OpenVPN is an open source VPN daemon by James Yonan. Because OpenVPN tries to be a universal VPN tool offering a great deal of flexibility, there are a lot of options on this manual page. If you're new to OpenVPN, you might want to skip ahead to the examples section where you will see how to construct simple VPNs on the command line without even needing a configuration file.
Also note that there's more documentation and examples on the OpenVPN web site: http://openvpn.net/
And if you would like to see a shorter version of this manual, see the openvpn usage message which can be obtained by running openvpn without any parameters.
OpenVPN is a robust and highly flexible VPN daemon. OpenVPN supports SSL/TLS security, ethernet bridging, TCP or UDP tunnel transport through proxies or NAT, support for dynamic IP addresses and DHCP, scalability to hundreds or thousands of users, and portability to most major OS platforms.
OpenVPN is tightly bound to the OpenSSL library, and derives much of its crypto capabilities from it.
OpenVPN supports conventional encryption using a pre-shared secret key (Static Key mode) or public key security (SSL/TLS mode) using client & server certificates. OpenVPN also supports non-encrypted TCP/UDP tunnels.
OpenVPN is designed to work with the TUN/TAP virtual networking interface that exists on most platforms.
Overall, OpenVPN aims to offer many of the key features of IPSec but with a relatively lightweight footprint.
If --config file is the only option to the openvpn command, the --config can be removed, and the command can be given as openvpn file
Note that configuration files can be nested to a reasonable depth.
Double quotation or single quotation characters ("", '') can be used to enclose single parameters containing whitespace, and "#" or ";" characters in the first column can be used to denote comments.
Note that OpenVPN 2.0 and higher performs backslash-based shell escaping for characters not in single quotations, so the following mappings should be observed:
\\ Maps to a single backslash character (\).
\" Pass a literal doublequote character ("), don't
interpret it as enclosing a parameter.
\[SPACE] Pass a literal space or tab character, don't
interpret it as a parameter delimiter.
For example on Windows, use double backslashes to represent pathnames:
secret "c:\\OpenVPN\\secret.key"
For examples of configuration files, see http://openvpn.net/examples.html
Here is an example configuration file:
# # Sample OpenVPN configuration file for # using a pre-shared static key. # # '#' or ';' may be used to delimit comments. # Use a dynamic tun device. dev tun # Our remote peer remote mypeer.mydomain # 10.1.0.1 is our local VPN endpoint # 10.1.0.2 is our remote VPN endpoint ifconfig 10.1.0.1 10.1.0.2 # Our pre-shared static key secret static.key
The OpenVPN client will try to connect to a server at host:port in the order specified by the list of --remote options.
proto indicates the protocol to use when connecting with the remote, and may be "tcp" or "udp".
The client will move on to the next host in the list, in the event of connection failure. Note that at any given time, the OpenVPN client will at most be connected to one server.
Note that since UDP is connectionless, connection failure is defined by the --ping and --ping-restart options.
Note the following corner case: If you use multiple --remote options, AND you are dropping root privileges on the client with --user and/or --group, AND the client is running a non-Windows OS, if the client needs to switch to a different server, and that server pushes back different TUN/TAP or route settings, the client may lack the necessary privileges to close and reopen the TUN/TAP interface. This could cause the client to exit with a fatal error.
If --remote is unspecified, OpenVPN will listen for packets from any IP address, but will not act on those packets unless they pass all authentication tests. This requirement for authentication is binding on all potential peers, even those from known and supposedly trusted IP addresses (it is very easy to forge a source IP address on a UDP packet).
When used in TCP mode, --remote will act as a filter, rejecting connections from any host which does not match host.
If host is a DNS name which resolves to multiple IP addresses, one will be randomly chosen, providing a sort of basic load-balancing and failover capability.
An OpenVPN client will try each connection profile sequentially until it achieves a successful connection.
--remote-random can be used to initially "scramble" the connection list.
Here is an example of connection profile usage:
client dev tun <connection> remote 198.19.34.56 1194 udp </connection> <connection> remote 198.19.34.56 443 tcp </connection> <connection> remote 198.19.34.56 443 tcp http-proxy 192.168.0.8 8080 http-proxy-retry </connection> <connection> remote 198.19.36.99 443 tcp http-proxy 192.168.0.8 8080 http-proxy-retry </connection> persist-key persist-tun pkcs12 client.p12 ns-cert-type server verb 3
First we try to connect to a server at 198.19.34.56:1194 using UDP. If that fails, we then try to connect to 198.19.34.56:443 using TCP. If that also fails, then try connecting through an HTTP proxy at 192.168.0.8:8080 to 198.19.34.56:443 using TCP. Finally, try to connect through the same proxy to a server at 198.19.36.99:443 using TCP.
The following OpenVPN options may be used inside of a <connection> block:
bind, connect-retry, connect-retry-max, connect-timeout, float, http-proxy, http-proxy-option, http-proxy-retry, http-proxy-timeout, local, lport, nobind, port, proto, remote, rport, socks-proxy, and socks-proxy-retry.
A defaulting mechanism exists for specifying options to apply to all <connection> profiles. If any of the above options (with the exception of remote ) appear outside of a <connection> block, but in a configuration file which has one or more <connection> blocks, the option setting will be used as a default for <connection> blocks which follow it in the configuration file.
For example, suppose the nobind option were placed in the sample configuration file above, near the top of the file, before the first <connection> block. The effect would be as if nobind were declared in all <connection> blocks below it.
The default protocol is udp when --proto is not specified.
For UDP operation, --proto udp should be specified on both peers.
For TCP operation, one peer must use --proto tcp-server and the other must use --proto tcp-client. A peer started with tcp-server will wait indefinitely for an incoming connection. A peer started with tcp-client will attempt to connect, and if that fails, will sleep for 5 seconds (adjustable via the --connect-retry option) and try again infinite or up to N retries (adjustable via the --connect-retry-max option). Both TCP client and server will simulate a SIGUSR1 restart signal if either side resets the connection.
OpenVPN is designed to operate optimally over UDP, but TCP capability is provided for situations where UDP cannot be used. In comparison with UDP, TCP will usually be somewhat less efficient and less robust when used over unreliable or congested networks.
This article outlines some of problems with tunneling IP over TCP:
http://sites.inka.de/sites/bigred/devel/tcp-tcp.html
There are certain cases, however, where using TCP may be advantageous from a security and robustness perspective, such as tunneling non-IP or application-level UDP protocols, or tunneling protocols which don't possess a built-in reliability layer.
auth-method should be one of "none", "basic", or "ntlm".
The auto flag causes OpenVPN to automatically determine the auth-method and query stdin or the management interface for username/password credentials, if required. This flag exists on OpenVPN 2.1 or higher.
VERSION version -- Set HTTP version number to version (default=1.0).
