SONiC Wiki中关于SONiC与P4的联动仿真实验(https://github.com/Azure/SONiC/wiki/SONiC-P4-Software-Switch)
该实验对了解SONiC和P4都有一定的帮助，之前一直用Barefoot原生Switchd和Stratum，所以借助这个实验来进一步学习一下SONiC，后面有机会再尝试用真实的P4设备跑SONiC对比一下。
SONiC-P4是一个软件交换机，其运行在P4模拟SAI行为模型软件交换机ASCI之上，SONiC-P4以docker容器部署，本实验也会以docker容器的方式运行SONiC-P4作为软件交换机。
网络拓扑
- switch1、switch2、host1、host2都以docker容器运行；
- 各容器的连接通过ovs中网桥连通
运行
- 下载仿真需要的文件sonic-p4-test.zip,解压出来，里面包含容器运行需要的脚本及配置文件等；
-
运行
./install_docker_ovs.sh安装docker和ovs，若系统已安装则忽略这步； -
原教程中使用
./load_image.sh来从sonic jenkins上下载sonic-p4对应docker镜像已经不能使用，这里我发现docker官仓中有这个镜像就直接从docker仓库下载了，运行docker pull alihasanahmedkhan/docker-sonic-p4下载镜像，若本地没有ubuntu 14.04镜像，运行docker pull ubuntu:14.04下载镜像； -
运行
./start.sh运行容器并连接网络拓扑，成功后可看到类似如下容器已启动：xxx@ubuntu:~$ docker ps CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES 2d8f8272ba4f ubuntu:14.04 "/bin/bash" 23 hours ago Up 23 hours host2 dc30fe84ac1e ubuntu:14.04 "/bin/bash" 23 hours ago Up 23 hours host1 4fe94a1879fd alihasanahmedkhan/docker-sonic-p4:latest "/bin/bash" 23 hours ago Up 23 hours switch2 9acb9dfc3696 alihasanahmedkhan/docker-sonic-p4:latest "/bin/bash" 23 hours ago Up 23 hours switch1两个ubuntu 14.04为host1和host2，两个alihasanahmedkhan/docker-sonic-p4为switch1和switch2
-
60s左右后，运行
./test.sh测试host1和host2的连通性，可得到如下结果：xxx@ubuntu:~/Downloads/sonic/p4-test$ ./test.sh [sudo] password for hui: PING 192.168.2.1 (192.168.2.1) 56(84) bytes of data. 64 bytes from 192.168.2.1: icmp_seq=1 ttl=64 time=8.54 ms --- 192.168.2.1 ping statistics --- 1 packets transmitted, 1 received, 0% packet loss, time 0ms rtt min/avg/max/mdev = 8.545/8.545/8.545/0.000 ms PING 192.168.2.2 (192.168.2.2) 56(84) bytes of data. 64 bytes from 192.168.2.2: icmp_seq=1 ttl=62 time=7.17 ms 64 bytes from 192.168.2.2: icmp_seq=2 ttl=62 time=33.1 ms 64 bytes from 192.168.2.2: icmp_seq=3 ttl=62 time=33.2 ms 64 bytes from 192.168.2.2: icmp_seq=4 ttl=62 time=32.5 ms 64 bytes from 192.168.2.2: icmp_seq=5 ttl=62 time=33.1 ms 64 bytes from 192.168.2.2: icmp_seq=6 ttl=62 time=35.4 ms - 运行
./stop.sh可结束仿真
实验详解
配置
switch1和switch2中相关配置都在解压后对应的switch1/etc和switch2/etc目录下，这些配置文件会映射到容器中，容器中各种SONiC进程运行过程中会使用到。
以switch1为例，容器启动过程start.sh中将配置及脚本通过docker数据卷-v参数把挂载到容器中:
sudo docker run --net=none --privileged --entrypoint /bin/bash --name switch1 -it -d -v $PWD/switch1:/sonic alihasanahmedkhan/docker-sonic-p4:latest
主要涉及如下配置：
- vlan配置，switch1/etc/config_db/vlan_config.json会生成到config_db.json配置到config_db
switch1/etc/config_db/vlan_config.json中vlan的配置:
{ "VLAN": { "Vlan15": { "members": [ "Ethernet0" ], "vlanid": "15" }, "Vlan10": { "members": [ "Ethernet1" ], "vlanid": "10" } }, "VLAN_MEMBER": { "Vlan15|Ethernet0": { "tagging_mode": "untagged" }, "Vlan10|Ethernet1": { "tagging_mode": "untagged" } }, "VLAN_INTERFACE": { "Vlan15|10.0.0.0/31": {}, "Vlan10|192.168.1.1/24": {} } }switch1/scripts/startup.sh中vlan配置的拷贝:
if [ -f /etc/sonic/config_db.json ]; then sonic-cfggen -j /etc/sonic/config_db.json -j /sonic/scripts/vlan_config.json --print-data > /tmp/config_db.json mv /tmp/config_db.json /etc/sonic/config_db.json else sonic-cfggen -j /sonic/etc/config_db/vlan_config.json --print-data > /etc/sonic/config_db.json fi - quagga中bgp配置
switch1/etc/quagga/bgpd.conf中bgp配置定义本端AS、bgp邻居:
hostname bgpd password zebra enable password zebra log file /var/log/quagga/bgpd.log ! router bgp 10001 bgp router-id 192.168.1.1 network 192.168.1.0 mask 255.255.255.0 neighbor 10.0.0.1 remote-as 10002 neighbor 10.0.0.1 timers 1 3 neighbor 10.0.0.1 send-community neighbor 10.0.0.1 allowas-in maximum-paths 64 ! access-list all permit anyswitch1/scripts/startup.sh中quagga配置的拷贝:
cp -rf /sonic/etc/quagga /etc/
运行机制
同样以switch1来分析一下SONiC-P4的运行机制，交换芯片自然是以支持P4的ASIC作为底层模型，交换系统以SONiC架构为网络OS，在SONiC中通常多个容器实现整个网络OS，本实验中一个容器就启动了switch，所以所有的SONiC框架组件都集成在这个容器中运行，这里从容器中进程到P4底层来一步步分析SONiC-P4的运行机制。
先说明一下host1，host2，switch1，switch2都是通过ovs中网桥连接起来的，具体都在start.sh中ovs命令实现：
sudo ovs-vsctl add-br switch1_switch2
sudo ovs-docker add-port switch1_switch2 sw_port0 switch1
sudo ovs-docker add-port switch1_switch2 sw_port0 switch2
sudo ovs-vsctl add-br host1_switch1
sudo ovs-docker add-port host1_switch1 sw_port1 switch1
sudo ovs-docker add-port host1_switch1 eth1 host1
sudo ovs-vsctl add-br host2_switch2
sudo ovs-docker add-port host2_switch2 sw_port1 switch2
sudo ovs-docker add-port host2_switch2 eth1 host2
xxx@ubuntu:~/Downloads/sonic/p4-test$ sudo ovs-vsctl show
[sudo] password for hui:
71e743dd-96df-4419-bfcf-c2b9fcdc582e
Bridge "switch1_switch2"
Port "2510a4f0883c4_l"
Interface "2510a4f0883c4_l"
Port "switch1_switch2"
Interface "switch1_switch2"
type: internal
Port "b993f47d0d894_l"
Interface "b993f47d0d894_l"
Bridge "host2_switch2"
Port "host2_switch2"
Interface "host2_switch2"
type: internal
Port "fdfbe4e3a0944_l"
Interface "fdfbe4e3a0944_l"
Port "ddccda4c0d1d4_l"
Interface "ddccda4c0d1d4_l"
Bridge "host1_switch1"
Port "b07b5c691a2d4_l"
Interface "b07b5c691a2d4_l"
Port "host1_switch1"
Interface "host1_switch1"
type: internal
Port "7c76ce7233714_l"
Interface "7c76ce7233714_l"
ovs_version: "2.12.0"
这里以switch1容器中运行的进程分析，switch2同样会存在对应情况。
查询进程列表及端口占用可得到类似如下结果:
root@9acb9dfc3696:/# ps aux
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
root 1 0.0 0.0 20248 2984 pts/0 Ss+ Jan27 0:00 /bin/bash
root 59 0.0 0.0 49644 18072 ? Ss Jan27 2:06 /usr/bin/python /usr/bin/supervisord
root 76 0.0 0.0 262900 2764 ? Sl Jan27 0:01 /usr/sbin/rsyslogd -n
root 82 0.1 0.0 43364 4468 ? Sl Jan27 3:25 /usr/bin/redis-server 127.0.0.1:6379
root 323 1.5 0.6 1010980 203952 ? Sl Jan27 44:54 simple_switch -i 0@sw_port0 -i 1@sw_port1 -i 2@sw_port2 -i 3@sw_port3 -i 4@sw_port4 -i 5@sw_port5 -i 6@sw_port6 -i 7@sw_port7 -i 7@sw_port7 -i 8@sw
