Read this in other languages: English, 简体中文.
- Use alternative DNS servers
- DNS name and server IP changes
- IKEv2-only VPN
- Internal VPN IPs and traffic
- Port forwarding to VPN clients
- Split tunneling
- Access VPN server's subnet
- Modify IPTables rules
- Deploy Google BBR congestion control algorithm
Clients are set to use Google Public DNS when the VPN is active. If another DNS provider is preferred, you may replace 8.8.8.8
and 8.8.4.4
in these files: /etc/ppp/options.xl2tpd
, /etc/ipsec.conf
and /etc/ipsec.d/ikev2.conf
(if exists). Then run service ipsec restart
and service xl2tpd restart
.
Advanced users can define VPN_DNS_SRV1
and optionally VPN_DNS_SRV2
when running the VPN setup script and the IKEv2 helper script. For example, if you want to use Cloudflare's DNS service:
sudo VPN_DNS_SRV1=1.1.1.1 VPN_DNS_SRV2=1.0.0.1 sh vpn.sh
sudo VPN_DNS_SRV1=1.1.1.1 VPN_DNS_SRV2=1.0.0.1 ikev2.sh --auto
In certain circumstances, you may want VPN clients to use the specified DNS server(s) only for resolving internal domain name(s), and use their locally configured DNS servers to resolve all other domain names. This can be configured using the modecfgdomains
option, e.g. modecfgdomains="internal.example.com, home"
. Add this option to section conn ikev2-cp
in /etc/ipsec.d/ikev2.conf
for IKEv2, and to section conn xauth-psk
in /etc/ipsec.conf
for IPsec/XAuth ("Cisco IPsec"). Then run service ipsec restart
. IPsec/L2TP mode does not support this option.
For IPsec/L2TP and IPsec/XAuth ("Cisco IPsec") modes, you may use a DNS name (e.g. vpn.example.com
) instead of an IP address to connect to the VPN server, without additional configuration. In addition, the VPN should generally continue to work after server IP changes, such as after restoring a snapshot to a new server with a different IP, although a reboot may be required.
For IKEv2 mode, if you want the VPN to continue to work after server IP changes, read this section. Alternatively, you may specify a DNS name to be used as the VPN server's address when setting up IKEv2. The DNS name must be a fully qualified domain name (FQDN). It will be included in the generated server certificate. Example:
sudo VPN_DNS_NAME='vpn.example.com' ikev2.sh --auto
Alternatively, you may customize IKEv2 setup options by running the helper script without the --auto
parameter.
Using Libreswan 4.2 or newer, advanced users can enable IKEv2-only mode on the VPN server. With IKEv2-only mode enabled, VPN clients can only connect to the VPN server using IKEv2. All IKEv1 connections (including IPsec/L2TP and IPsec/XAuth ("Cisco IPsec") modes) will be dropped.
To enable IKEv2-only mode, first install the VPN server and set up IKEv2 using instructions in the README. Then run the helper script and follow the prompts.
# Download the script
wget -nv -O ikev2onlymode.sh https://bit.ly/ikev2onlymode
# Run the script and follow the prompts
sudo bash ikev2onlymode.sh
To disable IKEv2-only mode, run the helper script again and select the appropriate option.
Alternatively, you may manually enable IKEv2-only mode. Click here for details.
Alternatively, you may manually enable IKEv2-only mode. First check Libreswan version using ipsec --version
, and update Libreswan if needed. Then edit /etc/ipsec.conf
on the VPN server. Append ikev1-policy=drop
to the end of the config setup
section, indented by two spaces. Save the file and run service ipsec restart
. When finished, you can run ipsec status
to verify that only the ikev2-cp
connection is enabled.
When connecting using IPsec/L2TP mode, the VPN server has internal IP 192.168.42.1
within the VPN subnet 192.168.42.0/24
. Clients are assigned internal IPs from 192.168.42.10
to 192.168.42.250
. To check which IP is assigned to a client, view the connection status on the VPN client.
