目前docker主要应用于单机环境，使用网桥模式，但如果想把多台主机网络互相，让多台主机内部的container互相通信，就得使用其他的软件来帮忙，可以使用Weave、Kubernetes、Flannel、SocketPlane或者openvswitch等，我这里就使用openvswitch来介绍docker多台主机网络互通。
先看一个使用openvswitch连接的架构图，连接的方式是vxlan
说明：
这里有2台主机，分别是NODEA与NODEB，系统是centos7，内核是3.18（默认centos7内核是3.10，但想使用vxlan，所以得升级，参考http://dl528888.blog.51cto.com/2382721/1609850）
docker是1.3.2版本，存储引擎是devicemapper。
每台主机里都有2个网桥ovs1与ovs2，ovs1是管理网络，连接内网网卡em1，ovs2是数据网络，docker测试机都连接这个ovs2，并且container创建的时候网络都是none，使用pipework指定固定ip。
然后2台主机使用vxlan连接网络。
重要：
我个人认为使用这个模式并且指定固定ip，适用于的环境主要是给研发或者个人的测试模式，如果是集群环境，没必要指定固定ip（我这里的集群就没有使用固定ip，使用动态ip，效果很好，后续给大家介绍集群）。
下面是部署方法
环境
一、安装openvswitch
我的版本是最新的2.3.1
1、安装基础环境
yum install gcc make python-devel openssl-devel kernel-devel graphviz \ kernel-debug-devel autoconf automake rpm-build redhat-rpm-config \ libtool
2、下载最新的包
wget http://openvswitch.org/releases/openvswitch-2.3.1.tar.gz
3、解压与打包
tar zxvf openvswitch-2.3.1.tar.gz mkdir -p ~/rpmbuild/SOURCES cp openvswitch-2.3.1.tar.gz ~/rpmbuild/SOURCES/ sed 's/openvswitch-kmod, //g' openvswitch-2.3.1/rhel/openvswitch.spec > openvswitch-2.3.1/rhel/openvswitch_no_kmod.spec rpmbuild -bb --without check openvswitch-2.3.1/rhel/openvswitch_no_kmod.spec
之后会在~/rpmbuild/RPMS/x86_64/里有2个文件
total 9500 -rw-rw-r-- 1 ovswitch ovswitch 2013688 Jan 15 03:20 openvswitch-2.3.1-1.x86_64.rpm -rw-rw-r-- 1 ovswitch ovswitch 7712168 Jan 15 03:20 openvswitch-debuginfo-2.3.1-1.x86_64.rpm
安装第一个就行
4、安装
yum localinstall ~/rpmbuild/RPMS/x86_64/openvswitch-2.3.1-1.x86_64.rpm
5、启动
systemctl start openvswitch
6、查看状态
[[email protected] tmp]# systemctl status openvswitch openvswitch.service - LSB: Open vSwitch switch Loaded: loaded (/etc/rc.d/init.d/openvswitch) Active: active (running) since Wed 2015-01-28 23:34:01 CST; 6 days ago CGroup: /system.slice/openvswitch.service ├─20314 ovsdb-server: monitoring pid 20315 (healthy) ├─20315 ovsdb-server /etc/openvswitch/conf.db -vconsole:emer -vsyslog:err -vfile:info --remote=punix:/var/run/openvswitch/db.sock --private-key=db:Open_vSwitch,SSL,p... ├─20324 ovs-vswitchd: monitoring pid 20325 (healthy) └─20325 ovs-vswitchd unix:/var/run/openvswitch/db.sock -vconsole:emer -vsyslog:err -vfile:info --mlockall --no-chdir --log-file=/var/log/openvswitch/ovs-vswitchd.log... Jan 28 23:34:01 ip-10-10-17-3 openvswitch[20291]: /etc/openvswitch/conf.db does not exist ... (warning). Jan 28 23:34:01 ip-10-10-17-3 openvswitch[20291]: Creating empty database /etc/openvswitch/conf.db [ OK ] Jan 28 23:34:01 ip-10-10-17-3 openvswitch[20291]: Starting ovsdb-server [ OK ] Jan 28 23:34:01 ip-10-10-17-3 ovs-vsctl[20316]: ovs|00001|vsctl|INFO|Called as ovs-vsctl --no-wait -- init -- set Open_vSwitch . db-version=7.6.2 Jan 28 23:34:01 ip-10-10-17-3 ovs-vsctl[20321]: ovs|00001|vsctl|INFO|Called as ovs-vsctl --no-wait set Open_vSwitch . ovs-version=2.3.1 "external-ids:system-id=\"6ea..."unknown\"" Jan 28 23:34:01 ip-10-10-17-3 openvswitch[20291]: Configuring Open vSwitch system IDs [ OK ] Jan 28 23:34:01 ip-10-10-17-3 openvswitch[20291]: Starting ovs-vswitchd [ OK ] Jan 28 23:34:01 ip-10-10-17-3 openvswitch[20291]: Enabling remote OVSDB managers [ OK ] Jan 28 23:34:01 ip-10-10-17-3 systemd[1]: Started LSB: Open vSwitch switch. Hint: Some lines were ellipsized, use -l to show in full.