AGENT user-agent -- Set HTTP "User-Agent" string to user-agent.
Set n to "infinite" to retry indefinitely.
By default, --resolv-retry infinite is enabled. You can disable by setting n=0.
Essentially, --float tells OpenVPN to accept authenticated packets from any address, not only the address which was specified in the --remote option.
Execute as:
cmd ip_address port_number
Don't use --ipchange in --mode server mode. Use a --client-connect script instead.
See the "Environmental Variables" section below for additional parameters passed as environmental variables.
Note that cmd can be a shell command with multiple arguments, in which case all OpenVPN-generated arguments will be appended to cmd to build a command line which will be passed to the script.
If you are running in a dynamic IP address environment where the IP addresses of either peer could change without notice, you can use this script, for example, to edit the /etc/hosts file with the current address of the peer. The script will be run every time the remote peer changes its IP address.
Similarly if our IP address changes due to DHCP, we should configure our IP address change script (see man page for dhcpcd(8) ) to deliver a SIGHUP or SIGUSR1 signal to OpenVPN. OpenVPN will then reestablish a connection with its most recently authenticated peer on its new IP address.
See examples section below for an example on setting up a TUN device.
You must use either tun devices on both ends of the connection or tap devices on both ends. You cannot mix them, as they represent different underlying network layers.
tun devices encapsulate IPv4 or IPv6 (OSI Layer 3) while tap devices encapsulate Ethernet 802.3 (OSI Layer 2).
If you set this directive on the server, the --server and --server-bridge directives will automatically push your chosen topology setting to clients as well. This directive can also be manually pushed to clients. Like the --dev directive, this directive must always be compatible between client and server.
mode can be one of:
net30 -- Use a point-to-point topology, by allocating one /30 subnet per client. This is designed to allow point-to-point semantics when some or all of the connecting clients might be Windows systems. This is the default on OpenVPN 2.0.
p2p -- Use a point-to-point topology where the remote endpoint of the client's tun interface always points to the local endpoint of the server's tun interface. This mode allocates a single IP address per connecting client. Only use when none of the connecting clients are Windows systems. This mode is functionally equivalent to the --ifconfig-pool-linear directive which is available in OpenVPN 2.0 and is now deprecated.
subnet -- Use a subnet rather than a point-to-point topology by configuring the tun interface with a local IP address and subnet mask, similar to the topology used in --dev tap and ethernet bridging mode. This mode allocates a single IP address per connecting client and works on Windows as well. Only available when server and clients are OpenVPN 2.1 or higher, or OpenVPN 2.0.x which has been manually patched with the --topology directive code. When used on Windows, requires version 8.2 or higher of the TAP-Win32 driver. When used on *nix, requires that the tun driver supports an ifconfig(8) command which sets a subnet instead of a remote endpoint IP address.
This option exists in OpenVPN 2.1 or higher.
On Windows systems, select the TAP-Win32 adapter which is named node in the Network Connections Control Panel or the raw GUID of the adapter enclosed by braces. The --show-adapters option under Windows can also be used to enumerate all available TAP-Win32 adapters and will show both the network connections control panel name and the GUID for each TAP-Win32 adapter.
For TUN devices, which facilitate virtual point-to-point IP connections, the proper usage of --ifconfig is to use two private IP addresses which are not a member of any existing subnet which is in use. The IP addresses may be consecutive and should have their order reversed on the remote peer. After the VPN is established, by pinging rn, you will be pinging across the VPN.
For TAP devices, which provide the ability to create virtual ethernet segments, --ifconfig is used to set an IP address and subnet mask just as a physical ethernet adapter would be similarly configured. If you are attempting to connect to a remote ethernet bridge, the IP address and subnet should be set to values which would be valid on the the bridged ethernet segment (note also that DHCP can be used for the same purpose).
This option, while primarily a proxy for the ifconfig(8) command, is designed to simplify TUN/TAP tunnel configuration by providing a standard interface to the different ifconfig implementations on different platforms.
--ifconfig parameters which are IP addresses can also be specified as a DNS or /etc/hosts file resolvable name.
For TAP devices, --ifconfig should not be used if the TAP interface will be getting an IP address lease from a DHCP server.
For example, if you have a configuration where the local host uses --ifconfig but the remote host does not, use --ifconfig-nowarn on the local host.
This option will also silence warnings about potential address conflicts which occasionally annoy more experienced users by triggering "false positive" warnings.
This option is intended as a convenience proxy for the route(8) shell command, while at the same time providing portable semantics across OpenVPN's platform space.
netmask default -- 255.255.255.255
gateway default -- taken from --route-gateway or the second parameter to --ifconfig when --dev tun is specified.
metric default -- taken from --route-metric otherwise 0.
The default can be specified by leaving an option blank or setting it to "default".
The network and gateway parameters can also be specified as a DNS or /etc/hosts file resolvable name, or as one of three special keywords:
vpn_gateway -- The remote VPN endpoint address (derived either from --route-gateway or the second parameter to --ifconfig when --dev tun is specified).
net_gateway -- The pre-existing IP default gateway, read from the routing table (not supported on all OSes).
remote_host -- The --remote address if OpenVPN is being run in client mode, and is undefined in server mode.
If dhcp is specified as the parameter, the gateway address will be extracted from a DHCP negotiation with the OpenVPN server-side LAN.
This option is designed to be useful in scenarios where DHCP is used to set tap adapter addresses. The delay will give the DHCP handshake time to complete before routes are added.
On Windows, --route-delay tries to be more intelligent by waiting w seconds (w=30 by default) for the TAP-Win32 adapter to come up before adding routes.
See the "Environmental Variables" section below for additional parameters passed as environmental variables.
Note that cmd can be a shell command with multiple arguments.
When used on the client, this option effectively bars the server from adding routes to the client's routing table, however note that this option still allows the server to set the TCP/IP properties of the client's TUN/TAP interface.
This option performs three steps:
(1) Create a static route for the --remote address which forwards to the pre-existing default gateway. This is done so that (3) will not create a routing loop.
(2) Delete the default gateway route.
(3) Set the new default gateway to be the VPN endpoint address (derived either from --route-gateway or the second parameter to --ifconfig when --dev tun is specified).
When the tunnel is torn down, all of the above steps are reversed so that the original default route is restored.
Option flags:
local -- Add the local flag if both OpenVPN servers are directly connected via a common subnet, such as with wireless. The local flag will cause step 1 above to be omitted.
def1 -- Use this flag to override the default gateway by using 0.0.0.0/1 and 128.0.0.0/1 rather than 0.0.0.0/0. This has the benefit of overriding but not wiping out the original default gateway.
bypass-dhcp -- Add a direct route to the DHCP server (if it is non-local) which bypasses the tunnel (Available on Windows clients, may not be available on non-Windows clients).
bypass-dns -- Add a direct route to the DNS server(s) (if they are non-local) which bypasses the tunnel (Available on Windows clients, may not be available on non-Windows clients).
