Download code from GitHub
The recipes in this chapter will provide an introduction into configuring OpenVPN. The recipes are based on a point-to-point style network, meaning that only a single client can connect at a time.
A point-to-point style network is very useful when connecting to a small number of sites or clients. It is easier to set up, as no certificates or Public Key Infrastructure (PKI) is required. Also, routing is slightly easier to configure, as no client-specific configuration files containing --iroute statements are required.
The drawbacks of a point-to-point style network are:
The lack of perfect forward secrecy— a key compromise may result in a total disclosure of previous sessions
The secret key must exist in plaintext form on each VPN peer
This recipe will explain the shortest setup possible when using OpenVPN. For this setup two computers are used that are connected over a network (LAN or Internet). We will use both a TUN-style network and a TAP-style network and will focus on the differences between them. A TUN device is used mostly for VPN tunnels where only IP-traffic is used. A TAP device allows full Ethernet frames to be passed over the OpenVPN tunnel, hence providing support for non-IP based protocols such as IPX and AppleTalk.
While this may seem useless at first glance, it can be very useful to quickly test whether OpenVPN can connect to a remote system.
Install OpenVPN 2.0 or higher on two computers. Make sure the computers are connected over a network. For this recipe, the server computer was running CentOS 5 Linux and OpenVPN 2.1.1 and the client was running Windows XP SP3 and OpenVPN 2.1.1.
We launch the server (listening)-side OpenVPN process for the TUN-style network:
[root@server]# openvpn --ifconfig 10.200.0.1 10.200.0.2 \ --dev tunThen we launch the client-side OpenVPN process:
[WinClient] C:\>"\Program Files\OpenVPN\bin\openvpn.exe" \ --ifconfig 10.200.0.2 10.200.0.1 --dev tun \ --remote openvpnserver.example.comThe following screenshot shows how a connection is established:
As soon as the connection is established, we can ping the other end of the tunnel.
Next, we stop the tunnel by pressing the F4 function key in the Command window and we restart both ends of the tunnel using the TAP device:
We launch the server (listening)-side OpenVPN process for the TAP-style network:
[root@server]# openvpn --ifconfig 10.200.0.1 255.255.255.0 \ --dev tapThen we launch the client-side OpenVPN process:
[WinClient] C:\>"\Program Files\OpenVPN\bin\openvpn.exe" \ --ifconfig 10.200.0.2 255.255.255.0 --dev tap \ --remote openvpnserver.example.com
The connection is established and we can again ping the other end of the tunnel.
The server listens on UDP port 1194, which is the OpenVPN default port for incoming connections. The client connects to the server on this port. After the initial handshake, the server configures the first available TUN device with IP address 10.200.0.1 and it expects the remote end (Peer address) to be 10.200.0.2.
The client does the opposite: after the initial handshake, the first TUN or TAP-Win32 device is configured with IP address 10.200.0.2. It expects the remote end (Peer address) to be 10.200.0.1. After this, the VPN is established.
In case of a TAP-style network, the server configures the first available TAP device with the IP address 10.200.0.01 and netmask 255.255.255.0. Similarly, the client is configured with IP address 10.200.0.2 and netmask 255.255.255.0.
In the previous example, we chose the UDP protocol. For this example, it would not have made any difference if we had chosen the TCP protocol, provided that we do that on the server side (the side without --remote):
[root@server]# openvpn --ifconfig 10.200.0.1 10.200.0.2 \
–-dev tun --proto tcp-server
And also on the client side:
[root@server]# openvpn --ifconfig 10.200.0.2 10.200.0.1 \
--dev tun --proto tcp-client
It is now possible to run non-IP traffic over the tunnel. For example, if AppleTalk is configured correctly on both sides, we can query a remote host using the aecho command:
aecho openvpnserver
22 bytes from 65280.1: aep_seq=0. time=26. ms
22 bytes from 65280.1: aep_seq=1. time=26. ms
22 bytes from 65280.1: aep_seq=2. time=27. ms
A tcpdump -nnel -i tap0 shows that the type of traffic is indeed non-IP based AppleTalk.
