docker自带了EXPOSE命令,可以通过编写dockerfile-p参数方便的映射container内部端口,但是对于已经运行的container,如果你想对外开放一个新的端口,只能编辑dockerfile然后重新build,有点不太方便。

其实docker本身使用了iptables来做端口映射的,所以我们可以通过一些简单的操作来实现动态映射运行中的container端口。

通过运行iptables命令可以看到具体的端口映射(下面的实例中IP为192.168.42.41的container开放了22和20280等端口)

[yaxin@ubox ~]$sudo iptables -nvxL
Chain INPUT (policy ACCEPT 262 packets, 529689 bytes)
    pkts      bytes target     prot opt in     out     source               destination
   14355   789552 DROP       tcp  --  *      *       0.0.0.0/0            0.0.0.0/0            tcp dpt:25

Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
    pkts      bytes target     prot opt in     out     source               destination
 5479459 653248187 DOCKER     all  --  *      docker0  0.0.0.0/0            0.0.0.0/0
   93990 314970368 ACCEPT     all  --  *      docker0  0.0.0.0/0            0.0.0.0/0            ctstate RELATED,ESTABLISHED
 4705395 2183219154 ACCEPT     all  --  docker0 !docker0  0.0.0.0/0            0.0.0.0/0
       0        0 ACCEPT     all  --  docker0 docker0  0.0.0.0/0            0.0.0.0/0

Chain OUTPUT (policy ACCEPT 282 packets, 622495 bytes)
    pkts      bytes target     prot opt in     out     source               destination

Chain DOCKER (1 references)
    pkts      bytes target     prot opt in     out     source               destination
     218    13193 ACCEPT     tcp  --  !docker0 docker0  0.0.0.0/0            192.168.42.41        tcp dpt:22280
 4868186 297463902 ACCEPT     tcp  --  !docker0 docker0  0.0.0.0/0            192.168.42.41        tcp dpt:20280
   78663 13128102 ACCEPT     tcp  --  !docker0 docker0  0.0.0.0/0            192.168.42.41        tcp dpt:22
      47     4321 ACCEPT     tcp  --  !docker0 docker0  0.0.0.0/0            192.168.42.41        tcp dpt:28159
[yaxin@ubox ~]$sudo iptables -t nat -nvxL
Chain PREROUTING (policy ACCEPT 210199 packets, 14035875 bytes)
    pkts      bytes target     prot opt in     out     source               destination
 1219483 82563968 DOCKER     all  --  *      *       0.0.0.0/0            0.0.0.0/0            ADDRTYPE match dst-type LOCAL

Chain INPUT (policy ACCEPT 197679 packets, 13316595 bytes)
    pkts      bytes target     prot opt in     out     source               destination

Chain OUTPUT (policy ACCEPT 271553 packets, 16671466 bytes)
    pkts      bytes target     prot opt in     out     source               destination
    1643    99251 DOCKER     all  --  *      *       0.0.0.0/0           !127.0.0.0/8          ADDRTYPE match dst-type LOCAL

Chain POSTROUTING (policy ACCEPT 271743 packets, 16682594 bytes)
    pkts      bytes target     prot opt in     out     source               destination
   13468   811013 MASQUERADE  all  --  *      !docker0  192.168.42.0/24      0.0.0.0/0
       0        0 MASQUERADE  tcp  --  *      *       192.168.42.41        192.168.42.41        tcp dpt:22280
       0        0 MASQUERADE  tcp  --  *      *       192.168.42.41        192.168.42.41        tcp dpt:20280
       0        0 MASQUERADE  tcp  --  *      *       192.168.42.41        192.168.42.41        tcp dpt:22
       0        0 MASQUERADE  tcp  --  *      *       192.168.42.41        192.168.42.41        tcp dpt:28159

Chain DOCKER (2 references)
    pkts      bytes target     prot opt in     out     source               destination
      22     1404 DNAT       tcp  --  !docker0 *       0.0.0.0/0            0.0.0.0/0            tcp dpt:22280 to:192.168.42.41:22280
     288    17156 DNAT       tcp  --  !docker0 *       0.0.0.0/0            0.0.0.0/0            tcp dpt:20280 to:192.168.42.41:20280
      93     5952 DNAT       tcp  --  !docker0 *       0.0.0.0/0            0.0.0.0/0            tcp dpt:22222 to:192.168.42.41:22
       8      512 DNAT       tcp  --  !docker0 *       0.0.0.0/0            0.0.0.0/0            tcp dpt:28159 to:192.168.42.41:28159

我们要做的就是根据自己container的情况配置iptables规则。

首先是filter这个表,我们要配置其放通转发,docker默认已经将所有的FORWARD规则放到了DOCKER这个自建链(chain)中了,这样方便配置,也方便查看。

然后再配置nat表为其配置DNAT,这个才是端口转发的核心配置。这个表中只需要配置POSTROUTINGDOCKER链即可,这里不讲为什么这么配置,如果想要深入了解iptables请google一下。

下面举个例子:

假如我有一个container,名字为nginx(通过运行docker ps命令即可查询),现在我在docker内部运行nginx程序,监听了8888端口,我希望外网可以通过8899端口(注意一下端口)访问