Using the def1 flag is highly recommended.
The MTU (Maximum Transmission Units) is the maximum datagram size in bytes that can be sent unfragmented over a particular network path. OpenVPN requires that packets on the control or data channels be sent unfragmented.
MTU problems often manifest themselves as connections which hang during periods of active usage.
It's best to use the --fragment and/or --mssfix options to deal with MTU sizing issues.
'no'
-- Never send DF (Don't Fragment) frames
'maybe'
-- Use per-route hints
'yes'
-- Always DF (Don't Fragment)
The max parameter is interpreted in the same way as the --link-mtu parameter, i.e. the UDP packet size after encapsulation overhead has been added in, but not including the UDP header itself.
The --fragment option only makes sense when you are using the UDP protocol ( --proto udp ).
--fragment adds 4 bytes of overhead per datagram.
See the --mssfix option below for an important related option to --fragment.
It should also be noted that this option is not meant to replace UDP fragmentation at the IP stack level. It is only meant as a last resort when path MTU discovery is broken. Using this option is less efficient than fixing path MTU discovery for your IP link and using native IP fragmentation instead.
Having said that, there are circumstances where using OpenVPN's internal fragmentation capability may be your only option, such as tunneling a UDP multicast stream which requires fragmentation.
The max parameter is interpreted in the same way as the --link-mtu parameter, i.e. the UDP packet size after encapsulation overhead has been added in, but not including the UDP header itself.
The --mssfix option only makes sense when you are using the UDP protocol for OpenVPN peer-to-peer communication, i.e. --proto udp.
--mssfix and --fragment can be ideally used together, where --mssfix will try to keep TCP from needing packet fragmentation in the first place, and if big packets come through anyhow (from protocols other than TCP), --fragment will internally fragment them.
Both --fragment and --mssfix are designed to work around cases where Path MTU discovery is broken on the network path between OpenVPN peers.
The usual symptom of such a breakdown is an OpenVPN connection which successfully starts, but then stalls during active usage.
If --fragment and --mssfix are used together, --mssfix will take its default max parameter from the --fragment max option.
Therefore, one could lower the maximum UDP packet size to 1300 (a good first try for solving MTU-related connection problems) with the following options:
--tun-mtu 1500 --fragment 1300 --mssfix
The TCP_NODELAY socket flag is useful in TCP mode, and causes the kernel to send tunnel packets immediately over the TCP connection without trying to group several smaller packets into a larger packet. This can result in a considerably improvement in latency.
This option is pushable from server to client, and should be used on both client and server for maximum effect.
OpenVPN uses the following algorithm to implement traffic shaping: Given a shaper rate of n bytes per second, after a datagram write of b bytes is queued on the TCP/UDP port, wait a minimum of (b / n) seconds before queuing the next write.
It should be noted that OpenVPN supports multiple tunnels between the same two peers, allowing you to construct full-speed and reduced bandwidth tunnels at the same time, routing low-priority data such as off-site backups over the reduced bandwidth tunnel, and other data over the full-speed tunnel.
Also note that for low bandwidth tunnels (under 1000 bytes per second), you should probably use lower MTU values as well (see above), otherwise the packet latency will grow so large as to trigger timeouts in the TLS layer and TCP connections running over the tunnel.
OpenVPN allows n to be between 100 bytes/sec and 100 Mbytes/sec.
If the optional bytes parameter is included, exit after n seconds of activity on tun/tap device produces a combined in/out byte count that is less than bytes.
This option has two intended uses:
(1) Compatibility with stateful firewalls. The periodic ping will ensure that a stateful firewall rule which allows OpenVPN UDP packets to pass will not time out.
(2) To provide a basis for the remote to test the existence of its peer using the --ping-exit option.
For example,
openvpn [options...] --inactive 3600 --ping 10 --ping-exit 60
when used on both peers will cause OpenVPN to exit within 60 seconds if its peer disconnects, but will exit after one hour if no actual tunnel data is exchanged.
This option is useful in cases where the remote peer has a dynamic IP address and a low-TTL DNS name is used to track the IP address using a service such as http://dyndns.org/ + a dynamic DNS client such as ddclient.
If the peer cannot be reached, a restart will be triggered, causing the hostname used with --remote to be re-resolved (if --resolv-retry is also specified).
In server mode, --ping-restart, --inactive, or any other type of internally generated signal will always be applied to individual client instance objects, never to whole server itself. Note also in server mode that any internally generated signal which would normally cause a restart, will cause the deletion of the client instance object instead.
In client mode, the --ping-restart parameter is set to 120 seconds by default. This default will hold until the client pulls a replacement value from the server, based on the --keepalive setting in the server configuration. To disable the 120 second default, set --ping-restart 0 on the client.
See the signals section below for more information on SIGUSR1.
Note that the behavior of SIGUSR1 can be modified by the --persist-tun, --persist-key, --persist-local-ip, and --persist-remote-ip options.
Also note that --ping-exit and --ping-restart are mutually exclusive and cannot be used together.
For example, --keepalive 10 60 expands as follows:
if mode server: ping 10 ping-restart 120 push "ping 10" push "ping-restart 60" else ping 10 ping-restart 60
SIGUSR1 is a restart signal similar to SIGHUP, but which offers finer-grained control over reset options.
This option can be combined with --user nobody to allow restarts triggered by the SIGUSR1 signal. Normally if you drop root privileges in OpenVPN, the daemon cannot be restarted since it will now be unable to re-read protected key files.
This option solves the problem by persisting keys across SIGUSR1 resets, so they don't need to be re-read.
Using this option ensures that key material and tunnel data are never written to disk due to virtual memory paging operations which occur under most modern operating systems. It ensures that even if an attacker was able to crack the box running OpenVPN, he would not be able to scan the system swap file to recover previously used ephemeral keys, which are used for a period of time governed by the --reneg options (see below), then are discarded.
The downside of using --mlock is that it will reduce the amount of physical memory available to other applications.
For --dev tun execute as:
cmd tun_dev tun_mtu link_mtu ifconfig_local_ip ifconfig_remote_ip [ init | restart ]
For --dev tap execute as:
cmd tap_dev tap_mtu link_mtu ifconfig_local_ip ifconfig_netmask [ init | restart ]
See the "Environmental Variables" section below for additional parameters passed as environmental variables.
Note that cmd can be a shell command with multiple arguments, in which case all OpenVPN-generated arguments will be appended to cmd to build a command line which will be passed to the shell.
Typically, cmd will run a script to add routes to the tunnel.