root 327 0.6 0.0 694416 26336 ? Sl Jan27 18:22 simple_switch -i 0@router_port1 -i 250@router_cpu_port --thrift-port 9091 --log-file bm_logs/router_log --log-flush --notifications-addr ipc:///tmp
root 366 0.0 0.0 294868 6320 ? Sl Jan27 2:34 /usr/bin/syncd -uN
root 373 0.0 0.0 182064 4888 ? Sl Jan27 0:22 /usr/bin/orchagent -d /var/log/swss -b 8192 -m 00:01:04:4c:49:f5
root 382 3.3 0.0 99776 3180 ? Sl Jan27 98:16 /usr/bin/portsyncd -p /port_config.ini
root 385 0.0 0.0 99844 3160 ? Sl Jan27 0:00 /usr/bin/intfsyncd
root 388 0.0 0.0 99844 3056 ? Sl Jan27 0:00 /usr/bin/neighsyncd
root 391 0.0 0.0 108148 3072 ? Sl Jan27 0:00 /usr/bin/teamsyncd
root 394 0.0 0.0 101012 3916 ? Sl Jan27 0:00 /usr/bin/fpmsyncd
root 397 0.0 0.0 99768 3828 ? Sl Jan27 0:09 /usr/bin/intfmgrd
root 400 0.0 0.0 99820 3780 ? Sl Jan27 0:09 /usr/bin/vlanmgrd
quagga 443 0.0 0.0 26408 2728 ? S Jan27 0:01 /usr/lib/quagga/zebra -A 127.0.0.1
quagga 445 0.0 0.0 31212 4172 ? S Jan27 0:55 /usr/lib/quagga/bgpd -A 127.0.0.1 -F
root 9707 0.0 0.0 20248 3212 pts/1 Ss 03:02 0:00 bash
root 11011 0.0 0.0 20052 2844 ? S 09:08 0:00 bash -c /usr/bin/arp_update; sleep 300
root 11028 0.0 0.0 4240 680 ? S 09:08 0:00 sleep 300
root@9acb9dfc3696:/# netstat -ntpl
Active Internet connections (only servers)
Proto Recv-Q Send-Q Local Address Foreign Address State PID/Program name
tcp 0 0 0.0.0.0:179 0.0.0.0:* LISTEN 445/bgpd
tcp 0 0 127.0.0.1:2620 0.0.0.0:* LISTEN 394/fpmsyncd
tcp 0 0 127.0.0.1:2601 0.0.0.0:* LISTEN 443/zebra
tcp 0 0 127.0.0.1:6379 0.0.0.0:* LISTEN 82/redis-server 127
tcp 0 0 127.0.0.1:2605 0.0.0.0:* LISTEN 445/bgpd
tcp6 0 0 :::179 :::* LISTEN 445/bgpd
tcp6 0 0 :::9090 :::* LISTEN 323/simple_switch
tcp6 0 0 :::9091 :::* LISTEN 327/simple_switch
SONiC-P4容器中supervisord会用于进程管理，SONiC相关进程及P4相关进程都会由supervisord启动，/usr/bin/start.sh中定义了supervisord管理的进程，我们可看到SONiC及P4的相关几个进程。
# P4进程
supervisorctl start bm_bridge
supervisorctl start bm_router
# SONiC进程
echo "Start syncd"
supervisorctl start syncd
echo "Start orchagent"
supervisorctl start orchagent
echo "Start portsyncd"
supervisorctl start portsyncd
echo "Start intfsyncd"
supervisorctl start intfsyncd
echo "Start neighsyncd"
supervisorctl start neighsyncd
echo "Start teamsyncd"
supervisorctl start teamsyncd
echo "Start fpmsyncd"
supervisorctl start fpmsyncd
echo "Start intfmgrd"
supervisorctl start intfmgrd
echo "Start vlanmgrd"
supervisorctl start vlanmgrd
echo "Start zebra"
supervisorctl start zebra
echo "Start bgpd"
supervisorctl start bgpd
if [ -f /etc/swss/config.d/default_config.json ]; then
swssconfig /etc/swss/config.d/default_config.json
fi
# Start arp_update when VLAN exists
VLAN=`sonic-cfggen -d -v 'VLAN.keys() | join(" ") if VLAN'`
if [ "$VLAN" != "" ]; then
echo "Start arp_update"
supervisorctl start arp_update
根据SONiC架构我们对应到SONiC关键进程，如上获取的进程列表我们可以发现多个容器进程合并到同一容器里运行了：
redis-server # database container的redis数据库进程
orchagent, portsyncd, intfsyncd, neighsyncd, intfmgrd, vlanmgrd, syncd # swss container对应的相关进程
