Difference between revisions of "VPP/Configure VPP TAP Interfaces For Container Routing"
From fd.io
< VPP
(Using VPP tap interfaces for inter-container routing) |
|||
| (One intermediate revision by one other user not shown) | |||
| Line 3: | Line 3: | ||
As docker simply consumes Linux Network Namespaces, the example can easily be translated to non-docker NetNS usecases. | As docker simply consumes Linux Network Namespaces, the example can easily be translated to non-docker NetNS usecases. | ||
The setup demonstrates VPP's ability to dynamically create linux tap interfaces and can be considered an alternative to the existing 'veth' (AF_PACKET) example here: [[VPP/Configure_VPP_As_A_Router_Between_Namespaces]] | The setup demonstrates VPP's ability to dynamically create linux tap interfaces and can be considered an alternative to the existing 'veth' (AF_PACKET) example here: [[VPP/Configure_VPP_As_A_Router_Between_Namespaces]] | ||
| − | |||
| − | |||
=== Setup === | === Setup === | ||
| Line 318: | Line 316: | ||
=== Adding a new tap interface === | === Adding a new tap interface === | ||
| − | Unlike the AF_PACKET interfaces, which must be specified at VPP startup, we can create a new linux TAP interface to VPP at any point through configuration. | + | Unlike the AF_PACKET interfaces, which must be specified at VPP startup (no longer true, see 'create host-interface'), we can create a new linux TAP interface to VPP at any point through configuration. |
Still on the Vagrant SSH session, see the following example: | Still on the Vagrant SSH session, see the following example: | ||
Latest revision as of 14:43, 9 April 2020
This example shows configuration of VPP as an IPv4 router between 2 docker containers and the host.
As docker simply consumes Linux Network Namespaces, the example can easily be translated to non-docker NetNS usecases. The setup demonstrates VPP's ability to dynamically create linux tap interfaces and can be considered an alternative to the existing 'veth' (AF_PACKET) example here: VPP/Configure_VPP_As_A_Router_Between_Namespaces
Contents
Setup
The Following script, when run on the FD.io Vagrant VM, will configure and install the relevant tools needed to create the environment described in the diagram below:
To if you do not have a VPP vagrant/development VM, please see instructions here: VPP/Build,_install,_and_test_images
#!/bin/bash
exposedockernetns () {
if [ "$1" == "" ]; then
echo "usage: $0 <container_name>"
echo "Exposes the netns of a docker container to the host"
exit 1
fi
pid=`docker inspect -f '{{.State.Pid}}' $1`
ln -s /proc/$pid/ns/net /var/run/netns/$1
echo "netns of ${1} exposed as /var/run/netns/${1}"
echo "try: ip netns exec ${1} ip addr list"
return 0
}
dockerrmf () {
#Cleanup all containers on the host (dead or alive).
docker kill `docker ps --no-trunc -aq` ; docker rm `docker ps --no-trunc -aq`
}
#Vagrant build box initial setup
if [ -a /etc/apt/sources.list.d/docker.list ]
then
echo "Docker APT Sources already configured. Not setting up Docker on this Vagrant Box"
echo "Cleaning up containers from previous run..."
dockerrmf
else
mkdir /var/run/netns
sudo apt-key adv --keyserver hkp://p80.pool.sks-keyservers.net:80 --recv-keys 58118E89F3A912897C070ADBF76221572C52609D
sudo echo "deb https://apt.dockerproject.org/repo ubuntu-trusty main" > /etc/apt/sources.list.d/docker.list
sudo apt-get update
sudo apt-get install -y docker-engine tmux uuid
fi
#Cleanup any old VPP Engine config
rm -f /home/vagrant/vpe-host.conf
#Stop any existing vpp instances started via init
stop vpp
#Create VPE Engine config
cat >/home/vagrant/vpe-host.conf <<EOL
tap connect tapcontainer1
tap connect taphost
tap connect tapcontainer2
set int ip addr tap-0 192.168.1.1/24
set int ip addr tap-1 192.168.2.1/24
set int ip addr tap-2 192.168.3.1/24
set int state tap-0 up
set int state tap-1 up
set int state tap-2 up
EOL
#Remove old netns simlink
rm -Rf /var/run/netns/*
mkdir /var/run/netns
#Start VPE, use our config
vpe unix {cli-listen 0.0.0.0:5002 startup-config /home/vagrant/vpe-host.conf } dpdk {no-pci no-huge num-mbufs 8192}
#Wait for VPE to configure and bring up interfaces
sleep 5
#Add host TAP IP
ip addr add 192.168.2.2/24 dev taphost
#Create a docker container
docker pull ubuntu
docker run --name "hasvppinterface1" ubuntu sleep 30000 &
docker run --name "hasvppinterface2" ubuntu sleep 30000 &
#Wait
sleep 5
#Expose our container to the 'ip netns exec' tools
exposedockernetns hasvppinterface1
exposedockernetns hasvppinterface2
#Move the 'tapcontainer1+2 VPP linux tap interface's into container1+2's network namespace respectivley.