Normally the up script is called after the TUN/TAP device is opened. In this context, the last command line parameter passed to the script will be init. If the --up-restart option is also used, the up script will be called for restarts as well. A restart is considered to be a partial reinitialization of OpenVPN where the TUN/TAP instance is preserved (the --persist-tun option will enable such preservation). A restart can be generated by a SIGUSR1 signal, a --ping-restart timeout, or a connection reset when the TCP protocol is enabled with the --proto option. If a restart occurs, and --up-restart has been specified, the up script will be called with restart as the last parameter.
The following standalone example shows how the --up script can be called in both an initialization and restart context. (NOTE: for security reasons, don't run the following example unless UDP port 9999 is blocked by your firewall. Also, the example will run indefinitely, so you should abort with control-c).
openvpn --dev tun --port 9999 --verb 4 --ping-restart 10 --up 'echo up' --down 'echo down' --persist-tun --up-restart
Note that OpenVPN also provides the --ifconfig option to automatically ifconfig the TUN device, eliminating the need to define an --up script, unless you also want to configure routes in the --up script.
If --ifconfig is also specified, OpenVPN will pass the ifconfig local and remote endpoints on the command line to the --up script so that they can be used to configure routes such as:
route add -net 10.0.0.0 netmask 255.255.255.0 gw $5
In --proto udp mode, this option normally requires the use of --ping to allow connection initiation to be sensed in the absence of tunnel data, since UDP is a "connectionless" protocol.
On Windows, this option will delay the TAP-Win32 media state transitioning to "connected" until connection establishment, i.e. the receipt of the first authenticated packet from the peer.
Note that if you reduce privileges by using --user and/or --group, your --down script will also run at reduced privilege.
This option should be used with caution, as there are good security reasons for having OpenVPN fail if it detects problems in a config file. Having said that, there are valid reasons for wanting new software features to gracefully degrade when encountered by older software versions.
This directive is designed to be pushed by the server to clients, and the prepending of "OPENVPN_" to the environmental variable is a safety precaution to prevent a LD_PRELOAD style attack from a malicious or compromised server.
0 --
Strictly no calling of external programs.
1 --
(Default) Only call built-in executables such as ifconfig, ip, route, or netsh.
2 --
Allow calling of built-in executables and user-defined scripts.
3 --
Allow passwords to be passed to scripts via environmental variables (potentially unsafe).
The method parameter indicates how OpenVPN should call external commands and scripts. Settings for method:
execve --
(default) Use execve() function on Unix family OSes and CreateProcess() on Windows.
system --
Use system() function (deprecated and less safe since the external program command
line is subject to shell expansion).
The --script-security option was introduced in OpenVPN 2.1_rc9. For configuration file compatibility with previous OpenVPN versions, use: --script-security 3 system
Use of this option is discouraged, but is provided as a temporary fix in situations where a recent version of OpenVPN must connect to an old version.
By setting user to nobody or somebody similarly unprivileged, the hostile party would be limited in what damage they could cause. Of course once you take away privileges, you cannot return them to an OpenVPN session. This means, for example, that if you want to reset an OpenVPN daemon with a SIGUSR1 signal (for example in response to a DHCP reset), you should make use of one or more of the --persist options to ensure that OpenVPN doesn't need to execute any privileged operations in order to restart (such as re-reading key files or running ifconfig on the TUN device).
This option is useful when you are running OpenVPN in --daemon mode, and you want to consolidate all of your OpenVPN control files in one location.
Since the chroot operation is delayed until after initialization, most OpenVPN options that reference files will operate in a pre-chroot context.
In many cases, the dir parameter can point to an empty directory, however complications can result when scripts or restarts are executed after the chroot operation.
Since the setcon operation is delayed until after initialization, OpenVPN can be restricted to just network-related system calls, whereas by applying the context before startup (such as the OpenVPN one provided in the SELinux Reference Policies) you will have to allow many things required only during initialization.
Like with chroot, complications can result when scripts or restarts are executed after the setcon operation, which is why you should really consider using the --persist-key and --persist-tun options.
The optional progname parameter will cause OpenVPN to report its program name to the system logger as progname. This can be useful in linking OpenVPN messages in the syslog file with specific tunnels. When unspecified, progname defaults to "openvpn".
When OpenVPN is run with the --daemon option, it will try to delay daemonization until the majority of initialization functions which are capable of generating fatal errors are complete. This means that initialization scripts can test the return status of the openvpn command for a fairly reliable indication of whether the command has correctly initialized and entered the packet forwarding event loop.
In OpenVPN, the vast majority of errors which occur after initialization are non-fatal.
The wait/nowait option must match what is specified in the inetd/xinetd config file. The nowait mode can only be used with --proto tcp-server. The default is wait. The nowait mode can be used to instantiate the OpenVPN daemon as a classic TCP server, where client connection requests are serviced on a single port number. For additional information on this kind of configuration, see the OpenVPN FAQ: http://openvpn.net/faq.html#oneport
This option precludes the use of --daemon, --local, or --remote. Note that this option causes message and error output to be handled in the same way as the --daemon option. The optional progname parameter is also handled exactly as in --daemon.
Also note that in wait mode, each OpenVPN tunnel requires a separate TCP/UDP port and a separate inetd or xinetd entry. See the OpenVPN 1.x HOWTO for an example on using OpenVPN with xinetd: http://openvpn.net/1xhowto.html
Note that on Windows, when OpenVPN is started as a service, logging occurs by default without the need to specify this option.
This option can only be used on non-Windows systems, when --proto udp is specified, and when --shaper is NOT specified.
Designed to be used to send messages to a controlling application which is receiving the OpenVPN log output.
signal can be set to "SIGHUP" or "SIGTERM". By default, no remapping occurs.
0 --
No output except fatal errors.
1 to 4 --
Normal usage range.
5 --
Output
R
and
W
characters to the console for each packet read and write, uppercase is
used for TCP/UDP packets and lowercase is used for TUN/TAP packets.
6 to 11 --
Debug info range (see errlevel.h for additional
information on debug levels).
Status can also be written to the syslog by sending a SIGUSR2 signal.
In a server mode setup, it is possible to selectively turn compression on or off for individual clients.
First, make sure the client-side config file enables selective compression by having at least one --comp-lzo directive, such as --comp-lzo no. This will turn off compression by default, but allow a future directive push from the server to dynamically change the on/off/adaptive setting.
Next in a --client-config-dir file, specify the compression setting for the client, for example:
comp-lzo yes push "comp-lzo yes"
The first line sets the comp-lzo setting for the server side of the link, the second sets the client side.
Adaptive compression tries to optimize the case where you have compression enabled, but you are sending predominantly uncompressible (or pre-compressed) packets over the tunnel, such as an FTP or rsync transfer of a large, compressed file. With adaptive compression, OpenVPN will periodically sample the compression process to measure its efficiency. If the data being sent over the tunnel is already compressed, the compression efficiency will be very low, triggering openvpn to disable compression for a period of time until the next re-sample test.
