin/cni
[root@linux-node2 ~]# mkdir /opt/kubernetes/bin/cni
[root@linux-node3 ~]# mkdir /opt/kubernetes/bin/cni
[root@linux-node1 src]# tar zxf cni-plugins-amd64-v0.7.1.tgz -C /opt/kubernetes/bin/cni
[root@linux-node1 src]# scp -r /opt/kubernetes/bin/cni/* 192.168.56.120:/opt/kubernetes/bin/cni/
[root@linux-node1 src]# scp -r /opt/kubernetes/bin/cni/* 192.168.56.130:/opt/kubernetes/bin/cni/

（2）创建Etcd的key

此步的操作是为了创建POD的网段，并在ETCD中存储，而后FLANNEL从ETCD中取出并进行分配

[root@linux-node1 src]# /opt/kubernetes/bin/etcdctl --ca-file /opt/kubernetes/ssl/ca.pem --cert-file /opt/kubernetes/ssl/flanneld.pem --key-file /opt/kubernetes/ssl/flanneld-key.pem \
      --no-sync -C https://192.168.56.110:2379,https://192.168.56.120:2379,https://192.168.56.130:2379 \
mk /kubernetes/network/config '{ "Network": "10.2.0.0/16", "Backend": { "Type": "vxlan", "VNI": 1 }}' >/dev/null >&

（3）启动flannel

[root@linux-node1 ~]# systemctl daemon-reload
[root@linux-node1 ~]# systemctl enable flannel
[root@linux-node1 ~]# chmod +x /opt/kubernetes/bin/*
[root@linux-node1 ~]# systemctl start flannel

[root@linux-node2 ~]# systemctl daemon-reload
[root@linux-node2 ~]# systemctl enable flannel
[root@linux-node2 ~]# chmod +x /opt/kubernetes/bin/*
[root@linux-node2 ~]# systemctl start flannel

[root@linux-node3 ~]# systemctl daemon-reload
[root@linux-node3 ~]# systemctl enable flannel
[root@linux-node3 ~]# chmod +x /opt/kubernetes/bin/*
[root@linux-node3 ~]# systemctl start flannel

可以看到每个节点上会多出一个flannel.1的网卡，不同的节点都在不同网段。

[root@linux-node1 ~]# ifconfig flannel.
flannel.: flags=<UP,BROADCAST,RUNNING,MULTICAST>  mtu
        inet 10.2.46.0  netmask 255.255.255.255  broadcast 0.0.0.0
        inet6 fe80::f4e6:1aff:fe7e:575b  prefixlen   scopeid 0x20<link>
        ether f6:e6:1a:7e::5b  txqueuelen   (Ethernet)
        RX packets   bytes  (0.0 B)
        RX errors   dropped   overruns   frame
        TX packets   bytes  (0.0 B)
        TX errors   dropped  overruns   carrier   collisions 

[root@linux-node2 ~]# ifconfig flannel.
flannel.: flags=<UP,BROADCAST,RUNNING,MULTICAST>  mtu
        inet 10.2.87.0  netmask 255.255.255.255  broadcast 0.0.0.0
        inet6 fe80::d4e5:72ff:fe3e:  prefixlen   scopeid 0x20<link>
        ether d6:e5::3e::  txqueuelen   (Ethernet)
        RX packets   bytes  (0.0 B)
        RX errors   dropped   overruns   frame
        TX packets   bytes  (0.0 B)
        TX errors   dropped  overruns   carrier   collisions 

[root@linux-node3 ~]# ifconfig flannel.
flannel.: flags=<UP,BROADCAST,RUNNING,MULTICAST>  mtu
        inet 10.2.33.0  netmask 255.255.255.255  broadcast 0.0.0.0
        ether be:cd:5a:4f:6b:d1  txqueuelen   (Ethernet)
        RX packets   bytes  (0.0 B)
        RX errors   dropped   overruns   frame
        TX packets   bytes  (0.0 B)
        TX errors   dropped  overruns   carrier   collisions 

（4）遇到的问题：Flannel无法启动

检查/opt/kubernetes/cfg/etcd.conf配置文件中的ETCD_LISTEN_CLIENT_URLS是否配置监听127.0.0.1:2379。依旧无法启动flannel，重新输入了一遍，正常了，暂时没发现其他原因，至于etcdctl无法获取key值，有待研究!!!

四、配置Docker使用Flannel

[root@linux-node1 ~]# vim /usr/lib/systemd/system/docker.service
[Unit] #在Unit下面修改After和增加Requires
After=network-online.target firewalld.service flannel.service  #让docker在flannel网络后面启动
Wants=network-online.target
Requires=flannel.service 

[Service] #增加EnvironmentFile=-/run/flannel/docker
Type=notify
EnvironmentFile=-/run/flannel/docker #加载环境文件，设置docker0的ip地址为flannel分配的ip地址
ExecStart=/usr/bin/dockerd $DOCKER_OPTS
[root@linux-node1 ~]# systemctl daemon-reload
[root@linux-node1 ~]# systemctl restart docker
[root@linux-node1 ~]# ifconfig docker0
docker0: flags=<UP,BROADCAST,MULTICAST>  mtu
        inet 10.2.46.1  netmask 255.255.255.0  broadcast 0.0.0.0
        ether ::1f:ef:9f:b5  txqueuelen   (Ethernet)
        RX packets   bytes  (0.0 B)
        RX errors   dropped   overruns   frame
        TX packets   bytes  (0.0 B)
        TX errors   dropped  overruns   carrier   collisions 

[root@linux-node2 ~]# ifconfig docker0
docker0: flags=<UP,BROADCAST,MULTICAST>  mtu
        inet 10.2.87.1  netmask 255.255.255.0  broadcast 0.0.0.0
        ether ::8a:a5::d7  txqueuelen   (Ethernet)
        RX packets   bytes  (0.0 B)
        RX errors   dropped   overruns   frame
        TX packets   bytes  (0.0 B)
        TX errors   dropped  overruns   carrier   collisions 

[root@linux-node3 ~]# ifconfig docker0
docker0: flags=<UP,BROADCAST,MULTICAST>  mtu
        inet 10.2.33.1  netmask 255.255.255.0  broadcast 0.0.0.0
        ether :::::  txqueuelen   (Ethernet)
        RX packets   bytes  (0.0 B)
        RX errors   dropped   overruns   frame
        TX packets   bytes  (0.0 B)
        TX errors   dropped  overruns   carrier   collisions 

总结

在kubectl get node时，会看到节点的状态READY，如果状态为NotReady，可以查看节点上的kubelet是否已经启动，如果未启动，进行启动。kubelet无法启动，要进行查看systemctl status kubelet或journalctl -xe看看是什么原因导致无法启动。遇到的一种情况是依赖docker，查看docker无法启动。再进一步排查docker无法启动的原因。