可以看到是正常运行状态
具体的安装详细步骤可以参考
https://github.com/openvswitch/ovs/blob/master/INSTALL.RHEL.md与http://www.linuxidc.com/Linux/2014-12/110272.htm
二、部署单机环境的docker
1、下载pipework
使用这个软件进行固定ip设置
cd /tmp/ git clone https://github.com/jpetazzo/pipework.git
2、在NODEA（ip是10.10.17.3）里运行下面命令
可以把下面内容复制到脚本里运行
#!/bin/bash #author: Deng Lei #email: [email protected] #删除docker测试机 docker rm `docker stop $(docker ps -a -q)` #删除已有的openvswitch交换机 ovs-vsctl list-br|xargs -I {} ovs-vsctl del-br {} #创建交换机 ovs-vsctl add-br ovs1 ovs-vsctl add-br ovs2 #把物理网卡加入ovs2 ovs-vsctl add-port ovs1 em1 ip link set ovs1 up ifconfig em1 0 ifconfig ovs1 10.10.17.3 ip link set ovs2 up ip addr add 172.16.0.3/16 dev ovs2 pipework_dir='/tmp/pipework' docker run --restart always --privileged -d --net="none" --name='test1' docker.ops-chukong.com:5000/centos6-http:new /usr/bin/supervisord $pipework_dir/pipework ovs2 test1 172.16.0.5/[email protected] docker run --restart always --privileged -d --net="none" --name='test2' docker.ops-chukong.com:5000/centos6-http:new /usr/bin/supervisord $pipework_dir/pipework ovs2 test2 172.16.0.6/[email protected]
根据自己的环境修改上面内容
运行脚本
[[email protected] tmp]# sh openvswitch_docker.sh 5a1139276ccd 03d866e20f58 6352f9ecd69450e332a13ec5dfecba106e58cf2a301e9b539a7e690cd61934f7 8e294816e5225dbee7e8442be1d9d5b4c6072d935b68e332b84196ea1db6f07c [[email protected] tmp]# docker ps -a CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES 8e294816e522 docker.ops-chukong.com:5000/centos6-http:new "/usr/bin/supervisor 3 seconds ago Up 2 seconds test2 6352f9ecd694 docker.ops-chukong.com:5000/centos6-http:new "/usr/bin/supervisor 3 seconds ago Up 3 seconds test1
可以看到已经启动了2个容器，分别是test1与test2
下面从本地登陆指定的ip试试
[[email protected] tmp]# ssh 172.16.0.5 The authenticity of host '172.16.0.5 (172.16.0.5)' can't be established. RSA key fingerprint is 39:7c:13:9f:d4:b0:d7:63:fc:ff:ae:e3:46:a4:bf:6b. Are you sure you want to continue connecting (yes/no)? yes Warning: Permanently added '172.16.0.5' (RSA) to the list of known hosts. [email protected]'s password: Last login: Mon Nov 17 14:10:39 2014 from 172.17.42.1 [email protected]:~ 18:57:50 # ifconfig eth1 Link encap:Ethernet HWaddr 26:39:B1:88:25:CC inet addr:172.16.0.5 Bcast:0.0.0.0 Mask:255.255.0.0 inet6 addr: fe80::2439:b1ff:fe88:25cc/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:65 errors:0 dropped:6 overruns:0 frame:0 TX packets:41 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:8708 (8.5 KiB) TX bytes:5992 (5.8 KiB) lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 inet6 addr: ::1/128 Scope:Host UP LOOPBACK RUNNING MTU:65536 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 b) TX bytes:0 (0.0 b) [email protected]:~ 18:57:51 # ping 172.16.0.6 -c 2 PING 172.16.0.6 (172.16.0.6) 56(84) bytes of data. 64 bytes from 172.16.0.6: icmp_seq=1 ttl=64 time=0.433 ms 64 bytes from 172.16.0.6: icmp_seq=2 ttl=64 time=0.040 ms --- 172.16.0.6 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 0.040/0.236/0.433/0.197 ms [email protected]:~ 18:58:03 # ping 172.16.0.3 -c 2 PING 172.16.0.3 (172.16.0.3) 56(84) bytes of data. 64 bytes from 172.16.0.3: icmp_seq=1 ttl=64 time=0.369 ms 64 bytes from 172.16.0.3: icmp_seq=2 ttl=64 time=0.045 ms --- 172.16.0.3 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 999ms rtt min/avg/max/mdev = 0.045/0.207/0.369/0.162 ms [email protected]:~ 18:58:09 # ping www.baidu.com -c 2 PING www.a.shifen.com (180.149.131.205) 56(84) bytes of data. 64 bytes from 180.149.131.205: icmp_seq=1 ttl=54 time=1.83 ms 64 bytes from 180.149.131.205: icmp_seq=2 ttl=54 time=1.81 ms --- www.a.shifen.com ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1003ms rtt min/avg/max/mdev = 1.816/1.827/1.839/0.044 ms