The management interface can also listen on a unix domain socket, for those platforms that support it. To use a unix domain socket, specify the unix socket pathname in place of IP and set port to 'unix'. While the default behavior is to create a unix domain socket that may be connected to by any process, the --management-client-user and --management-client-group directives can be used to restrict access.
The management interface provides a special mode where the TCP management link can operate over the tunnel itself. To enable this mode, set IP = "tunnel". Tunnel mode will cause the management interface to listen for a TCP connection on the local VPN address of the TUN/TAP interface.
While the management port is designed for programmatic control of OpenVPN by other applications, it is possible to telnet to the port, using a telnet client in "raw" mode. Once connected, type "help" for a list of commands.
For detailed documentation on the management interface, see the management-notes.txt file in the management folder of the OpenVPN source distribution.
It is strongly recommended that IP be set to 127.0.0.1 (localhost) to restrict accessibility of the management server to local clients.
This directive does not affect the --http-proxy username/password. It is always cached.
For more information and examples on how to build OpenVPN plug-in modules, see the README file in the plugin folder of the OpenVPN source distribution.
If you are using an RPM install of OpenVPN, see /usr/lib/openvpn/plugin. The documentation is in doc and the actual plugin modules are in lib.
Multiple plugin modules can be cascaded, and modules can be used in tandem with scripts. The modules will be called by OpenVPN in the order that they are declared in the config file. If both a plugin and script are configured for the same callback, the script will be called last. If the return code of the module/script controls an authentication function (such as tls-verify, auth-user-pass-verify, or client-connect), then every module and script must return success (0) in order for the connection to be authenticated.
For example, --server 10.8.0.0 255.255.255.0 expands as follows:
mode server
tls-server
push "topology [topology]"
if dev tun AND (topology == net30 OR topology == p2p):
ifconfig 10.8.0.1 10.8.0.2
if !nopool:
ifconfig-pool 10.8.0.4 10.8.0.251
route 10.8.0.0 255.255.255.0
if client-to-client:
push "route 10.8.0.0 255.255.255.0"
else if topology == net30:
push "route 10.8.0.1"
if dev tap OR (dev tun AND topology == subnet):
ifconfig 10.8.0.1 255.255.255.0
if !nopool:
ifconfig-pool 10.8.0.2 10.8.0.254 255.255.255.0
push "route-gateway 10.8.0.1"
Don't use --server if you are ethernet bridging. Use --server-bridge instead.
A helper directive similar to --server which is designed to simplify the configuration of OpenVPN's server mode in ethernet bridging configurations.
If --server-bridge is used without any parameters, it will enable a DHCP-proxy mode, where connecting OpenVPN clients will receive an IP address for their TAP adapter from the DHCP server running on the OpenVPN server-side LAN. Note that only clients that support the binding of a DHCP client with the TAP adapter (such as Windows) can support this mode. The optional nogw flag (advanced) indicates that gateway information should not be pushed to the client.
To configure ethernet bridging, you must first use your OS's bridging capability to bridge the TAP interface with the ethernet NIC interface. For example, on Linux this is done with the brctl tool, and with Windows XP it is done in the Network Connections Panel by selecting the ethernet and TAP adapters and right-clicking on "Bridge Connections".
Next you you must manually set the IP/netmask on the bridge interface. The gateway and netmask parameters to --server-bridge can be set to either the IP/netmask of the bridge interface, or the IP/netmask of the default gateway/router on the bridged subnet.
Finally, set aside a IP range in the bridged subnet, denoted by pool-start-IP and pool-end-IP, for OpenVPN to allocate to connecting clients.
For example, server-bridge 10.8.0.4 255.255.255.0 10.8.0.128 10.8.0.254 expands as follows:
mode server tls-server ifconfig-pool 10.8.0.128 10.8.0.254 255.255.255.0 push "route-gateway 10.8.0.4"
In another example, --server-bridge (without parameters) expands as follows:
mode server tls-server push "route-gateway dhcp"
Or --server-bridge nogw expands as follows:
mode server tls-server
This is a partial list of options which can currently be pushed: --route, --route-gateway, --route-delay, --redirect-gateway, --ip-win32, --dhcp-option, --inactive, --ping, --ping-exit, --ping-restart, --setenv, --persist-key, --persist-tun, --echo, --comp-lzo, --socket-flags, --sndbuf, --rcvbuf
This option must be associated with a specific client instance, which means that it must be specified either in a client instance config file using --client-config-dir or dynamically generated using a --client-connect script.
The goal of this option is to provide a long-term association between clients (denoted by their common name) and the virtual IP address assigned to them from the ifconfig-pool. Maintaining a long-term association is good for clients because it allows them to effectively use the --persist-tun option.
file is a comma-delimited ASCII file, formatted as <Common-Name>,<IP-address>.
If seconds = 0, file will be treated as read-only. This is useful if you would like to treat file as a configuration file.
Note that the entries in this file are treated by OpenVPN as suggestions only, based on past associations between a common name and IP address. They do not guarantee that the given common name will always receive the given IP address. If you want guaranteed assignment, use --ifconfig-push
This option is deprecated, and should be replaced with --topology p2p which is functionally equivalent.
The parameters local and remote-netmask are set according to the --ifconfig directive which you want to execute on the client machine to configure the remote end of the tunnel. Note that the parameters local and remote-netmask are from the perspective of the client, not the server. They may be DNS names rather than IP addresses, in which case they will be resolved on the server at the time of client connection.
This option must be associated with a specific client instance, which means that it must be specified either in a client instance config file using --client-config-dir or dynamically generated using a --client-connect script.
Remember also to include a --route directive in the main OpenVPN config file which encloses local, so that the kernel will know to route it to the server's TUN/TAP interface.
OpenVPN's internal client IP address selection algorithm works as follows:
1
-- Use
--client-connect script
generated file for static IP (first choice).
2
-- Use
--client-config-dir
file for static IP (next choice).
3
-- Use
--ifconfig-pool
allocation for dynamic IP (last choice).
This directive can be used to route a fixed subnet from the server to a particular client, regardless of where the client is connecting from. Remember that you must also add the route to the system routing table as well (such as by using the --route directive). The reason why two routes are needed is that the --route directive routes the packet from the kernel to OpenVPN. Once in OpenVPN, the --iroute directive routes to the specific client.
This option must be specified either in a client instance config file using --client-config-dir or dynamically generated using a --client-connect script.
The --iroute directive also has an important interaction with --push "route ...". --iroute essentially defines a subnet which is owned by a particular client (we will call this client A). If you would like other clients to be able to reach A's subnet, you can use --push "route ..." together with --client-to-client to effect this. In order for all clients to see A's subnet, OpenVPN must push this route to all clients EXCEPT for A, since the subnet is already owned by A. OpenVPN accomplishes this by not not pushing a route to a client if it matches one of the client's iroutes.