When connecting using IPsec/XAuth ("Cisco IPsec") or IKEv2 mode, the VPN server does NOT have an internal IP within the VPN subnet 192.168.43.0/24
. Clients are assigned internal IPs from 192.168.43.10
to 192.168.43.250
.
You may use these internal VPN IPs for communication. However, note that the IPs assigned to VPN clients are dynamic, and firewalls on client devices may block such traffic.
Advanced users may optionally assign static IPs to VPN clients. Expand for details.
IPsec/L2TP mode: Assign static IPs to VPN clients
The example below ONLY applies to IPsec/L2TP mode. Commands must be run as root
.
-
First, create a new VPN user for each VPN client that you want to assign a static IP to. Refer to Manage VPN Users. Helper scripts are included for convenience.
-
Edit
/etc/xl2tpd/xl2tpd.conf
on the VPN server. Replaceip range = 192.168.42.10-192.168.42.250
with e.g.ip range = 192.168.42.100-192.168.42.250
. This reduces the pool of auto-assigned IP addresses, so that more IPs are available to assign to clients as static IPs. -
Edit
/etc/ppp/chap-secrets
on the VPN server. For example, if the file contains:"username1" l2tpd "password1" * "username2" l2tpd "password2" * "username3" l2tpd "password3" *
Let's assume that you want to assign static IP
192.168.42.2
to VPN userusername2
, assign static IP192.168.42.3
to VPN userusername3
, while keepingusername1
unchanged (auto-assign from the pool). After editing, the file should look like:"username1" l2tpd "password1" * "username2" l2tpd "password2" 192.168.42.2 "username3" l2tpd "password3" 192.168.42.3
Note: The assigned static IP(s) must be from the subnet
192.168.42.0/24
, and must NOT be from the pool of auto-assigned IPs (seeip range
above). In addition,192.168.42.1
is reserved for the VPN server itself. In the example above, you can only assign static IP(s) from the range192.168.42.2-192.168.42.99
. -
(Important) Restart the xl2tpd service:
service xl2tpd restart
IPsec/XAuth ("Cisco IPsec") mode: Assign static IPs to VPN clients
The example below ONLY applies to IPsec/XAuth ("Cisco IPsec") mode. Commands must be run as root
.
-
First, create a new VPN user for each VPN client that you want to assign a static IP to. Refer to Manage VPN Users. Helper scripts are included for convenience.
-
Edit
/etc/ipsec.conf
on the VPN server. Replacerightaddresspool=192.168.43.10-192.168.43.250
with e.g.rightaddresspool=192.168.43.100-192.168.43.250
. This reduces the pool of auto-assigned IP addresses, so that more IPs are available to assign to clients as static IPs. -
Edit
/etc/ipsec.d/ikev2.conf
on the VPN server (if exists). Replacerightaddresspool=192.168.43.10-192.168.43.250
with the same value as the previous step. -
Edit
/etc/ipsec.d/passwd
on the VPN server. For example, if the file contains:username1:password1hashed:xauth-psk username2:password2hashed:xauth-psk username3:password3hashed:xauth-psk
Let's assume that you want to assign static IP
192.168.43.2
to VPN userusername2
, assign static IP192.168.43.3
to VPN userusername3
, while keepingusername1
unchanged (auto-assign from the pool). After editing, the file should look like:username1:password1hashed:xauth-psk username2:password2hashed:xauth-psk:192.168.42.2 username3:password3hashed:xauth-psk:192.168.42.3
Note: The assigned static IP(s) must be from the subnet
192.168.43.0/24
, and must NOT be from the pool of auto-assigned IPs (seerightaddresspool
above). In the example above, you can only assign static IP(s) from the range192.168.43.1-192.168.43.99
. -
(Important) Restart the IPsec service:
service ipsec restart
IKEv2 mode: Assign static IPs to VPN clients
The example below ONLY applies to IKEv2 mode. Commands must be run as root
.
-
First, create a new IKEv2 client certificate for each client that you want to assign a static IP to, and write down the name of each IKEv2 client. Refer to Add a client certificate.