ip link set tapcontainer1 netns hasvppinterface1
ip link set tapcontainer2 netns hasvppinterface2
#Give our in-container TAP interface's IP addresses and bring them up. Add routes back to the host TAP's via VPP.
ip netns exec hasvppinterface1 ip addr add 192.168.1.2/24 dev tapcontainer1
ip netns exec hasvppinterface1 ip link set tapcontainer1 up
ip netns exec hasvppinterface1 ip route add 192.168.2.0/24 via 192.168.1.1
ip netns exec hasvppinterface1 ip route add 192.168.3.0/24 via 192.168.1.1
ip netns exec hasvppinterface2 ip addr add 192.168.3.2/24 dev tapcontainer2
ip netns exec hasvppinterface2 ip link set tapcontainer2 up
ip netns exec hasvppinterface2 ip route add 192.168.2.0/24 via 192.168.3.1
ip netns exec hasvppinterface2 ip route add 192.168.1.0/24 via 192.168.3.1
#Let the host also know howto get to the container TAP via VPE
ip route add 192.168.1.0/24 via 192.168.2.1
ip route add 192.168.3.0/24 via 192.168.2.1
#Block ICMP out of the default docker0 container interfaces to prevent false positive results
ip netns exec hasvppinterface1 iptables -A OUTPUT -p icmp -o eth0 -j REJECT
ip netns exec hasvppinterface2 iptables -A OUTPUT -p icmp -o eth0 -j REJECT
#TEST!
echo "Pinging container1 via host > HostVPE > Container1 TAP"
ping -c3 192.168.1.2
echo "Pinging container2 via host > HostVPE > Container2 TAP"
ping -c3 192.168.1.2
echo "Ping from container1 via TAP > HostVPP > Host"
docker exec hasvppinterface1 ping -c3 192.168.2.2
echo "Ping from container2 via TAP > HostVPP > Host"
docker exec hasvppinterface2 ping -c3 192.168.2.2
echo "Ping from Container1 to Container2 via TAP > HostVPP > TAP"
docker exec hasvppinterface1 ping -c3 192.168.3.2
Running the Example
1. Access the FD.io build VM from your development host. Then sudo to root.
$ vagrant ssh vagrant@localhost$ sudo su -
2. Copy the script above into the VM (alternatively, place it in the vagrant directory of your cloned VPP repo and it will be available in the vagrant VM under the '/vagrant' directory. 3. Run the script to install docker, set up VPP and configure the relevant interfaces for VPP into each container. Pings will then be run to verify connectivity through VPP to each container.
root@localhost# chmod +x /vagrant/thisscript.sh root@localhost# /vagrant/thisscript.sh
Explore the Environment
The script output should have ended with successful pings between the two docker containers and the host as follows:
Pinging container1 via host > HostVPE > Container1 TAP PING 192.168.1.2 (192.168.1.2) 56(84) bytes of data. 64 bytes from 192.168.1.2: icmp_seq=2 ttl=63 time=0.261 ms 64 bytes from 192.168.1.2: icmp_seq=3 ttl=63 time=0.119 ms --- 192.168.1.2 ping statistics --- 3 packets transmitted, 2 received, 33% packet loss, time 2006ms rtt min/avg/max/mdev = 0.119/0.190/0.261/0.071 ms Pinging container2 via host > HostVPE > Container2 TAP PING 192.168.1.2 (192.168.1.2) 56(84) bytes of data. 64 bytes from 192.168.1.2: icmp_seq=1 ttl=63 time=0.423 ms 64 bytes from 192.168.1.2: icmp_seq=2 ttl=63 time=0.101 ms 64 bytes from 192.168.1.2: icmp_seq=3 ttl=63 time=0.162 ms --- 192.168.1.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 1998ms rtt min/avg/max/mdev = 0.101/0.228/0.423/0.140 ms Ping from container1 via TAP > HostVPP > Host 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=63 time=0.092 ms 64 bytes from 192.168.2.2: icmp_seq=2 ttl=63 time=0.177 ms 64 bytes from 192.168.2.2: icmp_seq=3 ttl=63 time=0.164 ms --- 192.168.2.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 2000ms rtt min/avg/max/mdev = 0.092/0.144/0.177/0.038 ms Ping from container2 via TAP > HostVPP > Host 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=63 time=0.166 ms 64 bytes from 192.168.2.2: icmp_seq=2 ttl=63 time=0.122 ms 64 bytes from 192.168.2.2: icmp_seq=3 ttl=63 time=0.162 ms --- 192.168.2.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 1999ms rtt min/avg/max/mdev = 0.122/0.150/0.166/0.019 ms Ping from Container1 to Container2 via TAP > HostVPP > TAP PING 192.168.3.2 (192.168.3.2) 56(84) bytes of data. 64 bytes from 192.168.3.2: icmp_seq=1 ttl=63 time=0.103 ms 64 bytes from 192.168.3.2: icmp_seq=2 ttl=63 time=0.164 ms 64 bytes from 192.168.3.2: icmp_seq=3 ttl=63 time=0.289 ms --- 192.168.3.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 2000ms rtt min/avg/max/mdev = 0.103/0.185/0.289/0.078 ms