  1. 找到docker为nginx分配的IP
[yaxin@ubox ~]$sudo docker inspect -f '{{.NetworkSettings.IPAddress}}' nginx
192.168.42.43
  1. 配置iptables中filter表的FORWARD(DOCKER)链
[yaxin@ubox ~]$sudo iptables -A DOCKER ! -i docker0 -o docker0 -p tcp --dport 8888 -d 192.168.42.43 -j ACCEPT
  1. 配置iptables中nat表的PREROUTING(DOCKER)和POSTROUTING链
[yaxin@ubox ~]$sudo iptables -t nat -A POSTROUTING -p tcp --dport 8888 -s 192.168.42.43 -d 192.168.42.43 -j MASQUERADE
[yaxin@ubox ~]$sudo iptables -t nat -A DOCKER ! -i dokcer0 -p tcp --dport 8899 -j DNAT --to-destination 192.168.42.43:8888
  1. 通过外网访问curl http://IP:8899就会显示nginx下配置的html页面

  2. 最后iptables规则

[yaxin@ubox ~]$sudo iptables -nvxL
Chain INPUT (policy ACCEPT 67893 packets, 212661547 bytes)
    pkts      bytes target     prot opt in     out     source               destination
   14364   790008 DROP       tcp  --  *      *       0.0.0.0/0            0.0.0.0/0            tcp dpt:25

Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
    pkts      bytes target     prot opt in     out     source               destination
 5479682 653269356 DOCKER     all  --  *      docker0  0.0.0.0/0            0.0.0.0/0
   94186 314986910 ACCEPT     all  --  *      docker0  0.0.0.0/0            0.0.0.0/0            ctstate RELATED,ESTABLISHED
 4705658 2183254076 ACCEPT     all  --  docker0 !docker0  0.0.0.0/0            0.0.0.0/0
       0        0 ACCEPT     all  --  docker0 docker0  0.0.0.0/0            0.0.0.0/0

Chain OUTPUT (policy ACCEPT 71253 packets, 222512872 bytes)
    pkts      bytes target     prot opt in     out     source               destination

Chain DOCKER (1 references)
    pkts      bytes target     prot opt in     out     source               destination
     218    13193 ACCEPT     tcp  --  !docker0 docker0  0.0.0.0/0            192.168.42.41        tcp dpt:22280
 4868186 297463902 ACCEPT     tcp  --  !docker0 docker0  0.0.0.0/0            192.168.42.41        tcp dpt:20280
   78663 13128102 ACCEPT     tcp  --  !docker0 docker0  0.0.0.0/0            192.168.42.41        tcp dpt:22
      47     4321 ACCEPT     tcp  --  !docker0 docker0  0.0.0.0/0            192.168.42.41        tcp dpt:28159
      27     4627 ACCEPT     tcp  --  !docker0 docker0  0.0.0.0/0            192.168.42.43        tcp dpt:8888
[yaxin@ubox ~]$sudo iptables -t nat -nvxL
Chain PREROUTING (policy ACCEPT 232 packets, 16606 bytes)
    pkts      bytes target     prot opt in     out     source               destination
 1220281 82620790 DOCKER     all  --  *      *       0.0.0.0/0            0.0.0.0/0            ADDRTYPE match dst-type LOCAL

Chain INPUT (policy ACCEPT 216 packets, 15671 bytes)
    pkts      bytes target     prot opt in     out     source               destination

Chain OUTPUT (policy ACCEPT 317 packets, 19159 bytes)
    pkts      bytes target     prot opt in     out     source               destination
    1644    99311 DOCKER     all  --  *      *       0.0.0.0/0           !127.0.0.0/8          ADDRTYPE match dst-type LOCAL

Chain POSTROUTING (policy ACCEPT 321 packets, 19367 bytes)
    pkts      bytes target     prot opt in     out     source               destination
   13512   813656 MASQUERADE  all  --  *      !docker0  192.168.42.0/24      0.0.0.0/0
       0        0 MASQUERADE  tcp  --  *      *       192.168.42.41        192.168.42.41        tcp dpt:22280
       0        0 MASQUERADE  tcp  --  *      *       192.168.42.41        192.168.42.41        tcp dpt:20280
       0        0 MASQUERADE  tcp  --  *      *       192.168.42.41        192.168.42.41        tcp dpt:22
       0        0 MASQUERADE  tcp  --  *      *       192.168.42.41        192.168.42.41        tcp dpt:28159
       0        0 MASQUERADE  tcp  --  *      *       192.168.42.43        192.168.42.43        tcp dpt:8888