teamsyncd # teamd container的teamsyncd进 #程
fpmsyncd, zebra, bgpd # bgp container对应的相关进程
P4进程：
simple_switch # P4模型虚机交换机进程，有两个进程，一个作为router，一个作为bridge
网络
实验中划分为两个AS，使用bgp路由协议，SONiC网络操作系统作为上层实现，底层P4模型模拟ASIC交换芯片，下面从这两个维度分析一下该实验网络，如下都以switch1容器举例，switch2同样对存在对应的系统及配置。
上层SONiC网络操作系统
switch bgp邻居状态：
# switch1
root@9acb9dfc3696:/# vtysh -c "show ip bgp sum"
BGP router identifier 192.168.1.1, local AS number 10001
RIB entries 3, using 336 bytes of memory
Peers 1, using 4656 bytes of memory
Neighbor V AS MsgRcvd MsgSent TblVer InQ OutQ Up/Down State/PfxRcd
10.0.0.1 4 10002 234804 234808 0 0 0 2d17h16m 1
Total number of neighbors 1
# switch2
root@4fe94a1879fd:/# vtysh -c "show ip bgp sum"
BGP router identifier 192.168.2.1, local AS number 10002
RIB entries 3, using 336 bytes of memory
Peers 1, using 4656 bytes of memory
Neighbor V AS MsgRcvd MsgSent TblVer InQ OutQ Up/Down State/PfxRcd
10.0.0.0 4 10001 235030 235031 0 0 0 2d17h20m 1
Total number of neighbors 1
switch生成路由如下：
# switch1
root@9acb9dfc3696:/# vtysh -c "show ip route"
Codes: K - kernel route, C - connected, S - static, R - RIP,
O - OSPF, I - IS-IS, B - BGP, P - PIM, A - Babel,
> - selected route, * - FIB route
C>* 10.0.0.0/31 is directly connected, Vlan15
C>* 127.0.0.0/8 is directly connected, lo
C>* 192.168.1.0/24 is directly connected, Vlan10
B>* 192.168.2.0/24 [20/0] via 10.0.0.1, Vlan15, 2d17h27m
root@9acb9dfc3696:/# ip route
10.0.0.0/31 dev Vlan15 proto kernel scope link src 10.0.0.0
192.168.1.0/24 dev Vlan10 proto kernel scope link src 192.168.1.1
192.168.2.0/24 via 10.0.0.1 dev Vlan15 proto zebra
# switch2
root@4fe94a1879fd:/# vtysh -c "show ip route"
Codes: K - kernel route, C - connected, S - static, R - RIP,
O - OSPF, I - IS-IS, B - BGP, P - PIM, A - Babel,
> - selected route, * - FIB route
C>* 10.0.0.0/31 is directly connected, Vlan14
C>* 127.0.0.0/8 is directly connected, lo
B>* 192.168.1.0/24 [20/0] via 10.0.0.0, Vlan14, 2d17h26m
C>* 192.168.2.0/24 is directly connected, Vlan9
root@4fe94a1879fd:/# ip route
10.0.0.0/31 dev Vlan14 proto kernel scope link src 10.0.0.1
192.168.1.0/24 via 10.0.0.0 dev Vlan14 proto zebra
192.168.2.0/24 dev Vlan9 proto kernel scope link src 192.168.2.1
switch vlan：
# switch1
root@9acb9dfc3696:/# show vlan config
Name VID Member Mode
------ ----- --------- --------
Vlan10 10 Ethernet1 untagged
Vlan15 15 Ethernet0 untagged
# switch2
root@4fe94a1879fd:/# show vlan config
Name VID Member Mode
------ ----- --------- --------
Vlan9 9 Ethernet1 untagged
Vlan14 14 Ethernet0 untagged
switch vlan interface：
# switch1
root@9acb9dfc3696:/# cat /proc/net/vlan/config
VLAN Dev name | VLAN ID
Name-Type: VLAN_NAME_TYPE_RAW_PLUS_VID_NO_PAD
Vlan10 | 10 | Bridge
Vlan15 | 15 | Bridge
root@9acb9dfc3696:/# ifconfig Vlan10
Vlan10 Link encap:Ethernet HWaddr 00:01:04:4c:49:f5