At this point, you can also verify we have two docker containers running and view their routing tables:
root@localhost:~# docker ps CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES 18ae72d3c11d ubuntu "sleep 30000" 55 seconds ago Up 54 seconds hasvppinterface2 f92742c796dd ubuntu "sleep 30000" 55 seconds ago Up 54 seconds hasvppinterface1 root@localhost:~# ip netns exec hasvppinterface1 ip route list default via 172.17.0.1 dev eth0 172.17.0.0/16 dev eth0 proto kernel scope link src 172.17.0.2 192.168.1.0/24 dev tapcontainer1 proto kernel scope link src 192.168.1.2 192.168.2.0/24 via 192.168.1.1 dev tapcontainer1 192.168.3.0/24 via 192.168.1.1 dev tapcontainer1 root@localhost:~# ip netns exec hasvppinterface2 ip route list default via 172.17.0.1 dev eth0 172.17.0.0/16 dev eth0 proto kernel scope link src 172.17.0.3 192.168.1.0/24 via 192.168.3.1 dev tapcontainer2 192.168.2.0/24 via 192.168.3.1 dev tapcontainer2 192.168.3.0/24 dev tapcontainer2 proto kernel scope link src 192.168.3.2
We can also connect to the VPP configuration interface (left exposed on localhost:5002 in this example) to show VPP interfaces, routing table and ARP entries as follows:
root@localhost:~# telnet localhost 5002
Trying ::1...
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
vpp# show interface
Name Idx State Counter Count
local0 0 down
pg/stream-0 1 down
pg/stream-1 2 down
pg/stream-2 3 down
pg/stream-3 4 down
tap-0 5 up rx packets 29
rx bytes 2458
tx packets 13
tx bytes 1162
drops 17
ip4 11
ip6 16
tap-1 6 up rx packets 21
rx bytes 1866
tx packets 12
tx bytes 1120
drops 9
ip4 12
ip6 8
tap-2 7 up rx packets 23
rx bytes 1926
tx packets 7
tx bytes 630
drops 16
ip4 6
ip6 16
vpp# show ip fib
Table 0, fib_index 0, flow hash: src dst sport dport proto
Destination Packets Bytes Adjacency
192.168.1.0/24 1 98 weight 1, index 3
arp tap-0 192.168.1.1/24
192.168.1.1/32 0 0 weight 1, index 4
local 192.168.1.1/24
192.168.1.2/32 5 490 weight 1, index 11
tap-0
IP4: e2:c9:1a:c6:5e:06 -> e2:c9:1a:c6:5e:06
192.168.2.0/24 0 0 weight 1, index 5
arp tap-1 192.168.2.1/24
192.168.2.1/32 0 0 weight 1, index 6
local 192.168.2.1/24
192.168.2.2/32 11 1078 weight 1, index 9
tap-1
IP4: 42:f0:8e:5f:65:64 -> 42:f0:8e:5f:65:64
192.168.3.0/24 0 0 weight 1, index 7
arp tap-2 192.168.3.1/24
192.168.3.1/32 0 0 weight 1, index 8
local 192.168.3.1/24
192.168.3.2/32 6 588 weight 1, index 10
tap-2
IP4: 8a:e2:eb:68:15:9d -> 8a:e2:eb:68:15:9d
vpp# show ip arp
Time FIB IP4 Stat Ethernet Interface
40.7780 0 192.168.1.2 e2:c9:1a:c6:5e:06 tap-0
34.7544 0 192.168.2.2 42:f0:8e:5f:65:64 tap-1
41.9191 0 192.168.3.2 8a:e2:eb:68:15:9d tap-2
vpp# quit
Run other commands in the containers
If you wish to run any other commands from within the VPP routed containers, you can create a shell to the container as follows:
root@localhost:~# docker exec -ti hasvppinterface1 bash root@f92742c796dd:/# echo "hello from container1" hello from container1 root@f92742c796dd:/# exit root@localhost:~# docker exec -ti hasvppinterface2 bash root@18ae72d3c11d:/# echo "hello from container2" hello from container2 root@18ae72d3c11d:/# ip route list default via 172.17.0.1 dev eth0 172.17.0.0/16 dev eth0 proto kernel scope link src 172.17.0.3 192.168.1.0/24 via 192.168.3.1 dev tapcontainer2 192.168.2.0/24 via 192.168.3.1 dev tapcontainer2 192.168.3.0/24 dev tapcontainer2 proto kernel scope link src 192.168.3.2
Non-Vagrant environments
The script used to create this demo is fairly well commented using very simple Linux shell commands, so should be easily re-produceable in other environments.
