summaryrefslogtreecommitdiff
path: root/app/openvpn/contrib/openvpn-fwmarkroute-1.00
diff options
context:
space:
mode:
Diffstat (limited to 'app/openvpn/contrib/openvpn-fwmarkroute-1.00')
-rw-r--r--app/openvpn/contrib/openvpn-fwmarkroute-1.00/README44
-rwxr-xr-xapp/openvpn/contrib/openvpn-fwmarkroute-1.00/fwmarkroute.down22
-rwxr-xr-xapp/openvpn/contrib/openvpn-fwmarkroute-1.00/fwmarkroute.up49
3 files changed, 115 insertions, 0 deletions
diff --git a/app/openvpn/contrib/openvpn-fwmarkroute-1.00/README b/app/openvpn/contrib/openvpn-fwmarkroute-1.00/README
new file mode 100644
index 00000000..66fe61ad
--- /dev/null
+++ b/app/openvpn/contrib/openvpn-fwmarkroute-1.00/README
@@ -0,0 +1,44 @@
+OpenVPN fwmark Routing
+Sean Reifschneider, <jafo@tummy.com>
+Thursday November 27, 2003
+==========================
+
+These scripts can be used with OpenVPN up and down scripts to set up
+routing on a Linux system such that the VPN traffic is sent via normal
+network connectivity, but other traffic to that network runs over the VPN.
+The idea is to allow encryption of data to the network the remote host is
+on, without interfering with the VPN traffic. You can't simply add a route
+to the remote network, becaues that will cause the VPN traffic to also try
+to run over the VPN, and breaks the VPN.
+
+These scripts use the Linux "fwmark" iptables rules to specify routing
+based not only on IP address, but also by port and protocol. This allows
+you to effectively say "if the packet is to this IP address on this port
+using this protocol, then use the normal default gateway, otherwise use the
+VPN gateway.
+
+This is set up on the client VPN system, not the VPN server. These scripts
+also set up all ICMP echo-responses to run across the VPN. You can
+comment the lines in the scripts to disable this, but I find this useful
+at coffee shops which have networks that block ICMP.
+
+To configure this, you need to set up these scripts as your up and down
+scripts in the config file. You will need to set these values in the
+config file:
+
+ up /etc/openvpn/fwmarkroute.up
+ down /etc/openvpn/fwmarkroute.down
+ up-restart
+ up-delay
+
+ setenv remote_netmask_bits 24
+
+Note: For this to work, you can't set the "user" or "group" config options,
+because then the scripts will not run as root.
+
+The last setting allows you to control the size of the network the remote
+system is on. The remote end has to be set up to route, probably with
+masquerading or NAT. The network this netmask relates to is calculated
+using the value of "remote" in the conf file.
+
+Sean
diff --git a/app/openvpn/contrib/openvpn-fwmarkroute-1.00/fwmarkroute.down b/app/openvpn/contrib/openvpn-fwmarkroute-1.00/fwmarkroute.down
new file mode 100755
index 00000000..87d67d4d
--- /dev/null
+++ b/app/openvpn/contrib/openvpn-fwmarkroute-1.00/fwmarkroute.down
@@ -0,0 +1,22 @@
+#!/bin/sh
+#
+# Bring down vpn routing.
+
+# calculate the network address
+remote_network=`ipcalc -n "$remote"/"$remote_netmask_bits"`
+remote_network="${remote_network#*=}"
+
+# clear routing via VPN
+ip route del "$remote_network"/"$remote_netmask_bits" via "$5" table vpn.out
+ip route del table vpnonly.out via "$5"
+iptables -D OUTPUT -t mangle -p "$proto" \
+ -d "$remote_network"/"$remote_netmask_bits" \
+ --dport "$remote_port" -j ACCEPT
+iptables -D OUTPUT -t mangle -d "$remote" -j MARK --set-mark 2
+
+# undo the ICMP ping tunneling
+iptables -D OUTPUT -t mangle --protocol icmp --icmp-type echo-request \
+ -j MARK --set-mark 3
+
+# flush route cache
+ip route flush cache
diff --git a/app/openvpn/contrib/openvpn-fwmarkroute-1.00/fwmarkroute.up b/app/openvpn/contrib/openvpn-fwmarkroute-1.00/fwmarkroute.up
new file mode 100755
index 00000000..661ec313
--- /dev/null
+++ b/app/openvpn/contrib/openvpn-fwmarkroute-1.00/fwmarkroute.up
@@ -0,0 +1,49 @@
+#!/bin/sh
+#
+# Bring up vpn routing.
+
+# calculate the network address
+remote_network=`ipcalc -n "$remote"/"$remote_netmask_bits"`
+remote_network="${remote_network#*=}"
+
+# add the stuff that doesn't change if it's not already there
+grep -q '^202 ' /etc/iproute2/rt_tables
+if [ "$?" -ne 0 ]
+then
+ echo 202 vpn.out >> /etc/iproute2/rt_tables
+fi
+grep -q '^203 ' /etc/iproute2/rt_tables
+if [ "$?" -ne 0 ]
+then
+ echo 203 vpnonly.out >> /etc/iproute2/rt_tables
+fi
+ip rule ls | grep -q 'lookup vpn.out *$'
+if [ "$?" -ne 0 ]
+then
+ ip rule add fwmark 2 table vpn.out
+fi
+ip rule ls | grep -q 'lookup vpnonly.out *$'
+if [ "$?" -ne 0 ]
+then
+ ip rule add fwmark 3 table vpnonly.out
+fi
+
+# route VPN traffic using the normal table
+iptables -A OUTPUT -t mangle -p "$proto" -d "$remote" --dport "$remote_port" \
+ -j ACCEPT
+
+# route all other traffic to that host via VPN
+iptables -A OUTPUT -t mangle -d "$remote_network"/"$remote_netmask_bits" \
+ -j MARK --set-mark 2
+
+# route all ICMP pings over the VPN
+iptables -A OUTPUT -t mangle --protocol icmp --icmp-type echo-request \
+ -j MARK --set-mark 3
+
+# NAT traffic going over the VPN, so it doesn't have an unknown address
+iptables -t nat -A POSTROUTING -o "$1" -j SNAT --to-source "$4"
+
+# add routing commands
+ip route add "$remote_network"/"$remote_netmask_bits" via "$5" table vpn.out
+ip route add table vpnonly.out via "$5"
+ip route flush cache