Swarm Mode负载均衡

时间 2019-12-19

原文原文链接

Libnetwork

Libnetwork最初是由libcontainer和Docker Engine中的网络相关的代码合并而成的，是Docker容器网络库，最核心的内容是其定义的Container Network Model（CNM）。node

Libnetwork CNM 定义了Docker容器的网络模型，按照该模型开发出的driver就能与docker daemon协同工做，实现容器网络。docker 原生的driver包括 none、bridge、overlay和macvlan，第三方driver包括flannel、weave、calico等。linux

CNM

CNM定义了以下三个组件：
nginx

Sandbox
Sandbox是Docker容器中一个网络配置的隔离环境，包含容器的interface、路由表和DNS设置。Linux Network Namespace是Sandbox的标准实现。Sandbox能够包含来自不一样 Network的Endpoint。docker
Endpoint
Endpoint是一个在Network中进行网络通信的接口（veth pair），用于将Sandbox接入Network。一个Endpoint只能属于一个Network，也只能属于一个Sandbox
Endpoint能够加入一个network，但多个Endpoint能够在一个Sandbox中共存。bash
Network
一个Network就是一个惟一的、可识别的endpoint组，组内endpoint能够相互通信。Network的实现能够是linux Bridge、VLAN等。

[root@swarm-manager ~]# ll /var/run/docker/netns/
total 0
-r--r--r-- 1 root root 0 Aug  5 10:45 1-i6xug49nwd
-r--r--r-- 1 root root 0 Aug  5 10:45 ingress_sbox
[root@swarm-manager ~]# nsenter --net=/var/run/docker/netns/ingress_sbox ifconfig
eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        inet 10.255.0.2  netmask 255.255.0.0  broadcast 0.0.0.0
        ether 02:42:0a:ff:00:02  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

eth1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.18.0.2  netmask 255.255.0.0  broadcast 0.0.0.0
        ether 02:42:ac:12:00:02  txqueuelen 0  (Ethernet)
        RX packets 90  bytes 75247 (73.4 KiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 123  bytes 10271 (10.0 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        loop  txqueuelen 0  (Local Loopback)
        RX packets 6  bytes 504 (504.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 6  bytes 504 (504.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0
[root@swarm-manager ~]# iptables -t mangle -S
-P PREROUTING ACCEPT
-P INPUT ACCEPT
-P FORWARD ACCEPT
-P OUTPUT ACCEPT
-P POSTROUTING ACCEPT
[root@swarm-manager ~]# ipvsadm
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port           Forward Weight ActiveConn InActConn

[root@swarm-manager ~]# nsenter --net=/var/run/docker/netns/1-i6xug49nwd ifconfig
br0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        inet 10.255.0.1  netmask 255.255.0.0  broadcast 0.0.0.0
        ether 3a:31:2c:7f:21:a8  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        loop  txqueuelen 0  (Local Loopback)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

veth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        ether 4e:75:b1:f9:5b:55  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

vxlan0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        ether 3a:31:2c:7f:21:a8  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

经过nsenter --net=<SandboxKey> ip add能够直接进入容器的Sandbox，查询相关的网络信息网络

load balancing

internal load balancing

vip（基于ipvs）

建立一个自定义的overlay网络：
只需在manager节点建立，当有Service链接该overlay网络时，将会自动在所分配的worker节点上自动建立该overlay网络。

[root@swarm-manager ~]# docker network create -d overlay --subnet 192.168.10.0/24 my-network
[root@swarm-manager ~]# docker network ls -f name=my-network
NETWORK ID          NAME                DRIVER              SCOPE
nvpbs39b6ctz        my-network          overlay             swarm
[root@swarm-manager ~]# docker network inspect -f {{.IPAM.Config}} my-network
[{192.168.10.0/24  192.168.10.1 map[]}]

建立一个链接到自定义的overlay网络的service（未指定"endpoint-mode"则默认为VIP模式）
Docker会为每一个overlay网络建立一个独立的network namespace，其中会有一个linux bridge br0，endpoint仍是由veth pair实现，一端链接到容器中的eth0，另外一端链接到network namespace的br0上。
br0除了链接全部的endpoint，还会链接一个vxlan设备，用于与其余host创建vxlan tunnel。容器之间的数据就是经过这个tunnel通讯的。

[root@swarm-manager ~]# docker service create --replicas 2 --name nginx-vip --network my-network nginx
[root@swarm-manager ~]# docker service inspect -f {{.Endpoint.VirtualIPs}} nginx-vip
[{nvpbs39b6ctzrfw6vj809kjbu 192.168.10.2/24}]