-
Edit
/etc/ipsec.d/ikev2.conf
on the VPN server. Replacerightaddresspool=192.168.43.10-192.168.43.250
with e.g.rightaddresspool=192.168.43.100-192.168.43.250
. This reduces the pool of auto-assigned IP addresses, so that more IPs are available to assign to clients as static IPs. -
Edit
/etc/ipsec.conf
on the VPN server. Replacerightaddresspool=192.168.43.10-192.168.43.250
with the same value as the previous step. -
Edit
/etc/ipsec.d/ikev2.conf
on the VPN server again. For example, if the file contains:conn ikev2-cp left=%defaultroute ... ...
Let's assume that you want to assign static IP
192.168.43.4
to IKEv2 clientclient1
, assign static IP192.168.43.5
to clientclient2
, while keeping other clients unchanged (auto-assign from the pool). After editing, the file should look like:conn ikev2-cp left=%defaultroute ... ... conn client1 rightid=@client1 rightaddresspool=192.168.43.4-192.168.43.4 also=ikev2-cp conn client2 rightid=@client2 rightaddresspool=192.168.43.5-192.168.43.5 also=ikev2-cp
Note: Add a new
conn
section for each client that you want to assign a static IP to. You must add a@
prefix to the client name forrightid=
. The client name must exactly match the name you specified when adding the client certificate. The assigned static IP(s) must be from the subnet192.168.43.0/24
, and must NOT be from the pool of auto-assigned IPs (seerightaddresspool
above). In the example above, you can only assign static IP(s) from the range192.168.43.1-192.168.43.99
. -
(Important) Restart the IPsec service:
service ipsec restart
Client-to-client traffic is allowed by default. If you want to disallow client-to-client traffic, run the following commands on the VPN server. Add them to /etc/rc.local
to persist after reboot.
iptables -I FORWARD 2 -i ppp+ -o ppp+ -s 192.168.42.0/24 -d 192.168.42.0/24 -j DROP
iptables -I FORWARD 3 -s 192.168.43.0/24 -d 192.168.43.0/24 -j DROP
iptables -I FORWARD 4 -i ppp+ -d 192.168.43.0/24 -j DROP
iptables -I FORWARD 5 -s 192.168.43.0/24 -o ppp+ -j DROP
In certain circumstances, you may want to forward port(s) on the VPN server to a connected VPN client. This can be done by adding IPTables rules on the VPN server.
Warning: Port forwarding will expose port(s) on the VPN client to the entire Internet, which could be a security risk! This is NOT recommended, unless your use case requires it.
Note: The internal VPN IPs assigned to VPN clients are dynamic, and firewalls on client devices may block forwarded traffic. To assign static IPs to VPN clients, refer to the previous section. To check which IP is assigned to a client, view the connection status on the VPN client.
Example 1: Forward TCP port 443 on the VPN server to the IPsec/L2TP client at 192.168.42.10
.
# Get default network interface name
netif=$(route 2>/dev/null | grep -m 1 '^default' | grep -o '[^ ]*$')
iptables -I FORWARD 2 -i "$netif" -o ppp+ -p tcp --dport 443 -j ACCEPT
iptables -t nat -A PREROUTING -p tcp --dport 443 -j DNAT --to 192.168.42.10
Example 2: Forward UDP port 123 on the VPN server to the IKEv2 (or IPsec/XAuth) client at 192.168.43.10
.
# Get default network interface name
netif=$(route 2>/dev/null | grep -m 1 '^default' | grep -o '[^ ]*$')
iptables -I FORWARD 2 -i "$netif" -d 192.168.43.0/24 -p udp --dport 123 -j ACCEPT
iptables -t nat -A PREROUTING -p udp --dport 123 -j DNAT --to 192.168.43.10
If you want the rules to persist after reboot, you may add these commands to /etc/rc.local
. To remove the added IPTables rules, run the commands again, but replace -I FORWARD 2
with -D FORWARD
, and replace -A PREROUTING
with -D PREROUTING
.