登陆后可以看到容器内的ip是指定的，并且能ping另外同一个网段的172.16.0.6，外网也能ping通。
下面进行vxlan测试，需要现在另外一个物理宿主机进行上面的脚本安装，然后在进行vxlan配置
3、在NODEB（ip是10.10.17.4）里运行
脚本内容是
#!/bin/bash #author: Deng Lei #email: [email protected] #删除docker测试机 docker rm `docker stop $(docker ps -a -q)` #删除已有的openvswitch交换机 ovs-vsctl list-br|xargs -I {} ovs-vsctl del-br {} #创建交换机 ovs-vsctl add-br ovs1 ovs-vsctl add-br ovs2 #把物理网卡加入ovs2 ovs-vsctl add-port ovs1 em1 ip link set ovs1 up ifconfig em1 0 ifconfig ovs1 10.10.17.4 ip link set ovs2 up ip addr add 172.16.0.4/16 dev ovs2 pipework_dir='/tmp/pipework' docker run --restart always --privileged -d --net="none" --name='test1' docker.ops-chukong.com:5000/centos6-http:new /usr/bin/supervisord $pipework_dir/pipework ovs2 test1 172.16.0.8/[email protected] docker run --restart always --privileged -d --net="none" --name='test2' docker.ops-chukong.com:5000/centos6-http:new /usr/bin/supervisord $pipework_dir/pipework ovs2 test2 172.16.0.9/[email protected]
运行这个脚本
[[email protected] tmp]# sh openvswitch_docker.sh 3999d60c5833 1b42d09f3311 a10c7b6f1141056e5276c44b348652e66921f322a96699903cf1c372858633d3 7e907cf62e593c27deb3ff455d1107074dfb975e68a3ac42ff844e84794322cd [[email protected] tmp]# docker ps -a CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES 7e907cf62e59 docker.ops-chukong.com:5000/centos6-http:new "/usr/bin/supervisor 3 seconds ago Up 2 seconds test2 a10c7b6f1141 docker.ops-chukong.com:5000/centos6-http:new "/usr/bin/supervisor 4 seconds ago Up 3 seconds test1
登陆分别的固定ip试试
[[email protected] tmp]# ssh 172.16.0.8 The authenticity of host '172.16.0.8 (172.16.0.8)' can't be established. RSA key fingerprint is 39:7c:13:9f:d4:b0:d7:63:fc:ff:ae:e3:46:a4:bf:6b. Are you sure you want to continue connecting (yes/no)? yes Warning: Permanently added '172.16.0.8' (RSA) to the list of known hosts. [email protected]'s password: Last login: Mon Nov 17 14:10:39 2014 from 172.17.42.1 [email protected]:~ 18:45:05 # ifconfig eth1 Link encap:Ethernet HWaddr CA:46:87:58:6C:BF inet addr:172.16.0.8 Bcast:0.0.0.0 Mask:255.255.0.0 inet6 addr: fe80::c846:87ff:fe58:6cbf/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:75 errors:0 dropped:2 overruns:0 frame:0 TX packets:41 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:10787 (10.5 KiB) TX bytes:5992 (5.8 KiB) lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 inet6 addr: ::1/128 Scope:Host UP LOOPBACK RUNNING MTU:65536 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 b) TX bytes:0 (0.0 b) [email protected]:~ 18:45:06 # ping 172.16.0.9 PING 172.16.0.9 (172.16.0.9) 56(84) bytes of data. 64 bytes from 172.16.0.9: icmp_seq=1 ttl=64 time=0.615 ms ^C --- 172.16.0.9 ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 531ms rtt min/avg/max/mdev = 0.615/0.615/0.615/0.000 ms [email protected]:~ 18:45:10 # ping 172.16.0.4 PING 172.16.0.4 (172.16.0.4) 56(84) bytes of data. 64 bytes from 172.16.0.4: icmp_seq=1 ttl=64 time=0.270 ms ^C --- 172.16.0.4 ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 581ms rtt min/avg/max/mdev = 0.270/0.270/0.270/0.000 ms [email protected]:~ 18:45:12 # ping www.baidu.com -c 2 PING www.a.shifen.com (180.149.131.236) 56(84) bytes of data. 64 bytes from 180.149.131.236: icmp_seq=1 ttl=54 time=1.90 ms 64 bytes from 180.149.131.236: icmp_seq=2 ttl=54 time=2.00 ms --- www.a.shifen.com ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1003ms rtt min/avg/max/mdev = 1.900/1.950/2.000/0.050 ms [email protected]:~