When this option is used, each client will "see" the other clients which are currently connected. Otherwise, each client will only see the server. Don't use this option if you want to firewall tunnel traffic using custom, per-client rules.
If the script wants to generate a dynamic config file to be applied on the server when the client connects, it should write it to the file named by $1.
See the --client-config-dir option below for options which can be legally used in a dynamically generated config file.
Note that the return value of script is significant. If script returns a non-zero error status, it will cause the client to be disconnected.
The exception to this rule is if the --client-disconnect script or plugins are cascaded, and at least one client-connect function succeeded, then ALL of the client-disconnect functions for scripts and plugins will be called on client instance object deletion, even in cases where some of the related client-connect functions returned an error status.
This file can specify a fixed IP address for a given client using --ifconfig-push, as well as fixed subnets owned by the client using --iroute.
One of the useful properties of this option is that it allows client configuration files to be conveniently created, edited, or removed while the server is live, without needing to restart the server.
The following options are legal in a client-specific context: --push, --push-reset, --iroute, --ifconfig-push, and --config.
When OpenVPN is tunneling data from a TUN/TAP device to a remote client over a TCP connection, it is possible that the TUN/TAP device might produce data at a faster rate than the TCP connection can support. When the number of output packets queued before sending to the TCP socket reaches this limit for a given client connection, OpenVPN will start to drop outgoing packets directed at this client.
The macro expands as follows:
if mode server: socket-flags TCP_NODELAY push "socket-flags TCP_NODELAY"
Note that this directive affects OpenVPN's internal routing table, not the kernel routing table.
This is an imperfect solution however, because in a real DoS scenario, legitimate connections might also be refused.
For the best protection against DoS attacks in server mode, use --proto udp and --tls-auth.
cmd will be executed with 3 parameters:
[1] operation --
"add", "update", or "delete" based on whether or not
the address is being added to, modified, or deleted from
OpenVPN's internal routing table.
[2] address --
The address being learned or unlearned. This can be
an IPv4 address such as "198.162.10.14", an IPv4 subnet
such as "198.162.10.0/24", or an ethernet MAC address (when
--dev tap
is being used) such as "00:FF:01:02:03:04".
[3] common name --
The common name on the certificate associated with the
client linked to this address. Only present for "add"
or "update" operations, not "delete".
On "add" or "update" methods, if the script returns a failure code (non-zero), OpenVPN will reject the address and will not modify its internal routing table.
Normally, the cmd script will use the information provided above to set appropriate firewall entries on the VPN TUN/TAP interface. Since OpenVPN provides the association between virtual IP or MAC address and the client's authenticated common name, it allows a user-defined script to configure firewall access policies with regard to the client's high-level common name, rather than the low level client virtual addresses.
OpenVPN will execute script as a shell command to validate the username/password provided by the client.
If method is set to "via-env", OpenVPN will call script with the environmental variables username and password set to the username/password strings provided by the client. Be aware that this method is insecure on some platforms which make the environment of a process publicly visible to other unprivileged processes.
If method is set to "via-file", OpenVPN will write the username and password to the first two lines of a temporary file. The filename will be passed as an argument to script, and the file will be automatically deleted by OpenVPN after the script returns. The location of the temporary file is controlled by the --tmp-dir option, and will default to the current directory if unspecified. For security, consider setting --tmp-dir to a volatile storage medium such as /dev/shm (if available) to prevent the username/password file from touching the hard drive.
The script should examine the username and password, returning a success exit code (0) if the client's authentication request is to be accepted, or a failure code (1) to reject the client.
This directive is designed to enable a plugin-style interface for extending OpenVPN's authentication capabilities.
To protect against a client passing a maliciously formed username or password string, the username string must consist only of these characters: alphanumeric, underbar ('_'), dash ('-'), dot ('.'), or at ('@'). The password string can consist of any printable characters except for CR or LF. Any illegal characters in either the username or password string will be converted to underbar ('_').
Care must be taken by any user-defined scripts to avoid creating a security vulnerability in the way that these strings are handled. Never use these strings in such a way that they might be escaped or evaluated by a shell interpreter.
For a sample script that performs PAM authentication, see sample-scripts/auth-pam.pl in the OpenVPN source distribution.
Options that will be compared for compatibility include dev-type, link-mtu, tun-mtu, proto, tun-ipv6, ifconfig, comp-lzo, fragment, keydir, cipher, auth, keysize, secret, no-replay, no-iv, tls-auth, key-method, tls-server, and tls-client.
This option requires that --disable-occ NOT be used.
If you use this directive, the entire responsibility of authentication will rest on your --auth-user-pass-verify script, so keep in mind that bugs in your script could potentially compromise the security of your VPN.
If you don't use this directive, but you also specify an --auth-user-pass-verify script, then OpenVPN will perform double authentication. The client certificate verification AND the --auth-user-pass-verify script will need to succeed in order for a client to be authenticated and accepted onto the VPN.
By default, OpenVPN will remap any character other than alphanumeric, underbar ('_'), dash ('-'), dot ('.'), and slash ('/') to underbar ('_'). The X509 Subject string as returned by the tls_id environmental variable, can additionally contain colon (':') or equal ('=').
While name remapping is performed for security reasons to reduce the possibility of introducing string expansion security vulnerabilities in user-defined authentication scripts, this option is provided for those cases where it is desirable to disable the remapping feature. Don't use this option unless you know what you are doing!
Not implemented on Windows.
pull tls-client
In particular, --pull allows the server to push routes to the client, so you should not use --pull or --client in situations where you don't trust the server to have control over the client's routing table.
If up is omitted, username/password will be prompted from the console.
The server configuration must specify an --auth-user-pass-verify script to verify the username/password provided by the client.
Normally used to prevent auth errors from being fatal on the client side, and to permit username/password requeries in case of error.
An AUTH_FAILED message is generated by the server if the client fails --auth-user-pass authentication, or if the server-side --client-connect script returns an error status when the client tries to connect.
type can be one of:
none --
Client will exit with a fatal error (this is the default).
nointeract --
Client will retry the connection without requerying for an
--auth-user-pass
username/password. Use this option for unattended clients.
interact --
Client will requery for an
--auth-user-pass
username/password and/or private key password before attempting a reconnection.
Note that while this option cannot be pushed, it can be controlled from the management interface.
The optional direction parameter enables the use of 4 distinct keys (HMAC-send, cipher-encrypt, HMAC-receive, cipher-decrypt), so that each data flow direction has a different set of HMAC and cipher keys. This has a number of desirable security properties including eliminating certain kinds of DoS and message replay attacks.
When the direction parameter is omitted, 2 keys are used bidirectionally, one for HMAC and the other for encryption/decryption.