This recipe uses OpenVPN secret keys to secure the VPN tunnel. It is very similar to the previous recipe but this time a shared secret key is used to encrypt the traffic between the client and the server.
Install OpenVPN 2.0 or higher on two computers. Make sure the computers are connected over a network. For this recipe, the server computer was running CentOS 5 Linux and OpenVPN 2.1.1 and the client was running Windows XP SP3 and OpenVPN 2.1.1.
First, we generate a secret key on the server (listener):
[root@server]# openvpn --genkey --secret secret.keyWe transfer this key to the client side over a secure channel (for example, using
scp).Next, we launch the server (listening)-side OpenVPN process:
[root@server]# openvpn --ifconfig 10.200.0.1 10.200.0.2 \ --dev tun --secret secret.keyThen, we launch the client-side OpenVPN process:
[WinClient] C:\>"\Program Files\OpenVPN\bin\openvpn.exe" \ --ifconfig 10.200.0.2 10.200.0.1 \ --dev tun --secret secret.key \ --remote openvpnserver.example.comThe connection is established:
This example works exactly as the very first: the server listens to the incoming connections on UDP port 1194. The client connects to the server on this port. After the initial handshake, the server configures the first available TUN device with IP address 10.200.0.1 and it expects the remote end (Peer address) to be 10.200.0.2. The client does the opposite.
By default, OpenVPN uses two symmetric keys when setting up a point-to-point connection:
A Cipher key to encrypt the contents of the packets being exchanged.
An HMAC key to sign packets. When packets arrive that are not signed using the appropriate HMAC key they are dropped immediately. This is the first line of defense against a "Denial of Service" attack.
The same set of keys are used on both ends and both keys are derived from the file specified using the --secret parameter.
An OpenVPN secret key file is formatted as follows:
#
# 2048 bit OpenVPN static key
#
-----BEGIN OpenVPN Static key V1-----
<16 lines of random bytes>
-----END OpenVPN Static key V1-----From the random bytes, the OpenVPN Cipher and HMAC keys are derived. Note that these keys are the same for each session!
As stated in the previous recipe, OpenVPN uses two symmetric keys when setting up a point-to-point connection. However, it is also possible to use shared, yet asymmetric keys in point-to-point mode. OpenVPN will use four keys in this case:
A Cipher key on the client side
An HMAC key on the client side
A Cipher key on the server side
An HMAC key on the server side
The same keying material is shared by both sides of the point-to-point connection but those keys that are derived for encrypting and signing the data are different for each side. This recipe explains how to set up OpenVPN in this manner and how the keys can be made visible.
For this recipe, we use the secret.key file from the previous recipe. Install OpenVPN 2.0 or higher on two computers. Make sure that the computers are connected over a network. For this recipe, the server computer was running CentOS 5 Linux and OpenVPN 2.1.1. The client was running Windows XP SP3 and OpenVPN 2.1.1.
We launch the server (listening) side OpenVPN process with an extra option to the
--secretparameter and with more verbose logging:[root@server]# openvpn \ --ifconfig 10.200.0.1 10.200.0.2 \ --dev tun --secret secret.key 0 \ --verb 7Then we launch the client-side OpenVPN process:
[WinClient] C:\>"\Program Files\OpenVPN\bin\openvpn.exe" \ --ifconfig 10.200.0.2 10.200.0.1 \ --dev tun --secret secret.key 1\ --remote openvpnserver \ --verb 7The connection will be established with a lot of debugging messages.