inet addr:192.168.1.1 Bcast:0.0.0.0 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:780 errors:0 dropped:0 overruns:0 frame:0
TX packets:780 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:21840 (21.3 KiB) TX bytes:32760 (31.9 KiB)
root@9acb9dfc3696:/# ifconfig Vlan15
Vlan15 Link encap:Ethernet HWaddr 00:01:04:4c:49:f5
inet addr:10.0.0.0 Bcast:0.0.0.0 Mask:255.255.255.254
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:471533 errors:0 dropped:0 overruns:0 frame:0
TX packets:471529 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:35490404 (33.8 MiB) TX bytes:42091665 (40.1 MiB)
# switch2
root@4fe94a1879fd:/# cat /proc/net/vlan/config
VLAN Dev name | VLAN ID
Name-Type: VLAN_NAME_TYPE_RAW_PLUS_VID_NO_PAD
Vlan14 | 14 | Bridge
Vlan9 | 9 | Bridge
root@4fe94a1879fd:/# ifconfig Vlan9
Vlan9 Link encap:Ethernet HWaddr 00:01:04:4c:49:f6
inet addr:192.168.2.1 Bcast:0.0.0.0 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:875 errors:0 dropped:0 overruns:0 frame:0
TX packets:2443 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:29484 (28.7 KiB) TX bytes:107590 (105.0 KiB)
root@4fe94a1879fd:/# ifconfig Vlan14
Vlan14 Link encap:Ethernet HWaddr 00:01:04:4c:49:f6
inet addr:10.0.0.1 Bcast:0.0.0.0 Mask:255.255.255.254
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:471765 errors:0 dropped:0 overruns:0 frame:0
TX packets:471769 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:35508077 (33.8 MiB) TX bytes:42112988 (40.1 MiB)
SONiC Redis数据库中数据：
root@9acb9dfc3696:/# redis-cli
127.0.0.1:6379> keys VLAN*
1) "VLAN_MEMBER_TABLE:Vlan10:Ethernet1"
2) "VLAN_TABLE:Vlan15"
3) "VLAN_TABLE:Vlan10"
4) "VLAN_MEMBER_TABLE:Vlan15:Ethernet0"
127.0.0.1:6379> keys ROUTE_TABLE*
1) "ROUTE_TABLE:192.168.1.0/24"
2) "ROUTE_TABLE:192.168.2.0/24"
3) "ROUTE_TABLE:10.0.0.0/31"
4) "ROUTE_TABLE:127.0.0.0/8"
127.0.0.1:6379> hgetall "VLAN_TABLE:Vlan10"
1) "admin_status"
2) "up"
127.0.0.1:6379> hgetall "VLAN_MEMBER_TABLE:Vlan10:Ethernet1"
1) "tagging_mode"
2) "untagged"
127.0.0.1:6379> hgetall "ROUTE_TABLE:192.168.1.0/24"
1) "nexthop"
2) ""
3) "ifname"
4) "Vlan10"
127.0.0.1:6379> hgetall "ROUTE_TABLE:192.168.2.0/24"
1) "nexthop"
2) "10.0.0.1"
3) "ifname"
4) "Vlan15"
127.0.0.1:6379> hgetall "ROUTE_TABLE:10.0.0.0/31"
1) "nexthop"
2) ""
3) "ifname"
4) "Vlan15"
127.0.0.1:6379>
底层P4模型交换芯片
运行机制中我们已经分析P4中分为两个进程，一个作为router，一个作为bridge，我们这里可以通过P4模型Thrift接口提供的Cli来查看一下交换芯片中转发表项进行分析。
通过/etc/supervisor/conf.d/supervisord.conf中bridge和router的Thrift接口监听端口分别为默认9090和设定的9091
[program:bm_router]
command=ip netns exec sw_net simple_switch -i 0@router_port1 -i 250@router_cpu_port --thrift-port 9091 --log-file bm_logs/router_log --log-flush --notifications-addr ipc:///tmp/bmv2-router-notifications.ipc /usr/share/p4-sai-bm/sai_router.json
priority=3
autostart=false
autorestart=false
stdout_logfile=syslog
stderr_logfile=syslog
[program:bm_bridge]