[root@swarm-node1 ~]# ls -lrt /var/run/docker/netns/
total 0
-r--r--r--. 1 root root 0 Aug  6 11:24 ingress_sbox
-r--r--r--. 1 root root 0 Aug  6 11:24 1-i6xug49nwd
-r--r--r--. 1 root root 0 Aug 10 16:30 1-vxe1cwk14a
-r--r--r--. 1 root root 0 Aug 10 16:30 18531514ffd0
[root@swarm-node1 ~]# nsenter --net=/var/run/docker/netns/1-vxe1cwk14a ifconfig
br0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        inet 192.168.10.1  netmask 255.255.255.0  broadcast 0.0.0.0
        ether 5a:f5:4c:f3:98:d5  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        loop  txqueuelen 0  (Local Loopback)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

veth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        ether 96:57:63:17:80:83  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

vxlan0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        ether 5a:f5:4c:f3:98:d5  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0
[root@swarm-node1 ~]# nsenter --net=/var/run/docker/netns/18531514ffd0 ifconfig
eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        inet 192.168.10.4  netmask 255.255.255.0  broadcast 0.0.0.0
        ether 02:42:c0:a8:0a:04  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

eth1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.18.0.3  netmask 255.255.0.0  broadcast 0.0.0.0
        ether 02:42:ac:12:00:03  txqueuelen 0  (Ethernet)
        RX packets 7484  bytes 16821353 (16.0 MiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 4780  bytes 321518 (313.9 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        loop  txqueuelen 0  (Local Loopback)
        RX packets 14  bytes 1717 (1.6 KiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 14  bytes 1717 (1.6 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

service-name将经过内置的DNS解析到VIP地址

[root@swarm-node1 ~]# docker ps
CONTAINER ID        IMAGE               COMMAND                  CREATED             STATUS              PORTS               NAMES
d66e400533af        nginx:latest        "nginx -g 'daemon ..."   18 minutes ago      Up 18 minutes       80/tcp              nginx-vip.2.p52ud6cmmgonl236dtvhuzibk
# docker exec -it d66e400533af sh
root@d66e400533af:/# nslookup nginx-vip
Server:         127.0.0.11
Address:        127.0.0.11#53

Non-authoritative answer:
Name:   nginx-vip
Address: 192.168.10.2

root@d66e400533af:/# nslookup tasks.nginx-vip
Server:         127.0.0.11
Address:        127.0.0.11#53

Non-authoritative answer:
Name:   tasks.nginx-vip
Address: 192.168.10.4
Name:   tasks.nginx-vip
Address: 192.168.10.3

Service IP：192.168.10.2在iptables的mangle表的OUTPUT链中被标记为0x112（274），IPVS经过该标记将Service IP转发到192.168.10.3和192.168.10.4的容器

[root@swarm-node1 ~]# docker inspect -f {{.NetworkSettings.SandboxKey}} d66e400533af
/var/run/docker/netns/18531514ffd0
[root@swarm-node1 ~]# nsenter --net=/var/run/docker/netns/18531514ffd0 sh
sh-4.2# ip add show eth0
102: eth0@if103: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP 
link/ether 02:42:c0:a8:0a:04 brd ff:ff:ff:ff:ff:ff link-netnsid 0
inet 192.168.10.4/24 scope global eth0
   valid_lft forever preferred_lft forever
inet 192.168.10.2/32 scope global eth0
   valid_lft forever preferred_lft forever
sh-4.2# ip route
default via 172.18.0.1 dev eth1 
172.18.0.0/16 dev eth1 proto kernel scope link src 172.18.0.3 
192.168.10.0/24 dev eth0 proto kernel scope link src 192.168.10.4 
sh-4.2# iptables -t mangle -S
-P PREROUTING ACCEPT
-P INPUT ACCEPT
-P FORWARD ACCEPT
-P OUTPUT ACCEPT
-P POSTROUTING ACCEPT
-A OUTPUT -d 192.168.10.2/32 -j MARK --set-xmark 0x112/0xffffffff
sh-4.2# ipvsadm
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port           Forward Weight ActiveConn InActConn
FWM  274 rr
-> 192.168.10.3:0                Masq    1      0          0
-> 192.168.10.4:0                Masq    1      0          0