可以看到结果跟NODEA（10.10.17.3）里运行的一样，登陆后可以看到容器内的ip是指定的，并且能ping另外同一个网段的172.16.0.9，外网也能ping通
然后在试试能否ping通对方的em1网卡与对方ovs2的ip
4、在NODEA里测试
[email protected]:~ 18:58:48 # ping 10.10.17.3 -c 2 PING 10.10.17.3 (10.10.17.3) 56(84) bytes of data. 64 bytes from 10.10.17.3: icmp_seq=1 ttl=64 time=0.317 ms 64 bytes from 10.10.17.3: icmp_seq=2 ttl=64 time=0.042 ms --- 10.10.17.3 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 0.042/0.179/0.317/0.138 ms [email protected]:~ 18:58:52 # ping 10.10.17.4 -c 2 PING 10.10.17.4 (10.10.17.4) 56(84) bytes of data. 64 bytes from 10.10.17.4: icmp_seq=1 ttl=63 time=1.35 ms 64 bytes from 10.10.17.4: icmp_seq=2 ttl=63 time=0.271 ms --- 10.10.17.4 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1001ms rtt min/avg/max/mdev = 0.271/0.814/1.357/0.543 ms [email protected]:~ 18:58:56 # ping 172.16.0.3 -c 2 PING 172.16.0.3 (172.16.0.3) 56(84) bytes of data. 64 bytes from 172.16.0.3: icmp_seq=1 ttl=64 time=0.330 ms 64 bytes from 172.16.0.3: icmp_seq=2 ttl=64 time=0.040 ms --- 172.16.0.3 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 0.040/0.185/0.330/0.145 ms [email protected]:~ 18:59:04 # ping 172.16.0.4 -c 2 PING 172.16.0.4 (172.16.0.4) 56(84) bytes of data. From 172.16.0.5 icmp_seq=1 Destination Host Unreachable From 172.16.0.5 icmp_seq=2 Destination Host Unreachable --- 172.16.0.4 ping statistics --- 2 packets transmitted, 0 received, +2 errors, 100% packet loss, time 3007ms pipe 2
能ping通自己的em1与10.10.17.4的em1网卡，并且对方的ovs2的ip也能ping通，但ovs2里的主机无法ping通
5、在NODEB里测试
[email protected]:~ 18:59:35 # ping 10.10.17.4 -c2 PING 10.10.17.4 (10.10.17.4) 56(84) bytes of data. 64 bytes from 10.10.17.4: icmp_seq=1 ttl=64 time=0.306 ms 64 bytes from 10.10.17.4: icmp_seq=2 ttl=64 time=0.032 ms --- 10.10.17.4 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 0.032/0.169/0.306/0.137 ms [email protected]:~ 18:59:48 # ping 10.10.17.3 -c2 PING 10.10.17.3 (10.10.17.3) 56(84) bytes of data. 64 bytes from 10.10.17.3: icmp_seq=1 ttl=63 time=0.752 ms 64 bytes from 10.10.17.3: icmp_seq=2 ttl=63 time=0.268 ms --- 10.10.17.3 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 0.268/0.510/0.752/0.242 ms [email protected]:~ 18:59:51 # ping 172.16.0.4 -c2 PING 172.16.0.4 (172.16.0.4) 56(84) bytes of data. 64 bytes from 172.16.0.4: icmp_seq=1 ttl=64 time=0.215 ms 64 bytes from 172.16.0.4: icmp_seq=2 ttl=64 time=0.037 ms --- 172.16.0.4 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 0.037/0.126/0.215/0.089 ms [email protected]:~ 18:59:57 # ping 172.16.0.3 -c2 PING 172.16.0.3 (172.16.0.3) 56(84) bytes of data. From 172.16.0.8 icmp_seq=1 Destination Host Unreachable From 172.16.0.8 icmp_seq=2 Destination Host Unreachable --- 172.16.0.3 ping statistics --- 2 packets transmitted, 0 received, +2 errors, 100% packet loss, time 3006ms pipe 2
结果也是一样，能ping通自己的em1与NODEA（10.10.17.3）的em1网卡，并且对方的ovs2的ip也能ping通，但ovs2里的主机无法ping通
6、vxlan设置
在NODEA里运行