command=ip netns exec sw_net simple_switch -i 0@sw_port0 -i 1@sw_port1 -i 2@sw_port2 -i 3@sw_port3 -i 4@sw_port4 -i 5@sw_port5 -i 6@sw_port6 -i 7@sw_port7 -i 7@sw_port7 -i 8@sw_port8 -i 9@sw_port9 -i 10@sw_port10 -i 11@sw_port11 -i 12@sw_port12 -i 13@sw_port13 -i 14@sw_port14 -i 15@sw_port15 -i 16@sw_port16 -i 17@sw_port17 -i 18@sw_port18 -i 19@sw_port19 -i 20@sw_port20 -i 21@sw_port21 -i 22@sw_port22 -i 23@sw_port23 -i 24@sw_port24 -i 25@sw_port25 -i 26@sw_port26 -i 27@sw_port27 -i 28@sw_port28 -i 29@sw_port29 -i 30@sw_port30 -i 31@sw_port31 -i 250@cpu_port -i 251@router_port0 --log-file bm_logs/bridge_log --log-flush /usr/share/p4-sai-bm/sai_bridge.json &
priority=3
autostart=false
autorestart=false
stdout_logfile=syslog
stderr_logfile=syslog
通过Cli连接9090端口查看bridge中vlan相关转发表项：
root@9acb9dfc3696:/# simple_switch_CLI --thrift-port 9090
Obtaining JSON from switch...
Done
Control utility for runtime P4 table manipulation
RuntimeCmd: show_tables
table_bridge_id_1d [implementation=None, mk=ingress_metadata.bridge_port(exact, 8)]
table_bridge_id_1q [implementation=None, mk=ingress_metadata.vid(exact, 12)]
table_bridge_loopback_filter [implementation=None, mk=egress_metadata.bridge_port(exact, 8)]
table_broadcast [implementation=None, mk=ingress_metadata.bridge_id(exact, 12)]
table_cpu_forward [implementation=None, mk=cpu_header_valid(valid, 1)]
table_drop_tagged_internal [implementation=None, mk=ingress_metadata.drop_tagged(exact, 1)]
table_drop_untagged_internal [implementation=None, mk=ingress_metadata.drop_untagged(exact, 1)]
table_egress_br_port_to_if [implementation=None, mk=egress_metadata.bridge_port(exact, 8)]
table_egress_clone_internal [implementation=None, mk=standard_metadata.instance_type(exact, 32)]
table_egress_lag [implementation=None, mk=egress_metadata.out_if(exact, 8), egress_metadata.hash_val(exact, 6)]
table_egress_set_vlan [implementation=None, mk=egress_metadata.bridge_port(exact, 8)]
table_egress_vbridge_STP [implementation=None, mk=egress_metadata.bridge_port(exact, 8)]
table_egress_vlan_filtering [implementation=None, mk=egress_metadata.bridge_port(exact, 8), ingress_metadata.vid(exact, 12)]
table_egress_vlan_tag [implementation=None, mk=egress_metadata.out_if(exact, 8), ingress_metadata.vid(exact, 12), vlan_valid(valid, 1)]
table_egress_xSTP [implementation=None, mk=egress_metadata.bridge_port(exact, 8), ingress_metadata.stp_id(exact, 3)]
table_fdb [implementation=None, mk=ethernet.dstAddr(exact, 48), ingress_metadata.bridge_id(exact, 12)]
table_flood [implementation=None, mk=ingress_metadata.bridge_id(exact, 12)]
table_hostif_vlan_tag [implementation=None, mk=cpu_header.dst(exact, 16), vlan_valid(valid, 1)]
table_ingress_lag [implementation=None, mk=standard_metadata.ingress_port(exact, 9)]
table_ingress_vlan_filtering [implementation=None, mk=ingress_metadata.bridge_port(exact, 8), ingress_metadata.vid(exact, 12)]