Chain DOCKER (2 references)
    pkts      bytes target     prot opt in     out     source               destination
      22     1404 DNAT       tcp  --  !docker0 *       0.0.0.0/0            0.0.0.0/0            tcp dpt:22280 to:192.168.42.41:22280
     288    17156 DNAT       tcp  --  !docker0 *       0.0.0.0/0            0.0.0.0/0            tcp dpt:20280 to:192.168.42.41:20280
      93     5952 DNAT       tcp  --  !docker0 *       0.0.0.0/0            0.0.0.0/0            tcp dpt:22222 to:192.168.42.41:22
       8      512 DNAT       tcp  --  !docker0 *       0.0.0.0/0            0.0.0.0/0            tcp dpt:28159 to:192.168.42.41:28159
       4      208 DNAT       tcp  --  !dokcer0 *       0.0.0.0/0            0.0.0.0/0            tcp dpt:8899 to:192.168.42.43:8888

当然,使用手动配置也是比较麻烦的,所以我写了一个脚本来自动配置端口映射,使用方法脚本中有说明

#!/bin/bash
# filename: docker_expose.sh

if [ `id -u` -ne 0 ];then
    echo "[EROOR] Please use root to run this script"
    exit 23
fi

if [ $# -ne 3 ];then
    echo "Usage: $0 <container_name> <add|del> [[<machine_ip>:]<machine_port>:]<container_port>[/<protocol_type>]"
    exit 1
fi

IPV4_RE='(([0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\.){3}([0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])'

container_name=$1
action=$2
arguments=$3

# check action
if [ "$action"x != "add"x -a "$action"x != "del"x ];then
    echo "[ERROR] Please use add or del parameter to add port map or delete port map"
    exit 654
fi
if [ "$action"x == "add"x ];then
    action="A"
else
    action="D"
fi

# get container ip by container name
container_ip=`docker inspect -f '{{.NetworkSettings.IPAddress}}' $container_name 2> /dev/null`
if [ -z $container_ip ];then
    echo "[ERROR] Get container's (${container_name}) IP error, please ensure you have this container"
    exit 2
fi

# parse arguments
protocol_type=`echo $arguments | awk -F '/' '{print $2}'`
if [ -z $protocol_type ];then
    protocol_type="tcp"
fi

# check protocol
if [ "$protocol_type"x != "tcp"x -a "$protocol_type"x != "udp"x ];then
    echo "[ERROR] Only tcp or udp protocol is allowed"
    exit 99
fi

machine_ip=''
machine_port=''
container_port=''
# split the left arguments
arguments=${arguments%/*}
machine_ip=`echo $arguments | awk -F ':' '{print $1}'`
machine_port=`echo $arguments | awk -F ':' '{print $2}'`
container_port=`echo $arguments | awk -F ':' '{print $3}'`
if [ -z $machine_port ];then
    # arguments is: 234
    container_port=$machine_ip
    machine_port=$machine_ip
    unset machine_ip
elif [ -z $container_port ];then
    # arguments is: 234:456
    container_port=$machine_ip
    machine_port=$machine_port
    unset machine_ip
fi

# check port number function
_check_port_number() {
    local port_num=$1
    if ! echo $port_num | egrep "^[0-9]+$" &> /dev/null;then
        echo "[ERROR] Invalid port number $port_num"
        exit 3
    fi
    if [ $port_num -gt 65535 -o $port_num -lt 1 ];then
        echo "[ERROR] Port number $port_num is out of range(1-56635)"
        exit 4
    fi
}

# check port and ip address
_check_port_number $container_port
_check_port_number $machine_port

if [ ! -z $machine_ip ];then
    if ! echo $machine_ip | egrep "^${IPV4_RE}$" &> /dev/null;then
        echo "[ERROR] Invalid Ip Adress $machine_ip"
        exit 5
    fi

    # check which interface bind the IP
    for interface in `ifconfig -s | sed -n '2,$p' | awk '{print $1}'`;do
        interface_ip=`ifconfig $interface | awk '/inet addr/{print substr($2,6)}'`
        if [ "$interface_ip"x == "$machine_ip"x ];then
            interface_name=$interface
            break
        fi
    done

    if [ -z $interface_name ];then
        echo "[ERROR] Can not find interface bind with $machine_ip"
        exit 98
    fi
fi

# run iptables command
echo "[INFO] Now start to change rules to iptables"
echo "[INFO] Changing POSTROUTING chain of nat table"
iptables -t nat -${action} POSTROUTING -p ${protocol_type} --dport ${container_port} -s ${container_ip} -d ${container_ip} -j MASQUERADE
if [ -z $interface_name ];then
    echo "[INFO] Changing DOCKER chain of filter table"
    iptables -${action} DOCKER ! -i docker0 -o docker0 -p ${protocol_type} --dport ${container_port} -d ${container_ip} -j ACCEPT
    echo "[INFO] Changing DOCKER chain of nat table"
    iptables -t nat -${action} DOCKER ! -i docker0 -p ${protocol_type} --dport ${machine_port} -j DNAT --to-destination ${container_ip}:${container_port}
else
    echo "[INFO] Changing DOCKER chain of filter table"
    iptables -${action} DOCKER -i $interface_name -o docker0 -p ${protocol_type} --dport ${container_port} -d ${container_ip} -j ACCEPT
    echo "[INFO] Changing DOCKER chain of nat table"
    iptables -t nat -${action} DOCKER -i $interface_name -p ${protocol_type} --dport ${machine_port} -j DNAT --to-destination ${container_ip}:${container_port}
fi