ovs-vsctl add-port ovs2 vx1 -- set interface vx1 type=vxlan options:remote_ip=10.10.17.4 [[email protected] tmp]# ovs-vsctl show d895d78b-8c89-49bc-b429-da6a4a2dcb3a Bridge "ovs1" Port "em1" Interface "em1" Port "ovs1" Interface "ovs1" type: internal Bridge "ovs2" Port "veth1pl15561" Interface "veth1pl15561" Port "veth1pl15662" Interface "veth1pl15662" Port "vx1" Interface "vx1" type: vxlan options: {remote_ip="10.10.17.4"} Port "ovs2" Interface "ovs2" type: internal ovs_version: "2.3.1"
在NODEB里运行
ovs-vsctl add-port ovs2 vx1 -- set interface vx1 type=vxlan options:remote_ip=10.10.17.3 [[email protected] tmp]# ovs-vsctl show 5a4b1bcd-3a91-4670-9c60-8fabbff37e85 Bridge "ovs2" Port "veth1pl28665" Interface "veth1pl28665" Port "ovs2" Interface "ovs2" type: internal Port "vx1" Interface "vx1" type: vxlan options: {remote_ip="10.10.17.3"} Port "veth1pl28766" Interface "veth1pl28766" Bridge "ovs1" Port "ovs1" Interface "ovs1" type: internal Port "em1" Interface "em1" ovs_version: "2.3.1"
现在NODEA与NODEB这2台物理机的网络都是互通的，容器的网络也是互通。
然后在NODEA（10.10.17.3）里ping NODEB（10.10.17.4）的ovs2 ip与容器的ip
[[email protected] tmp]# ssh 172.16.0.5 [email protected]'s password: Last login: Tue Feb 3 19:04:30 2015 from 172.16.0.3 [email protected]:~ 19:04:38 # ping 172.16.0.4 -c 2 PING 172.16.0.4 (172.16.0.4) 56(84) bytes of data. 64 bytes from 172.16.0.4: icmp_seq=1 ttl=64 time=0.623 ms 64 bytes from 172.16.0.4: icmp_seq=2 ttl=64 time=0.272 ms --- 172.16.0.4 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1001ms rtt min/avg/max/mdev = 0.272/0.447/0.623/0.176 ms [email protected]:~ 19:04:41 # ping 172.16.0.8 -c 2 PING 172.16.0.8 (172.16.0.8) 56(84) bytes of data. 64 bytes from 172.16.0.8: icmp_seq=1 ttl=64 time=1.76 ms 64 bytes from 172.16.0.8: icmp_seq=2 ttl=64 time=0.277 ms --- 172.16.0.8 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1002ms rtt min/avg/max/mdev = 0.277/1.023/1.769/0.746 ms [email protected]:~ 19:04:44 # ping 172.16.0.9 -c 2 PING 172.16.0.9 (172.16.0.9) 56(84) bytes of data. 64 bytes from 172.16.0.9: icmp_seq=1 ttl=64 time=1.83 ms 64 bytes from 172.16.0.9: icmp_seq=2 ttl=64 time=0.276 ms --- 172.16.0.9 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1002ms rtt min/avg/max/mdev = 0.276/1.053/1.830/0.777 ms [email protected]:~
可以看到可以在NODEA（10.10.17.3）里ping通NODEB（10.10.17.4）的ovs2 ip与交换机下面的容器ip
如果各自设置vxlan，还是无法连接请看看iptables里是否给ovs1进行了input放行
[[email protected] tmp]# cat /etc/sysconfig/iptables # Generated by iptables-save v1.4.7 on Fri Dec 6 10:59:13 2013 *filter :INPUT DROP [0:0] :FORWARD ACCEPT [0:0] :OUTPUT ACCEPT [1:83] -A INPUT -m state --state RELATED,ESTABLISHED -j ACCEPT -A INPUT -p icmp -j ACCEPT -A INPUT -i lo -j ACCEPT -A INPUT -i em1 -j ACCEPT -A INPUT -i ovs1 -j ACCEPT -A INPUT -p tcp -m multiport --dports 50020 -j ACCEPT -A INPUT -p tcp -j REJECT --reject-with tcp-reset -A FORWARD -p tcp -m tcp --tcp-flags FIN,SYN,RST,ACK RST -m limit --limit 1/sec -j ACCEPT COMMIT # Completed on Fri Dec 6 10:59:13 2013 *nat :PREROUTING ACCEPT [2:269] :POSTROUTING ACCEPT [1739:127286] :OUTPUT ACCEPT [1739:127286] :DOCKER - [0:0] -A PREROUTING -m addrtype --dst-type LOCAL -j DOCKER -A POSTROUTING -s 172.16.0.0/8 ! -d 172.16.0.0/8 -j MASQUERADE -A OUTPUT ! -d 127.0.0.0/8 -m addrtype --dst-type LOCAL -j DOCKER COMMIT
在NODEB里测试