table_l2_trap [implementation=None, mk=ethernet.dstAddr(exact, 48)]
table_l3_interface [implementation=None, mk=ethernet.dstAddr(exact, 48), ingress_metadata.bridge_id(exact, 12)]
table_lag_hash [implementation=None, mk=egress_metadata.out_if(exact, 8)]
table_learn_fdb [implementation=None, mk=ethernet.srcAddr(exact, 48), ingress_metadata.bridge_id(exact, 12)]
table_mc_fdb [implementation=None, mk=ethernet.dstAddr(exact, 48), ingress_metadata.bridge_id(exact, 12)]
table_mc_l2_sg_g [implementation=None, mk=ingress_metadata.bridge_id(exact, 12), ipv4.srcAddr(exact, 32), ipv4.dstAddr(exact, 32)]
table_mc_lookup_mode [implementation=None, mk=ingress_metadata.vid(exact, 12)]
table_port_configurations [implementation=None, mk=ingress_metadata.l2_if(exact, 8)]
table_port_ingress_interface_type [implementation=None, mk=ingress_metadata.l2_if(exact, 8)]
table_port_set_packet_vid_internal [implementation=None, mk=ingress_metadata.is_tagged(exact, 1)]
table_subport_ingress_interface_type [implementation=None, mk=ingress_metadata.l2_if(exact, 8), ingress_metadata.vid(exact, 12)]
table_trap_id [implementation=None, mk=ingress_metadata.trap_id(exact, 11)]
table_unknown_multicast_ipv4 [implementation=None, mk=ingress_metadata.bridge_id(exact, 12)]
table_unknown_multicast_nonip [implementation=None, mk=ingress_metadata.bridge_id(exact, 12)]
table_vbridge_STP [implementation=None, mk=ingress_metadata.bridge_port(exact, 8)]
table_xSTP [implementation=None, mk=ingress_metadata.bridge_port(exact, 8), ingress_metadata.stp_id(exact, 3)]
table_xSTP_instance [implementation=None, mk=ingress_metadata.vid(exact, 12)]
RuntimeCmd: table_dump table_egress_vlan_tag
==========
TABLE ENTRIES
**********
Dumping entry 0x1000000
Match key:
* egress_metadata.out_if: EXACT 01
* ingress_metadata.vid : EXACT 000a
* vlan_valid : VALID 01
Action entry: action_forward_vlan_untag -
**********
Dumping entry 0x1000001
Match key:
* egress_metadata.out_if: EXACT 00
* ingress_metadata.vid : EXACT 000f
* vlan_valid : VALID 01
Action entry: action_forward_vlan_untag -
**********
Dumping entry 0x1000002
Match key:
* egress_metadata.out_if: EXACT fb
* ingress_metadata.vid : EXACT 000a
* vlan_valid : VALID 00
Action entry: action_forward_vlan_tag - 00, 00, 0a
**********
Dumping entry 0x1000003
Match key:
* egress_metadata.out_if: EXACT fb
* ingress_metadata.vid : EXACT 000f
* vlan_valid : VALID 00
Action entry: action_forward_vlan_tag - 00, 00, 0f
**********
Dumping entry 0x20
Match key:
* egress_metadata.out_if: EXACT fb
* ingress_metadata.vid : EXACT 0001
* vlan_valid : VALID 00
Action entry: action_forward_vlan_tag - 00, 00, 01
==========
Dumping default entry
EMPTY
==========
通过Cli连接9091端口查看router中bgp生成的路由转发表项：
root@9acb9dfc3696:/# simple_switch_CLI --thrift-port 9091
Obtaining JSON from switch...