If we look through the server-side messages (searching for
crypt), we can find the negotiated keys on the server side. Note that the output has been reformatted for clarity:… Static Encrypt: Cipher 'BF-CBC' initialized with 128 bit key … Static Encrypt: CIPHER KEY: 80797ddc 547fbdef 79eb353f 2a1f3d1f … Static Encrypt: Using 160 bit message hash 'SHA1' for HMAC authentication … Static Encrypt: HMAC KEY: c752f254 cc4ac230 83bd8daf 6141e73d 844764d8 … Static Decrypt: Cipher 'BF-CBC' initialized with 128 bit key … Static Decrypt: CIPHER KEY: 8cf9abdd 371392b1 14b51523 25302c99 … Static Decrypt: Using 160 bit message hash 'SHA1' for HMAC authentication … Static Decrypt: HMAC KEY: 39e06d8e 20c0d3c6 0f63b3e7 d94f35af bd744b27
On the client side, we will find the same keys but the 'Encrypt' and 'Decrypt' keys have been reversed:
… Static Encrypt:
Cipher 'BF-CBC' initialized with 128 bit key
… Static Encrypt:
CIPHER KEY: 8cf9abdd 371392b1 14b51523 25302c99
… Static Encrypt:
Using 160 bit message hash 'SHA1' for HMAC authentication
… Static Encrypt:
HMAC KEY: 39e06d8e 20c0d3c6 0f63b3e7 d94f35af bd744b27
… Static Decrypt:
Cipher 'BF-CBC' initialized with 128 bit key
… Static Decrypt:
CIPHER KEY: 80797ddc 547fbdef 79eb353f 2a1f3d1f
… Static Decrypt:
Using 160 bit message hash 'SHA1' for HMAC authentication
… Static Decrypt:
HMAC KEY: c752f254 cc4ac230 83bd8daf 6141e73d 844764d8
If you look at the keys carefully, you can see that each one of them is mirrored on the client and the server side.
OpenVPN derives all keys from the static.key file, provided that there is enough entropy (randomness) in the file to reliably generate four keys. All keys generated using the following will have enough entropy:
$ openvpn –-genkey –-secret secret.key
An OpenVPN static key file is 2048 bits in size. The Cipher keys are each 128 bits, whereas the HMAC keys are 160 bits each, for a total of 776 bits. This allows OpenVPN to easily generate four random keys from the static key file, even if a cipher is chosen that requires a larger initialization key.
The same secret key files are used in a client/server setup when the following parameter is used: tls-auth ta.key.
Chapter 2's recipe, Setting up the public and private keys, in which the
tls-authkey is generated in a very similar manner.
In the very first recipe, we created a tunnel in which the data traffic was not encrypted. To create a completely plain text tunnel, we also disable the HMAC authentication. This can be useful when debugging a bad connection, as all traffic over the tunnel can now easily be monitored. In this recipe, we will look at how to do this. This type of tunnel is also useful when doing performance measurements, as it is the least CPU-intensive tunnel that can be established.
Install OpenVPN 2.0 or higher on two computers. Make sure the computers are connected over a network. For this recipe, the server computer was running CentOS 5 Linux and OpenVPN 2.1.1 and the client was running Fedora 13 Linux and OpenVPN 2.1.1.
As we are not using any encryption, no secret keys are needed.
Launch the server (listening)-side OpenVPN process:
[root@server]# openvpn \ --ifconfig 10.200.0.1 10.200.0.2 \ --dev tun -–auth noneThen launch the client-side OpenVPN process:
[root@client]# openvpn \ --ifconfig 10.200.0.2 10.200.0.1 \ --dev tun –-auth none\ --remote openvpnserver.example.comThe connection is established with two warning messages in the output:
… ******* WARNING *******: null cipher specified, no encryption will be used
… ******* WARNING *******: null MAC specified, no authentication will be used
With this setup, absolutely no encryption is performed. All the traffic that is sent over the tunnel is encapsulated in an OpenVPN packet and then sent "as-is".
To actually view the traffic, we can use tcpdump:
Set up the connection as outlined.