[[email protected] tmp]# ssh 172.16.0.8 [email protected]'s password: Last login: Tue Feb 3 18:59:35 2015 from 172.16.0.4 [email protected]:~ 19:08:08 # ping 172.16.0.3 -c 2 PING 172.16.0.3 (172.16.0.3) 56(84) bytes of data. 64 bytes from 172.16.0.3: icmp_seq=1 ttl=64 time=1.48 ms 64 bytes from 172.16.0.3: icmp_seq=2 ttl=64 time=0.289 ms --- 172.16.0.3 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1002ms rtt min/avg/max/mdev = 0.289/0.889/1.489/0.600 ms [email protected]:~ 19:08:13 # ping 172.16.0.5 -c 2 PING 172.16.0.5 (172.16.0.5) 56(84) bytes of data. 64 bytes from 172.16.0.5: icmp_seq=1 ttl=64 time=1.27 ms 64 bytes from 172.16.0.5: icmp_seq=2 ttl=64 time=0.289 ms --- 172.16.0.5 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1001ms rtt min/avg/max/mdev = 0.289/0.783/1.277/0.494 ms [email protected]:~ 19:08:16 # ping 172.16.0.6 -c 2 PING 172.16.0.6 (172.16.0.6) 56(84) bytes of data. 64 bytes from 172.16.0.6: icmp_seq=1 ttl=64 time=1.32 ms 64 bytes from 172.16.0.6: icmp_seq=2 ttl=64 time=0.275 ms --- 172.16.0.6 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1001ms rtt min/avg/max/mdev = 0.275/0.800/1.326/0.526 ms [email protected]:~
结果也是一样，设置了vxlan就可以2个宿主机的所有服务器进行通信。
目前是2个节点的vxlan，如果是3个节点呢
7、vxlan多节点应用（超过2个节点）
架构图为
新节点是NODEC（ip是10.10.21.199）
环境为
部署单机环境，脚本内容是
#!/bin/bash #author: Deng Lei #email: [email protected] #删除docker测试机 docker rm `docker stop $(docker ps -a -q)` #删除已有的openvswitch交换机 ovs-vsctl list-br|xargs -I {} ovs-vsctl del-br {} #创建交换机 ovs-vsctl add-br ovs1 ovs-vsctl add-br ovs2 #把物理网卡加入ovs2 ovs-vsctl add-port ovs1 em1 ip link set ovs1 up ifconfig em1 0 ifconfig ovs1 10.10.21.199 ip link set ovs2 up ip addr add 172.16.0.11/16 dev ovs2 pipework_dir='/tmp/pipework' docker run --restart always --privileged -d --net="none" --name='test1' docker.ops-chukong.com:5000/centos6-http:new /usr/bin/supervisord $pipework_dir/pipework ovs2 test1 172.16.0.12/[email protected] docker run --restart always --privileged -d --net="none" --name='test2' docker.ops-chukong.com:5000/centos6-http:new /usr/bin/supervisord $pipework_dir/pipework ovs2 test2 172.16.0.13/[email protected]
运行脚本
19:14:13 # sh openvswitch_docker.sh 534db79be57a c014d47212868080a2a3c473b916ccbc9a45ecae4ab564807104faeb7429e1d3 5a98c459438d6e2c37d040f18282509763acc16d63010adc72852585dc255666 测试 [email protected]:/tmp 19:17:00 # ssh 172.16.0.12 [email protected]'s password: Last login: Mon Nov 17 14:10:39 2014 from 172.17.42.1 [email protected]:~ 19:17:08 # ifconfig eth1 Link encap:Ethernet HWaddr 4A:B3:FE:B7:01:C7 inet addr:172.16.0.12 Bcast:0.0.0.0 Mask:255.255.0.0 inet6 addr: fe80::48b3:feff:feb7:1c7/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:117 errors:0 dropped:0 overruns:0 frame:0 TX packets:89 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:20153 (19.6 KiB) TX bytes:16600 (16.2 KiB) lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 inet6 addr: ::1/128 Scope:Host UP LOOPBACK RUNNING MTU:65536 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 b) TX bytes:0 (0.0 b) [email protected]:~ 19:17:10 # ping 172.16.0.11 -c 2 PING 172.16.0.11 (172.16.0.11) 56(84) bytes of data. 64 bytes from 172.16.0.11: icmp_seq=1 ttl=64 time=0.240 ms 64 bytes from 172.16.0.11: icmp_seq=2 ttl=64 time=0.039 ms --- 172.16.0.11 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 0.039/0.139/0.240/0.101 ms [email protected]:~ 19:17:16 # ping 172.16.0.12 -c 2 PING 172.16.0.12 (172.16.0.12) 56(84) bytes of data. 64 bytes from 172.16.0.12: icmp_seq=1 ttl=64 time=0.041 ms 64 bytes from 172.16.0.12: icmp_seq=2 ttl=64 time=0.026 ms --- 172.16.0.12 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 999ms rtt min/avg/max/mdev = 