With split tunneling, VPN clients will only send traffic for specific destination subnet(s) through the VPN tunnel. Other traffic will NOT go through the VPN tunnel. Split tunneling has some limitations, and is not supported by all VPN clients.
Advanced users can optionally enable split tunneling for the IPsec/XAuth ("Cisco IPsec") and/or IKEv2 modes. Expand for details. IPsec/L2TP mode does NOT support this feature.
IPsec/XAuth ("Cisco IPsec") mode: Enable split tunneling
The example below ONLY applies to IPsec/XAuth ("Cisco IPsec") mode. Commands must be run as root
.
- Edit
/etc/ipsec.conf
on the VPN server. In the sectionconn xauth-psk
, replaceleftsubnet=0.0.0.0/0
with the subnet(s) you want VPN clients to send traffic through the VPN tunnel. For example:
For a single subnet:For multiple subnets (useleftsubnet=10.123.123.0/24
leftsubnets
instead):leftsubnets="10.123.123.0/24,10.100.0.0/16"
- (Important) Restart the IPsec service:
service ipsec restart
IKEv2 mode: Enable split tunneling
The example below ONLY applies to IKEv2 mode. Commands must be run as root
.
- Edit
/etc/ipsec.d/ikev2.conf
on the VPN server. In the sectionconn ikev2-cp
, replaceleftsubnet=0.0.0.0/0
with the subnet(s) you want VPN clients to send traffic through the VPN tunnel. For example:
For a single subnet:For multiple subnets (useleftsubnet=10.123.123.0/24
leftsubnets
instead):leftsubnets="10.123.123.0/24,10.100.0.0/16"
- (Important) Restart the IPsec service:
service ipsec restart
After connecting to the VPN, VPN clients can generally access services running on other devices that are within the same local subnet as the VPN server, without additional configuration. For example, if the VPN server's local subnet is 192.168.0.0/24
, and an Nginx server is running on IP 192.168.0.2
, VPN clients can use IP 192.168.0.2
to access the Nginx server.
Please note, additional configuration is required if the VPN server has multiple network interfaces (e.g. eth0
and eth1
), and you want VPN clients to access the local subnet behind the network interface that is NOT for Internet access. In this scenario, you must run the following commands to add IPTables rules. To persist after reboot, you may add these commands to /etc/rc.local
.
# Replace eth1 with the name of the network interface
# on the VPN server that you want VPN clients to access
netif=eth1
iptables -I FORWARD 2 -i "$netif" -o ppp+ -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT
iptables -I FORWARD 2 -i ppp+ -o "$netif" -j ACCEPT
iptables -I FORWARD 2 -i "$netif" -d 192.168.43.0/24 -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT
iptables -I FORWARD 2 -s 192.168.43.0/24 -o "$netif" -j ACCEPT
iptables -t nat -I POSTROUTING -s 192.168.43.0/24 -o "$netif" -m policy --dir out --pol none -j MASQUERADE
iptables -t nat -I POSTROUTING -s 192.168.42.0/24 -o "$netif" -j MASQUERADE
If you want to modify the IPTables rules after install, edit /etc/iptables.rules
and/or /etc/iptables/rules.v4
(Ubuntu/Debian), or /etc/sysconfig/iptables
(CentOS/RHEL). Then reboot your server.
Note: If using Rocky Linux, AlmaLinux or CentOS/RHEL 8 and firewalld was active during VPN setup, nftables may be configured. In this case, edit /etc/sysconfig/nftables.conf
instead of /etc/sysconfig/iptables
.
After the VPN server is set up, the performance can be improved by deploying the Google BBR congestion control algorithm.
This is usually done by modifying the configuration file /etc/sysctl.conf
. However, some Linux distributions may additionally require updates to the Linux kernel.
For detailed deployment methods, please refer to this document.
Copyright (C) 2021-2022 Lin Song
This work is licensed under the Creative Commons Attribution-ShareAlike 3.0 Unported License
Attribution required: please include my name in any derivative and let me know how you have improved it!