Start
tcpdumpand listen on the network interface, not the tunnel interface itself:[root]@client]# tcpdump -w -I eth0 -s 0 host openvpnserver \ | stringsNow, send some text across the tunnel, using something like
nc(Netcat). First, launchncon the server side:[server]$ nc -l 31000On the client side, launchncin client mode and type the wordshelloandgoodbye.[client]$ nc 10.200.0.1 3100 hello goodbyeIn the
tcpdumpwindow, you should now see:
Point-to-point style networks are great if you want to connect two networks together over a static, encrypted tunnel. If you only have a small number of endpoints (fewer than four), then it is far easier than using a client/server setup as described in Chapter 2, Client-server IP-only Networks.
For this recipe, we use the following network layout:
Install OpenVPN 2.0 or higher on two computers. Make sure the computers are connected over a network. For this recipe, the server computer was running CentOS 5 Linux and OpenVPN 2.1.1 and the client was running Fedora 13 Linux and OpenVPN 2.1.1. We'll use the secret.key file from the OpenVPN Secret keys recipe here.
First we establish the connection, but we also make sure OpenVPN daemonized itself:
[root@server]# openvpn \ --ifconfig 10.200.0.1 10.200.0.2 \ --dev tun --secret secret.key \ --daemon --log /tmp/openvpnserver.logThen we launch the client-side OpenVPN process:
[client]$ openvpn \ --ifconfig 10.200.0.2 10.200.0.1 \ --dev tun --secret secret.key \ --remote openvpnserver \ --daemon --log /tmp/openvpnclient.logThe connection is established:
[server]$ tail -1 /tmp/openvpnserver.log Initialization Sequence Completed
Now we add routing:
On the server side, we add a static route:
[root@server]# route add -net 192.168.4.0/24 gw 10.200.0.2On the client side, we need to do two things:
Make sure that you have IP traffic forwarding enabled. On Linux this can be achieved using the following:
[root@client]# sysctl -w net.ipv4.ip_forward=1Make sure that on the Windows client on the client-side LAN there is a route back to the OpenVPN server:
C:> route add 10.200.0.0 mask 255.255.255.0 192.168.4.5Here 192.168.4.5 is the LAN IP address of the OpenVPN client.
From the server, we can now ping machines on the client LAN. First we ping the LAN IP of the OpenVPN client:
[root@server]# ping -c 2 192.168.4.5 PING 192.168.4.5 (192.168.4.5) 56(84) bytes of data. 64 bytes from 192.168.4.5: icmp_seq=0 ttl=64 time=31.7 ms 64 bytes from 192.168.4.5: icmp_seq=1 ttl=64 time=31.3 ms --- 192.168.4.5 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1000ms rtt min/avg/max/mdev = 31.359/31.537/31.716/0.251 ms, pipe 2And next the LAN IP of a machine on the OpenVPN client LAN:
[root@server]# ping -c 2 192.168.4.164 [server]$ ping -c 2 192.168.4.164 PING 192.168.4.164 (192.168.4.164) 56(84) bytes of data. 64 bytes from 192.168.4.164: icmp_seq=0 ttl=63 time=31.9 ms 64 bytes from 192.168.4.164: icmp_seq=1 ttl=63 time=31.4 ms --- 192.168.4.164 ping statistics --- 2 packets transmitted, 2 received, 0% packet loss, time 1001ms rtt min/avg/max/mdev = 31.486/31.737/31.989/0.308 ms, pipe 2
In our network setup, the LAN we want to reach is behind the OpenVPN client, so we have to add a route to the server:
[server]$ route add -net 192.168.4.0/24 gw 10.200.0.2
On the client side, we need to do two things:
Make sure that the routing is enabled. If you want routing to remain enabled after a reboot, edit the file
/etc/sysctl.cnf.net.ipv4.ip_forward = 1We also need to make sure that on the client LAN there is a route back to the OpenVPN server. This can be done by adding a route to the LAN gateway or by adding a static route to each of the machines on the client LAN. In this recipe, we added a route to a Windows client that is in the same LAN as the OpenVPN client:
C:> route add 10.200.0.0 mask 255.255.255.0 192.168.4.5where 192.168.4.5 is the LAN IP address of the OpenVPN client
.