0.026/0.033/0.041/0.009 ms [email protected]:~ 19:17:18 # ping 172.16.0.13 -c 2 PING 172.16.0.13 (172.16.0.13) 56(84) bytes of data. 64 bytes from 172.16.0.13: icmp_seq=1 ttl=64 time=0.316 ms 64 bytes from 172.16.0.13: icmp_seq=2 ttl=64 time=0.040 ms --- 172.16.0.13 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 0.040/0.178/0.316/0.138 ms [email protected]:~ 19:17:21 # ping www.baidu.com -c 2 PING www.a.shifen.com (180.149.131.205) 56(84) bytes of data. 64 bytes from 180.149.131.205: icmp_seq=1 ttl=54 time=2.65 ms 64 bytes from 180.149.131.205: icmp_seq=2 ttl=54 time=1.93 ms --- www.a.shifen.com ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1002ms rtt min/avg/max/mdev = 1.939/2.295/2.651/0.356 ms [email protected]:~
可以看到可以ping通本地的ovs2 的ip与交换机下面是ip
[email protected]:~ 19:17:26 # ping 10.10.17.3 -c 2 PING 10.10.17.3 (10.10.17.3) 56(84) bytes of data. 64 bytes from 10.10.17.3: icmp_seq=1 ttl=63 time=0.418 ms 64 bytes from 10.10.17.3: icmp_seq=2 ttl=63 time=0.213 ms --- 10.10.17.3 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 0.213/0.315/0.418/0.104 ms [email protected]:~ 19:18:05 # ping 10.10.17.4 -c 2 PING 10.10.17.4 (10.10.17.4) 56(84) bytes of data. 64 bytes from 10.10.17.4: icmp_seq=1 ttl=63 time=0.865 ms 64 bytes from 10.10.17.4: icmp_seq=2 ttl=63 time=0.223 ms --- 10.10.17.4 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 0.223/0.544/0.865/0.321 ms [email protected]:~ 19:18:08 # ping 172.16.0.3 -c 2 PING 172.16.0.3 (172.16.0.3) 56(84) bytes of data. From 172.16.0.12 icmp_seq=1 Destination Host Unreachable From 172.16.0.12 icmp_seq=2 Destination Host Unreachable --- 172.16.0.3 ping statistics --- 2 packets transmitted, 0 received, +2 errors, 100% packet loss, time 3006ms pipe 2 [email protected]:~ 19:18:19 # ping 172.16.0.4 -c 2 PING 172.16.0.4 (172.16.0.4) 56(84) bytes of data. From 172.16.0.12 icmp_seq=1 Destination Host Unreachable From 172.16.0.12 icmp_seq=2 Destination Host Unreachable --- 172.16.0.4 ping statistics --- 2 packets transmitted, 0 received, +2 errors, 100% packet loss, time 3004ms pipe 2 [email protected]:~
可以看到能ping通NODEA（10.10.17.3）与NODEB（10.10.17.4）（em1网卡都是走物理交换机），但他们2个的ovs2都无法ping通
下面是在NODEC（10.10.21.199）里与10.10.17.3做一个vxlan
19:19:22 # ovs-vsctl add-port ovs2 vx1 -- set interface vx1 type=vxlan options:remote_ip=10.10.17.3 [email protected]:/tmp 19:20:01 # ovs-vsctl show 96259d0f-b794-49fd-81fb-3251ede9c2a5 Bridge "ovs2" Port "veth1pl30126" Interface "veth1pl30126" Port "vx1" Interface "vx1" type: vxlan options: {remote_ip="10.10.17.3"} Port "veth1pl30247" Interface "veth1pl30247" Port "ovs2" Interface "ovs2" type: internal Bridge "ovs1" Port "ovs1" Interface "ovs1" type: internal Port "em1" Interface "em1" ovs_version: "2.3.1"
然后还需要在NODEA（10.10.17.3）里配置
[[email protected] x86_64]# ovs-vsctl add-port ovs2 vx2 -- set interface vx2 type=vxlan options:remote_ip=10.10.21.199 [[email protected] x86_64]# ovs-vsctl show d895d78b-8c89-49bc-b429-da6a4a2dcb3a Bridge "ovs1" Port "em1" Interface "em1" Port "ovs1" Interface "ovs1" type: internal Bridge "ovs2" Port "veth1pl15561" Interface "veth1pl15561" Port "veth1pl15662" Interface "veth1pl15662" Port "vx2" Interface "vx2" type: vxlan options: {remote_ip="10.10.21.199"} Port "vx1" Interface "vx1" type: vxlan options: {remote_ip="10.10.17.4"} Port "ovs2" Interface "ovs2" type: internal ovs_version: "2.3.1"