VIP模式中，swarm mode为容器分配了一个链接到overlay网络（my-network）的网卡"eth0@if103"且生成了VIP，同时也分配了一个链接到docker_gwbridge网络的网卡"eth1@if105"用于链接外部网络。
全部链接到my-network网络中的容器能够经过service-name或者VIP来访问service，经过service-name访问时，将先经过内置的DNS服务解析获取到VIP。负载均衡

dnsrr（DNS round-robin）

建立一个链接到自定义的overlay网络的service，并指定endpoint-mode为dnsrrtcp

[root@swarm-manager ~]# docker service create --endpoint-mode dnsrr --replicas 2 --name nginx-dnsrr --network my-network nginx
[root@swarm-manager ~]# docker service inspect -f {{.Spec.EndpointSpec.Mode}} nginx-dnsrr
dnsrr
[root@swarm-node1 ~]# docker inspect -f {{.NetworkSettings.SandboxKey}} b68f0b4465b4
/var/run/docker/netns/b4efcf686a74
[root@swarm-node1 ~]# ls -lrt /var/run/docker/netns/
total 0
-r--r--r--. 1 root root 0 Aug  6 11:24 ingress_sbox
-r--r--r--. 1 root root 0 Aug  6 11:24 1-i6xug49nwd
-r--r--r--. 1 root root 0 Aug 10 16:30 1-vxe1cwk14a
-r--r--r--. 1 root root 0 Aug 10 16:30 18531514ffd0
-r--r--r--. 1 root root 0 Aug 10 18:08 b4efcf686a74
[root@swarm-node1 ~]# docker ps
CONTAINER ID        IMAGE               COMMAND                  CREATED              STATUS              PORTS               NAMES
b68f0b4465b4        nginx:latest        "nginx -g 'daemon ..."   About a minute ago   Up About a minute   80/tcp              nginx-dnsrr.2.kpx2tqqmdugdpwpdwynbyzf9j
d66e400533af        nginx:latest        "nginx -g 'daemon ..."   2 hours ago          Up 2 hours          80/tcp              nginx-vip.2.p52ud6cmmgonl236dtvhuzibk

service-name将经过内置的DNS解析到每一个容器的overlay网络地址

[root@swarm-node1 ~]# docker exec -it b68f0b4465b4 bash
root@b68f0b4465b4:/# nslookup nginx-dnsrr
Server:         127.0.0.11
Address:        127.0.0.11#53

Non-authoritative answer:
Name:   my-nginx-dnsrr
Address: 192.168.10.5
Name:   my-nginx-dnsrr
Address: 192.168.10.6
root@b68f0b4465b4:/# ifconfig
eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        inet 192.168.10.6  netmask 255.255.255.0  broadcast 0.0.0.0
        ether 02:42:c0:a8:0a:06  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

eth1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.18.0.4  netmask 255.255.0.0  broadcast 0.0.0.0
        ether 02:42:ac:12:00:04  txqueuelen 0  (Ethernet)
        RX packets 7656  bytes 16832156 (16.0 MiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 5053  bytes 343122 (335.0 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        loop  txqueuelen 0  (Local Loopback)
        RX packets 6  bytes 788 (788.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 6  bytes 788 (788.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

ingress load balancing

建立一个链接到自定义的overlay网络的service，并将容器的80端口映射host的80端口

[root@swarm-manager ~]# docker service create --replicas 2 --name nginx-ingress --network my-network --publish 80:80 nginx
[root@swarm-manager ~]# docker service ls
ID                  NAME                MODE                REPLICAS            IMAGE               PORTS
7yzee08a9ryq        nginx-dnsrr         replicated          2/2                 nginx:latest
qx5epc99yu8q        nginx-vip           replicated          2/2                 nginx:latest
udiaexlplqq2        nginx-ingress       replicated          2/2                 nginx:latest        *:80->80/tcp
[root@swarm-manager ~]# docker service inspect -f {{.Endpoint.VirtualIPs}} nginx-ingress
[{i6xug49nwdsxauqqpli3apvym 10.255.0.5/16} {vxe1cwk14avlfp2xjgymhkhdl 192.168.10.7/24}]
[root@swarm-node1 ~]# docker ps
CONTAINER ID        IMAGE               COMMAND                  CREATED             STATUS              PORTS               NAMES
42aed469b4cc        nginx:latest        "nginx -g 'daemon ..."   About an hour ago   Up About an hour    80/tcp              nginx-ingress.2.u00j8ich8cv4bo0grx1hbsr7q
b68f0b4465b4        nginx:latest        "nginx -g 'daemon ..."   About an hour ago   Up About an hour    80/tcp              nginx-dnsrr.2.kpx2tqqmdugdpwpdwynbyzf9j
d66e400533af        nginx:latest        "nginx -g 'daemon ..."   3 hours ago         Up 3 hours          80/tcp              nginx-vip.2.p52ud6cmmgonl236dtvhuzibk
[root@swarm-node1 ~]# docker exec -it 42aed469b4cc bash
root@42aed469b4cc:/# ifconfig
eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        inet 10.255.0.7  netmask 255.255.0.0  broadcast 0.0.0.0
        ether 02:42:0a:ff:00:07  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

eth1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.18.0.5  netmask 255.255.0.0  broadcast 0.0.0.0
        ether 02:42:ac:12:00:05  txqueuelen 0  (Ethernet)
        RX packets 4049  bytes 10745529 (10.2 MiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 3164  bytes 211730 (206.7 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