On the openvpn-users mailing list, a large number of the problems reported have to do with routing issues. Most of them have little to do with the OpenVPN itself but more with understanding the routing and the flow of packets over the network. Chapter 8, Troubleshooting OpenVPN: Routing Issues, provides some recipes to diagnose and fix the most common routing problems.
It is also possible to add the appropriate routes when the tunnel first comes up. This can be done using the --route statement:
[server]$ openvpn \
--ifconfig 10.200.0.1 10.200.0.2 \
--dev tun --secret secret.key \
--daemon --log /var/log/openvpnserver-1.5.log \
--route 192.168.4.0 255.255.255.0
Note that on the client LAN the route back to the server still has to be set manually.
The last recipe of this chapter, 3-way routing, in which a more complicated setup using three remote sites is explained.
Chapter 8, Troubleshooting OpenVPN: Routing Issues
Most recipes in this book can be carried out without using configuration files. However, in most real-life cases a configuration file is much easier to use than a lengthy command-line. It is important to know that OpenVPN actually treats configuration file entries and command-line parameters identically. The only difference is that all command-line parameters start with a double dash ("--") whereas the configuration file entries do not. This makes it very easy to overrule the configuration file entries using an extra command-line parameter.
Install OpenVPN 2.0 or higher on two computers. Make sure the computers are connected over a network. For this recipe, the server computer was running CentOS 5 Linux and OpenVPN 2.1.1 and the client was running Windows XP SP3 and OpenVPN 2.1.1 In this recipe we'll use the secret.key file from the OpenVPN Secret keys recipe.
Create a configuration file based on an earlier recipe:
dev tun port 1194 ifconfig 10.200.0.1 10.200.0.2 secret secret.key remote openvpnserver.example.com verb 3Save this file as
example1-6-client.conf.We launch the server (listening)-side OpenVPN process on a non-standard port:
[root@server]# openvpn \ --ifconfig 10.200.0.1 10.200.0.2 \ --dev tun --secret secret.key \ --port 11000Then we launch the client-side OpenVPN process and add an extra command-line parameter:
[WinClient] C:\>"\Program Files\OpenVPN\bin\openvpn.exe" \ --config client.conf \ --port 11000
The connection is established:
The command-line and the configuration file are read and parsed from left to right. This means that most options that are specified before the configuration file can be overruled by entries in that file. Similarly, options specified after the following directive overrule the entries in that file:
--config client.conf
Hence, the following option overruled the line 'port 1194' from the configuration file:
--port 11000
However, some options can be specified multiple times, in which case the first occurrence "wins". In that case, it is also possible to specify the option before specifying the --config directive.
Here is another example to show the importance of the ordering of the command-line parameters:
C:\>"\Program Files\OpenVPN\bin\openvpn.exe" \
--verb 0 \
--config client.conf \
--port 11000
This produces the exact same connection log as shown before. The 'verb 3' from the client.conf configuration file overruled the --verb 0 as specified on the command line. However, with the following command line:
C:\>"\Program Files\OpenVPN\bin\openvpn.exe" \
--config client.conf \
--port 11000 \
--verb 0
Then the connection log shows the following:
… NOTE: OpenVPN 2.1 requires '--script-security 2' or higher to call user-defined scripts or executables
This shows all the other messages that have been muted.
Some of the newer features of OpenVPN 2.1 deviate slightly from this principle, notably the <connection> blocks and the inline certificates. See Chapter 12, OpenVPN 2.1 specifics for more details.
In this recipe, we set up a complete site-to-site network, using most of the built-in security features that OpenVPN offers. It is intended as a "one-stop-shop" example of how to set up a point-to-point network.
We use the following network layout:
Install OpenVPN 2.0 or higher on two computers. Make sure that the computers are connected over a network. For this recipe, the server computer was running CentOS 5 Linux and OpenVPN 2.1.1 and the client was running Fedora 13 Linux and OpenVPN 2.1.1. We'll use the secret.key file from the OpenVPN Secret keys recipe here.