Done
Control utility for runtime P4 table manipulation
RuntimeCmd: show_tables
table_cpu_forward [implementation=None, mk=cpu_header_valid(valid, 1)]
table_egress_L3_if [implementation=None, mk=router_metadata.egress_rif(exact, 8)]
table_egress_clone_internal [implementation=None, mk=standard_metadata.instance_type(exact, 32)]
table_ingress_l3_if [implementation=None, mk=standard_metadata.ingress_port(exact, 9), vlan.vid(exact, 12)]
table_ingress_vrf [implementation=None, mk=router_metadata.ingress_rif(exact, 8)]
table_ip2me_trap [implementation=None, mk=l4_metadata.srcPort(exact, 16), l4_metadata.dstPort(exact, 16), ipv4.protocol(exact, 8)]
table_l3_trap_id [implementation=None, mk=ingress_metadata.trap_id(exact, 11)]
table_neighbor [implementation=None, mk=router_metadata.egress_rif(exact, 8), router_metadata.next_hop_dst_ip(exact, 32)]
table_next_hop [implementation=None, mk=router_metadata.next_hop_id(exact, 8)]
table_next_hop_group [implementation=None, mk=router_metadata.next_hop_group_id(exact, 3)]
table_pre_l3_trap [implementation=None, mk=vlan.etherType(ternary, 16), ipv4.dstAddr(lpm, 32), arp_ipv4.opcode(ternary, 16)]
table_router [implementation=None, mk=router_metadata.ingress_vrf(exact, 8), ipv4.dstAddr(lpm, 32)]
table_ttl [implementation=None, mk=ipv4.ttl(exact, 8)]
RuntimeCmd: table_dump table_router
==========
TABLE ENTRIES
**********
Dumping entry 0x0
Match key:
* router_metadata.ingress_vrf: EXACT 00
* ipv4.dstAddr : LPM 00000000/0
Action entry: _drop -
**********
Dumping entry 0x1
Match key:
* router_metadata.ingress_vrf: EXACT 00
* ipv4.dstAddr : LPM c0a80100/24
Action entry: action_set_erif_set_nh_dstip_from_pkt - 02
**********
Dumping entry 0x2
Match key:
* router_metadata.ingress_vrf: EXACT 00
* ipv4.dstAddr : LPM c0a80101/32
Action entry: action_set_ip2me -
**********
Dumping entry 0x3
Match key:
* router_metadata.ingress_vrf: EXACT 00
* ipv4.dstAddr : LPM 0a000000/31
Action entry: action_set_erif_set_nh_dstip_from_pkt - 03
**********
Dumping entry 0x4
Match key:
* router_metadata.ingress_vrf: EXACT 00
* ipv4.dstAddr : LPM 0a000000/32
Action entry: action_set_ip2me -
**********
Dumping entry 0x5
Match key:
* router_metadata.ingress_vrf: EXACT 00
* ipv4.dstAddr : LPM c0a80200/24
Action entry: action_set_nhop_id - 00
==========
Dumping default entry
Action entry: _drop -
==========
通过bridge和router中转发表项我们可以查到前面实验中针对vlan的配置及bgp设定生成的路由表项已经有下发对应的表项到P4模型中的ASIC芯片，具体router及bridge的P4数据面实现可通过如下P4代码实现详细了解。
https://github.com/Mellanox/SAI-P4-BM/blob/master/p4-switch/sai-p4-bm/p4src
Reference
https://github.com/Azure/SONiC/wiki/SONiC-P4-Software-Switch
https://github.com/Azure/SONiC/wiki/Architecture?spm=a2c6h.12873639.0.0.980036b0oeCGJJ
https://github.com/Mellanox/SAI-P4-BM
https://hub.docker.com/r/alihasanahmedkhan/docker-sonic-p4
https://zhuanlan.zhihu.com/p/99887221
非常不错
您好，我在进行实验时进行到运行./start.sh时，出现connect: Network is unreachable的提示，请问这种情况怎么解决，如能帮我解答，不胜感激！！！
不好意思，我的表达有误，是运行./test.sh时出现错误，更正一下。
test.sh就是进入host容器执行ping操作，具体看容器是否正常运行，并且手动进入容器内部ping一下试试。