eth2: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1450
        inet 192.168.10.9  netmask 255.255.255.0  broadcast 0.0.0.0
        ether 02:42:c0:a8:0a:09  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        loop  txqueuelen 0  (Local Loopback)
        RX packets 4  bytes 620 (620.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 4  bytes 620 (620.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0
root@42aed469b4cc:/# ip add show eth0
110: eth0@if111: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default 
    link/ether 02:42:0a:ff:00:07 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 10.255.0.7/16 scope global eth0
       valid_lft forever preferred_lft forever
    inet 10.255.0.5/32 scope global eth0
       valid_lft forever preferred_lft forever
       root@42aed469b4cc:/# ip add show eth2
root@42aed469b4cc:/# ip add show eth2
114: eth2@if115: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default 
    link/ether 02:42:c0:a8:0a:09 brd ff:ff:ff:ff:ff:ff link-netnsid 2
    inet 192.168.10.9/24 scope global eth2
       valid_lft forever preferred_lft forever
    inet 192.168.10.7/32 scope global eth2
       valid_lft forever preferred_lft forever

容器中除了lo，还建立了3个网卡：ide

eth0：链接到ingress的网络，经过routing mesh提供外部服务访问。
eth1：当service设置了端口映射时，swarm会为Service中的每一个容器另外分配一块网卡，链接到docker_gwbridge网络。当容器内部主动往外发送数据时，由docker_gwbridge SNAT转发至外部网络。
eth2：链接到my-network的overlay网络。

routing mesh

若是在建立Service时映射了端口，swarm mode将会经过routing mesh在全部节点上监听80端口，即便节点上未建立相应的容器，并经过iptables作反向NAT，当客户端访问集群中的任意节点的80端口，swarm负载均衡会将请求路由到一个活动的容器。若不想在未建立容器的节点上监听published端口，则可在映射端口时经过--publish mode=host,target=80,published=8080进行指定。oop

建立完Service以后会发现Virtual IPs有两个入口，其中10.255.0.5链接到ingress网络，192.168.10.7是链接自定义的my-network网络。当外部客户端访问服务时，swarm负载均衡的流程以下:

一、用户访问swarm-node1上的Nginx服务（172.16.100.21:80）
二、iptables将根据-A DOCKER-INGRESS -p tcp -m tcp --dport 80 -j DNAT --to-destination 172.18.0.2:80的规则将请求转发至ingress sanbox中的172.18.0.2:80。

[root@swarm-node1 ~]# iptables -t nat -S
-P PREROUTING ACCEPT
-P INPUT ACCEPT
-P OUTPUT ACCEPT
-P POSTROUTING ACCEPT
-N DOCKER
-N DOCKER-INGRESS
-A PREROUTING -m addrtype --dst-type LOCAL -j DOCKER-INGRESS
-A PREROUTING -m addrtype --dst-type LOCAL -j DOCKER
-A OUTPUT -m addrtype --dst-type LOCAL -j DOCKER-INGRESS
-A OUTPUT ! -d 127.0.0.0/8 -m addrtype --dst-type LOCAL -j DOCKER
-A POSTROUTING -o docker_gwbridge -m addrtype --src-type LOCAL -j MASQUERADE
-A POSTROUTING -s 172.17.0.0/16 ! -o docker0 -j MASQUERADE
-A POSTROUTING -s 172.18.0.0/16 ! -o docker_gwbridge -j MASQUERADE
-A DOCKER -i docker0 -j RETURN
-A DOCKER -i docker_gwbridge -j RETURN
-A DOCKER-INGRESS -p tcp -m tcp --dport 80 -j DNAT --to-destination 172.18.0.2:80
-A DOCKER-INGRESS -j RETURN

三、ingress sanbox中的iptable根据不一样的端口设置不一样的mark（0x114/276）。

[root@swarm-node1 ~]# nsenter --net=/var/run/docker/netns/ingress_sbox iptables -t mangle -S
[root@swarm-node1 ~]# iptables -t mangle -S
-P PREROUTING ACCEPT
-P INPUT ACCEPT
-P FORWARD ACCEPT
-P OUTPUT ACCEPT
-P POSTROUTING ACCEPT
-A PREROUTING -p tcp -m tcp --dport 80 -j MARK --set-xmark 0x114/0xffffffff
-A OUTPUT -d 10.255.0.5/32 -j MARK --set-xmark 0x114/0xffffffff

四、ipvs将根据不一样的mark转发到对应的real server（容器namespace）；

[root@swarm-node1 ~]# nsenter --net=/var/run/docker/netns/ingress_sbox ipvsadm
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
FWM  276 rr
  -> 10.255.0.6:0                 Masq    1      0          0
  -> 10.255.0.7:0                 Masq    1      0          0