Make sure routing (IP forwarding) is configured on both the server and client.
Create the server configuration file:
dev tun proto udp local openvpnserver.example.com lport 1194 remote openvpnclient.example.com rport 1194 secret secret.key 0 ifconfig 10.200.0.1 10.200.0.2 route 192.168.4.0 255.255.255.0 user nobody group nobody persist-tun persist-key keepalive 10 60 ping-timer-rem verb 3 daemon log-append /tmp/openvpn.logSave it as
example1-7-server.conf.On the client side, we create the configuration file:
dev tun proto udp local openvpnclient.example.com lport 1194 remote openvpnserver.example.com rport 1194 secret secret.key 1 ifconfig 10.200.0.2 10.200.0.1 route 172.31.32.0 255.255.255.0 user nobody group nobody persist-tun persist-key keepalive 10 60 ping-timer-rem verb 3 daemon log-append /tmp/openvpn.logSave it as
example1-7-client.conf.We start the tunnel on both ends:
[root@server]# openvpn --config example1-7-server.confAnd:
[root@client]# openvpn --config client.confNow our site-to-site tunnel is established.
Check the log files on both the client and server, to verify that the connection has been established.
After the connection comes up, the machines on the LANs behind both end points can be reached over the OpenVPN tunnel.
For example, when we ping a machine on the client-side LAN from the server, we see the following:
The client and server configuration files are very similar:
The server listens only on one interface and one UDP port
The server accepts connections only from a single IP address and port
The client has these options mirrored
Here is the set of configuration options:
user nobody
group nobody
persist-tun
persist-key
keepalive 10 60
ping-timer-remThey are used to make the connection more robust and secure, as follows:
The OpenVPN process runs as user
nobody, groupnobody, after the initial connection is established. Even if somebody is able to take control of the OpenVPN process itself he would still only be usernobodyand notroot. Note that on some Linux distributions the groupnogroupis used instead.The
persist-tunandpersist-keyoptions are used to ensure that the connection comes back up automatically if the underlying network is disrupted. These options are necessary when usinguser nobodyandgroup nobody(orgroup nogroup).The
keepaliveandping-timer-remoptions cause OpenVPN to send a periodic 'ping' message over the tunnel to ensure that both ends of the tunnel remain up and running.
This point-to-point setup can also be used to evade restrictive firewalls. The data stream between the two endpoints is not recognizable and very hard to decipher. When OpenVPN is run in client/server (see Chapter 2, Multi-client TUN-style Networks), the traffic is recognizable as OpenVPN traffic due to the initial TLS handshake.
Chapter 8, Troubleshooting OpenVPN: Routing Issues, in which the most common routing issues are explained.
For a small number (less than four) of fixed endpoints, a point-to-point setup is very flexible. In this recipe, we set up three OpenVPN tunnels between three sites, including routing between the endpoints. By setting up three tunnels, we create a redundant routing so that all sites are connected even if one of the tunnels is disrupted.
We use the following network layout:
Install OpenVPN 2.0 or higher on two computers. Make sure the computers are connected over a network. In this recipe, the tunnel endpoints were running CentOS 5 Linux or Fedora 13 Linux and OpenVPN 2.1.1. Make sure that the routing (IP forwarding) is configured on all the OpenVPN endpoints.