The direction parameter should always be complementary on either side of the connection, i.e. one side should use "0" and the other should use "1", or both sides should omit it altogether.
The direction parameter requires that file contains a 2048 bit key. While pre-1.5 versions of OpenVPN generate 1024 bit key files, any version of OpenVPN which supports the direction parameter, will also support 2048 bit key file generation using the --genkey option.
Static key encryption mode has certain advantages, the primary being ease of configuration.
There are no certificates or certificate authorities or complicated negotiation handshakes and protocols. The only requirement is that you have a pre-existing secure channel with your peer (such as ssh ) to initially copy the key. This requirement, along with the fact that your key never changes unless you manually generate a new one, makes it somewhat less secure than TLS mode (see below). If an attacker manages to steal your key, everything that was ever encrypted with it is compromised. Contrast that to the perfect forward secrecy features of TLS mode (using Diffie Hellman key exchange), where even if an attacker was able to steal your private key, he would gain no information to help him decrypt past sessions.
Another advantageous aspect of Static Key encryption mode is that it is a handshake-free protocol without any distinguishing signature or feature (such as a header or protocol handshake sequence) that would mark the ciphertext packets as being generated by OpenVPN. Anyone eavesdropping on the wire would see nothing but random-looking data.
OpenVPN's usage of HMAC is to first encrypt a packet, then HMAC the resulting ciphertext.
In static-key encryption mode, the HMAC key is included in the key file generated by --genkey. In TLS mode, the HMAC key is dynamically generated and shared between peers via the TLS control channel. If OpenVPN receives a packet with a bad HMAC it will drop the packet. HMAC usually adds 16 or 20 bytes per packet. Set alg=none to disable authentication.
For more information on HMAC see http://www.cs.ucsd.edu/users/mihir/papers/hmac.html
For more information on blowfish, see http://www.counterpane.com/blowfish.html
To see other ciphers that are available with OpenVPN, use the --show-ciphers option.
OpenVPN supports the CBC, CFB, and OFB cipher modes, however CBC is recommended and CFB and OFB should be considered advanced modes.
Set alg=none to disable encryption.
Set alg=none to disable the PRNG and use the OpenSSL RAND_bytes function instead for all of OpenVPN's pseudo-random number needs.
If engine-name is specified, use a specific crypto engine. Use the --show-engines standalone option to list the crypto engines which are supported by OpenSSL.
OpenVPN provides datagram replay protection by default.
Replay protection is accomplished by tagging each outgoing datagram with an identifier that is guaranteed to be unique for the key being used. The peer that receives the datagram will check for the uniqueness of the identifier. If the identifier was already received in a previous datagram, OpenVPN will drop the packet. Replay protection is important to defeat attacks such as a SYN flood attack, where the attacker listens in the wire, intercepts a TCP SYN packet (identifying it by the context in which it occurs in relation to other packets), then floods the receiving peer with copies of this packet.
OpenVPN's replay protection is implemented in slightly different ways, depending on the key management mode you have selected.
In Static Key mode or when using an CFB or OFB mode cipher, OpenVPN uses a 64 bit unique identifier that combines a time stamp with an incrementing sequence number.
When using TLS mode for key exchange and a CBC cipher mode, OpenVPN uses only a 32 bit sequence number without a time stamp, since OpenVPN can guarantee the uniqueness of this value for each key. As in IPSec, if the sequence number is close to wrapping back to zero, OpenVPN will trigger a new key exchange.
To check for replays, OpenVPN uses the sliding window algorithm used by IPSec.
By default n is 64 (the IPSec default) and t is 15 seconds.
This option is only relevant in UDP mode, i.e. when either --proto udp is specifed, or no --proto option is specified.
When OpenVPN tunnels IP packets over UDP, there is the possibility that packets might be dropped or delivered out of order. Because OpenVPN, like IPSec, is emulating the physical network layer, it will accept an out-of-order packet sequence, and will deliver such packets in the same order they were received to the TCP/IP protocol stack, provided they satisfy several constraints.
(a) The packet cannot be a replay (unless --no-replay is specified, which disables replay protection altogether).
(b) If a packet arrives out of order, it will only be accepted if the difference between its sequence number and the highest sequence number received so far is less than n.
(c) If a packet arrives out of order, it will only be accepted if it arrives no later than t seconds after any packet containing a higher sequence number.
If you are using a network link with a large pipeline (meaning that the product of bandwidth and latency is high), you may want to use a larger value for n. Satellite links in particular often require this.
If you run OpenVPN at --verb 4, you will see the message "Replay-window backtrack occurred [x]" every time the maximum sequence number backtrack seen thus far increases. This can be used to calibrate n.
There is some controversy on the appropriate method of handling packet reordering at the security layer.
Namely, to what extent should the security layer protect the encapsulated protocol from attacks which masquerade as the kinds of normal packet loss and reordering that occur over IP networks?
The IPSec and OpenVPN approach is to allow packet reordering within a certain fixed sequence number window.
OpenVPN adds to the IPSec model by limiting the window size in time as well as sequence space.
OpenVPN also adds TCP transport as an option (not offered by IPSec) in which case OpenVPN can adopt a very strict attitude towards message deletion and reordering: Don't allow it. Since TCP guarantees reliability, any packet loss or reordering event can be assumed to be an attack.
In this sense, it could be argued that TCP tunnel transport is preferred when tunneling non-IP or UDP application protocols which might be vulnerable to a message deletion or reordering attack which falls within the normal operational parameters of IP networks.
So I would make the statement that one should never tunnel a non-IP protocol or UDP application protocol over UDP, if the protocol might be vulnerable to a message deletion or reordering attack that falls within the normal operating parameters of what is to be expected from the physical IP layer. The problem is easily fixed by simply using TCP as the VPN transport layer.
This option will strengthen protection against replay attacks, especially when you are using OpenVPN in a dynamic context (such as with --inetd) when OpenVPN sessions are frequently started and stopped.
This option will keep a disk copy of the current replay protection state (i.e. the most recent packet timestamp and sequence number received from the remote peer), so that if an OpenVPN session is stopped and restarted, it will reject any replays of packets which were already received by the prior session.
This option only makes sense when replay protection is enabled (the default) and you are using either --secret (shared-secret key mode) or TLS mode with --tls-auth.
OpenVPN uses an IV by default, and requires it for CFB and OFB cipher modes (which are totally insecure without it). Using an IV is important for security when multiple messages are being encrypted/decrypted with the same key.
IV is implemented differently depending on the cipher mode used.
In CBC mode, OpenVPN uses a pseudo-random IV for each packet.
In CFB/OFB mode, OpenVPN uses a unique sequence number and time stamp as the IV. In fact, in CFB/OFB mode, OpenVPN uses a datagram space-saving optimization that uses the unique identifier for datagram replay protection as the IV.