之前在NODEA（10.10.17.3）里与NODE（10.10.17.4）做的vxlan使用vx1，这里NODEA（10.10.17.3）与NODEC（10.10.21.199）就使用vx2端口
然后在NODEA（10.10.17.3）里ping NODEC（10.10.21.199）的ovs2 ip与交换机下面的ip
[[email protected] x86_64]# ping 172.16.0.11 -c 2 PING 172.16.0.11 (172.16.0.11) 56(84) bytes of data. 64 bytes from 172.16.0.11: icmp_seq=1 ttl=64 time=1.48 ms 64 bytes from 172.16.0.11: icmp_seq=2 ttl=64 time=0.244 ms --- 172.16.0.11 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1001ms rtt min/avg/max/mdev = 0.244/0.865/1.487/0.622 ms [[email protected] x86_64]# ping 172.16.0.12 -c 2 PING 172.16.0.12 (172.16.0.12) 56(84) bytes of data. 64 bytes from 172.16.0.12: icmp_seq=1 ttl=64 time=1.54 ms 64 bytes from 172.16.0.12: icmp_seq=2 ttl=64 time=0.464 ms --- 172.16.0.12 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1001ms rtt min/avg/max/mdev = 0.464/1.006/1.549/0.543 ms [[email protected] x86_64]# ping 172.16.0.13 -c 2 PING 172.16.0.13 (172.16.0.13) 56(84) bytes of data. 64 bytes from 172.16.0.13: icmp_seq=1 ttl=64 time=1.73 ms 64 bytes from 172.16.0.13: icmp_seq=2 ttl=64 time=0.232 ms --- 172.16.0.13 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1001ms rtt min/avg/max/mdev = 0.232/0.983/1.735/0.752 ms
可以看到是通的
在NODEC（10.10.21.199）里ping NODEA（10.10.17.3）的ovs2的ip与交换机下面的ip
[email protected]:/tmp 19:23:17 # ping 172.16.0.3 -c 2 PING 172.16.0.3 (172.16.0.3) 56(84) bytes of data. 64 bytes from 172.16.0.3: icmp_seq=1 ttl=64 time=0.598 ms 64 bytes from 172.16.0.3: icmp_seq=2 ttl=64 time=0.300 ms --- 172.16.0.3 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 999ms rtt min/avg/max/mdev = 0.300/0.449/0.598/0.149 ms [email protected]:/tmp 19:23:26 # ping 172.16.0.5 -c 2 PING 172.16.0.5 (172.16.0.5) 56(84) bytes of data. 64 bytes from 172.16.0.5: icmp_seq=1 ttl=64 time=1.23 ms 64 bytes from 172.16.0.5: icmp_seq=2 ttl=64 time=0.214 ms --- 172.16.0.5 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1001ms rtt min/avg/max/mdev = 0.214/0.726/1.239/0.513 ms [email protected]:/tmp 19:23:29 # ping 172.16.0.6 -c 2 PING 172.16.0.6 (172.16.0.6) 56(84) bytes of data. 64 bytes from 172.16.0.6: icmp_seq=1 ttl=64 time=1.00 ms 64 bytes from 172.16.0.6: icmp_seq=2 ttl=64 time=0.226 ms --- 172.16.0.6 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1001ms rtt min/avg/max/mdev = 0.226/0.613/1.000/0.387 ms [email protected]:/tmp
也是通的，然后从NODEC（10.10.21.199） ping NODEB（10.10.17.4）的ovs2的ip与其交换机的ip
[email protected]:/tmp 19:55:05 # ping -c 10 172.16.0.4 PING 172.16.0.4 (172.16.0.4) 56(84) bytes of data. 64 bytes from 172.16.0.4: icmp_seq=1 ttl=64 time=1.99 ms 64 bytes from 172.16.0.4: icmp_seq=2 ttl=64 time=0.486 ms 64 bytes from 172.16.0.4: icmp_seq=3 ttl=64 time=0.395 ms 64 bytes from 172.16.0.4: icmp_seq=4 ttl=64 time=0.452 ms 64 bytes from 172.16.0.4: icmp_seq=5 ttl=64 time=0.457 ms 64 bytes from 172.16.0.4: icmp_seq=6 ttl=64 time=0.461 ms 64 bytes from 172.16.0.4: icmp_seq=7 ttl=64 time=0.457 ms 64 bytes from 172.16.0.4: icmp_seq=8 ttl=64 time=0.428 ms 64 bytes from 172.16.0.4: icmp_seq=9 ttl=64 time=0.492 ms 64 bytes from 172.16.0.4: icmp_seq=10 ttl=64 time=0.461 ms --- 172.16.0.4 ping statistics --- 10 packets transmitted, 10 received, 0% packet loss, time 9000ms rtt min/avg/max/mdev = 0.395/0.608/1.995/0.463 ms
可以看到是通的，平均延迟0.608，并且可以发现使用了vxlan，3个节点，如果想全部互通，只需要2个线连接就行。
如果使用gre模式，3个节点就需要3个线了，架构图为
目前使用docker结合openvswitch的vxlan模式就把多台主机的docker连接起来，这样很多测试就方便很多，但还是建议把这样的方式作为测试环境。
转载于:https://blog.51cto.com/dl528888/1611491