We generate three static keys:
[root@siteA]# openvpn –-genkey –-secret AtoB.key [root@siteA]# openvpn –-genkey –-secret AtoC.key [root@siteA]# openvpn –-genkey –-secret BtoC.keyTransfer these keys to all endpoints over a secure channel (for example, using
scp).Create the server (listener) configuration file named
example1-8-serverBtoA.conf:dev tun proto udp port 1194 secret AtoB.key 0 ifconfig 10.200.0.1 10.200.0.2 route 192.168.4.0 255.255.255.0 vpn_gateway 5 route 192.168.6.0 255.255.255.0 vpn_gateway 10 route-delay keepalive 10 60 verb 3 Next, create example1-8-serverCtoA.conf: dev tun proto udp port 1195 secret AtoC.key 0 ifconfig 10.200.0.5 10.200.0.6 route 192.168.4.0 255.255.255.0 vpn_gateway 5 route 192.168.5.0 255.255.255.0 vpn_gateway 10 route-delay keepalive 10 60 verb 3 and example1-8-serverBtoC.conf: dev tun proto udp port 1196 secret BtoC.key 0 ifconfig 10.200.0.9 10.200.0.10 route 192.168.4.0 255.255.255.0 vpn_gateway 10 route 192.168.6.0 255.255.255.0 vpn_gateway 5 route-delay keepalive 10 60 verb 3 Now, create the client (connector) configuration files example1-8-clientAtoB.conf: dev tun proto udp remote siteB port 1194 secret AtoB.key 1 ifconfig 10.200.0.2 10.200.0.1 route 192.168.5.0 255.255.255.0 vpn_gateway 5 route 192.168.6.0 255.255.255.0 vpn_gateway 10 route-delay keepalive 10 60 verb 3 Also, create example1-8-clientAtoC.conf file: dev tun proto udp remote siteC port 1195 secret AtoC.key 1 ifconfig 10.200.0.6 10.200.0.5 route 192.168.5.0 255.255.255.0 vpn_gateway 10 route 192.168.6.0 255.255.255.0 vpn_gateway 5 route-delay verb 3 and finally the example1-8-clientCtoB.conf: dev tun proto udp remote siteB port 1196 secret BtoC.key 1 ifconfig 10.200.0.10 10.200.0.9 route 192.168.4.0 255.255.255.0 vpn_gateway 10 route 192.168.5.0 255.255.255.0 vpn_gateway 5 route-delay keepalive 10 60 verb 3
First, we start all the listener tunnels:
[root@siteB]# openvpn --config example1-8-serverBtoA.conf
[root@siteB]# openvpn --config example1-8-serverBtoC.conf
[root@siteC]# openvpn --config example1-8-serverCtoA.conf
These are followed by the connector tunnels:
[root@siteA]# openvpn --config example1-8-clientAtoB.conf
[root@siteA]# openvpn --config example1-8-clientAtoC.conf
[root@siteC]# openvpn --config example1-8-clientCtoB.conf
And with that, our three-way site-to-site network is established.
It can clearly be seen that the number of configuration files gets out of hand too quickly. In principle, two tunnels would have been sufficient to connect three remote sites, but then there would have been no redundancy.
With the third tunnel and with the configuration options:
route 192.168.5.0 255.255.255.0 vpn_gateway 5
route 192.168.6.0 255.255.255.0 vpn_gateway 10
route-delay
keepalive 10 60
There are always 2 routes to each remote network.
For example, site A has two routes to site B (LAN 192.168.5.0/24), as seen from the following routing table:
[siteA]$ ip route show
[…]
192.168.5.0/24 via 10.200.0.1 dev tun0 metric 5
192.168.5.0/24 via 10.200.0.5 dev tun1 metric 10
[…]A route:
This setup has the advantage that if one tunnel fails, then after 60 seconds, the connection and its corresponding routes are dropped and are restarted. The backup route to the other network then automatically takes over and all three sites can reach each other again.
When the "direct" tunnel is restored the direct routes are also restored and the network traffic will automatically choose the best path to the remote site.
In this recipe, we connect three remote sites. This results in six different configuration files that provide the limitations of the point-to-point setup. In general, to connect N possible sites with full redundancy, you will have N * ( N – 1 ) configuration files. This is manageable for up to four sites, but after that, a server/multiple-client setup as described in the next chapters is much easier.
Chapter 8, Troubleshooting OpenVPN: Routing Issues, in which the most common routing issues are explained.