Adding a new tap interface
Unlike the AF_PACKET interfaces, which must be specified at VPP startup (no longer true, see 'create host-interface'), we can create a new linux TAP interface to VPP at any point through configuration. Still on the Vagrant SSH session, see the following example:
### Show current VPP and linux interfaces
root@localhost:~# vppctl show int
Name Idx State Counter Count
local0 0 down
pg/stream-0 1 down
pg/stream-1 2 down
pg/stream-2 3 down
pg/stream-3 4 down
tap-0 5 up rx packets 29
rx bytes 2458
tx packets 13
tx bytes 1162
drops 17
ip4 11
ip6 16
tap-1 6 up rx packets 21
rx bytes 1866
tx packets 12
tx bytes 1120
drops 9
ip4 12
ip6 8
tap-2 7 up rx packets 23
rx bytes 1926
tx packets 7
tx bytes 630
drops 16
ip4 6
ip6 16
root@localhost:~# ip link list
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP mode DEFAULT group default qlen 1000
link/ether 08:00:27:b1:94:b1 brd ff:ff:ff:ff:ff:ff
3: docker0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default
link/ether 02:42:53:6f:28:38 brd ff:ff:ff:ff:ff:ff
5: taphost: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 4352 qdisc pfifo_fast state UNKNOWN mode DEFAULT group default qlen 500
link/ether 42:f0:8e:5f:65:64 brd ff:ff:ff:ff:ff:ff
8: veth27cd9d5: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master docker0 state UP mode DEFAULT group default
link/ether 72:7c:93:5f:a9:d8 brd ff:ff:ff:ff:ff:ff
10: veth9d6aa0f: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master docker0 state UP mode DEFAULT group default
link/ether 26:32:9e:07:5a:e0 brd ff:ff:ff:ff:ff:ff
### Add a new tap interface
vppctl tap connect newtap
### Show interfaces on VPP and Linux
root@localhost:~# vppctl show interface
Name Idx State Counter Count
local0 0 down
pg/stream-0 1 down
pg/stream-1 2 down
pg/stream-2 3 down
pg/stream-3 4 down
tap-0 5 up rx packets 29
rx bytes 2458
tx packets 13
tx bytes 1162
drops 17
ip4 11
ip6 16
tap-1 6 up rx packets 21
rx bytes 1866
tx packets 12
tx bytes 1120
drops 9
ip4 12
ip6 8
tap-2 7 up rx packets 23
rx bytes 1926
tx packets 7
tx bytes 630
drops 16
ip4 6
ip6 16
tap-3 8 down rx packets 7
rx bytes 578
drops 7
root@localhost:~# ip link list
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP mode DEFAULT group default qlen 1000
link/ether 08:00:27:b1:94:b1 brd ff:ff:ff:ff:ff:ff
3: docker0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default
link/ether 02:42:53:6f:28:38 brd ff:ff:ff:ff:ff:ff
5: taphost: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 4352 qdisc pfifo_fast state UNKNOWN mode DEFAULT group default qlen 500
link/ether 42:f0:8e:5f:65:64 brd ff:ff:ff:ff:ff:ff
8: veth27cd9d5: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master docker0 state UP mode DEFAULT group default
link/ether 72:7c:93:5f:a9:d8 brd ff:ff:ff:ff:ff:ff
10: veth9d6aa0f: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master docker0 state UP mode DEFAULT group default
link/ether 26:32:9e:07:5a:e0 brd ff:ff:ff:ff:ff:ff
11: newtap: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 4352 qdisc pfifo_fast state UNKNOWN mode DEFAULT group default qlen 500
link/ether da:28:dd:bd:e2:93 brd ff:ff:ff:ff:ff:ff