The typical usage of --test-crypto would be something like this:
openvpn --test-crypto --secret key
or
openvpn --test-crypto --secret key --verb 9
This option is very useful to test OpenVPN after it has been ported to a new platform, or to isolate problems in the compiler, OpenSSL crypto library, or OpenVPN's crypto code. Since it is a self-test mode, problems with encryption and authentication can be debugged independently of network and tunnel issues.
To use TLS mode, each peer that runs OpenVPN should have its own local certificate/key pair ( --cert and --key ), signed by the root certificate which is specified in --ca.
When two OpenVPN peers connect, each presents its local certificate to the other. Each peer will then check that its partner peer presented a certificate which was signed by the master root certificate as specified in --ca.
If that check on both peers succeeds, then the TLS negotiation will succeed, both OpenVPN peers will exchange temporary session keys, and the tunnel will begin passing data.
The OpenVPN distribution contains a set of scripts for managing RSA certificates & keys, located in the easy-rsa subdirectory.
The easy-rsa package is also rendered in web form here: http://openvpn.net/easyrsa.html
openssl req -nodes -new -x509 -keyout ca.key -out ca.crt
Then edit your openssl.cnf file and edit the certificate variable to point to your new root certificate ca.crt.
For testing purposes only, the OpenVPN distribution includes a sample CA certificate (ca.crt). Of course you should never use the test certificates and test keys distributed with OpenVPN in a production environment, since by virtue of the fact that they are distributed with OpenVPN, they are totally insecure.
openssl dhparam -out dh1024.pem 1024
to generate your own, or use the existing dh1024.pem file included with the OpenVPN distribution. Diffie Hellman parameters may be considered public.
openssl req -nodes -new -keyout mycert.key -out mycert.csr
If your certificate authority private key lives on another machine, copy the certificate signing request (mycert.csr) to this other machine (this can be done over an insecure channel such as email). Now sign the certificate with a command such as:
openssl ca -out mycert.crt -in mycert.csr
Now copy the certificate (mycert.crt) back to the peer which initially generated the .csr file (this can be over a public medium). Note that the openssl ca command reads the location of the certificate authority key from its configuration file such as /usr/share/ssl/openssl.cnf -- note also that for certificate authority functions, you must set up the files index.txt (may be empty) and serial (initialize to 01 ).
0
(default) -- Try to determind automatically.
1
-- Use sign.
2
-- Use sign recover.
4
-- Use decrypt.
8
-- Use unwrap.
Use this option instead of --cert and --key.
This makes it possible to use any smart card, supported by Windows, but also any kind of certificate, residing in the Cert Store, where you have access to the private key. This option has been tested with a couple of different smart cards (GemSAFE, Cryptoflex, and Swedish Post Office eID) on the client side, and also an imported PKCS12 software certificate on the server side.
To select a certificate, based on a substring search in the certificate's subject:
cryptoapicert "SUBJ:Peter Runestig"
To select a certificate, based on certificate's thumbprint:
cryptoapicert "THUMB:f6 49 24 41 01 b4 ..."
The thumbprint hex string can easily be copy-and-pasted from the Windows Certificate Store GUI.
After OpenVPN negotiates a TLS session, a new set of keys for protecting the tunnel data channel is generated and exchanged over the TLS session.
In method 1 (the default for OpenVPN 1.x), both sides generate random encrypt and HMAC-send keys which are forwarded to the other host over the TLS channel.
In method 2, (the default for OpenVPN 2.0) the client generates a random key. Both client and server also generate some random seed material. All key source material is exchanged over the TLS channel. The actual keys are generated using the TLS PRF function, taking source entropy from both client and server. Method 2 is designed to closely parallel the key generation process used by TLS 1.0.
Note that in TLS mode, two separate levels of keying occur:
(1) The TLS connection is initially negotiated, with both sides of the connection producing certificates and verifying the certificate (or other authentication info provided) of the other side. The --key-method parameter has no effect on this process.
(2) After the TLS connection is established, the tunnel session keys are separately negotiated over the existing secure TLS channel. Here, --key-method determines the derivation of the tunnel session keys.
When using dual-factor authentication, note that this default value may cause the end user to be challenged to reauthorize once per hour.
Also, keep in mind that this option can be used on both the client and server, and whichever uses the lower value will be the one to trigger the renegotiation. A common mistake is to set --reneg-sec to a higher value on either the client or server, while the other side of the connection is still using the default value of 3600 seconds, meaning that the renegotiation will still occur once per 3600 seconds. The solution is to increase --reneg-sec on both the client and server, or set it to 0 on one side of the connection (to disable), and to your chosen value on the other side.
If the daemon is reset by a signal or --ping-restart, it will allow one new connection.
--single-session can be used with --ping-exit or --inactive to create a single dynamic session that will exit when finished.
In a nutshell, --tls-auth enables a kind of "HMAC firewall" on OpenVPN's TCP/UDP port, where TLS control channel packets bearing an incorrect HMAC signature can be dropped immediately without response.
file (required) is a key file which can be in one of two formats:
(1) An OpenVPN static key file generated by --genkey (required if direction parameter is used).
(2) A freeform passphrase file. In this case the HMAC key will be derived by taking a secure hash of this file, similar to the md5sum(1) or sha1sum(1) commands.
OpenVPN will first try format (1), and if the file fails to parse as a static key file, format (2) will be used.
See the --secret option for more information on the optional direction parameter.
--tls-auth is recommended when you are running OpenVPN in a mode where it is listening for packets from any IP address, such as when --remote is not specified, or --remote is specified with --float.
The rationale for this feature is as follows. TLS requires a multi-packet exchange before it is able to authenticate a peer. During this time before authentication, OpenVPN is allocating resources (memory and CPU) to this potential peer. The potential peer is also exposing many parts of OpenVPN and the OpenSSL library to the packets it is sending. Most successful network attacks today seek to either exploit bugs in programs (such as buffer overflow attacks) or force a program to consume so many resources that it becomes unusable. Of course the first line of defense is always to produce clean, well-audited code. OpenVPN has been written with buffer overflow attack prevention as a top priority. But as history has shown, many of the most widely used network applications have, from time to time, fallen to buffer overflow attacks.
So as a second line of defense, OpenVPN offers this special layer of authentication on top of the TLS control channel so that every packet on the control channel is authenticated by an HMAC signature and a unique ID for replay protection. This signature will also help protect against DoS (Denial of Service) attacks. An important rule of thumb in reducing vulnerability to DoS attacks is to minimize the amount of resources a potential, but as yet unauthenticated, client is able to consume.
--tls-auth does this by signing every TLS control channel packet with an HMAC signature, including packets which are sent before the TLS level has had a chance to authenticate the peer. The result is that packets without the correct signature can be dropped immediately upon reception, before they have a chance to consume additional system resources such as by initiating a TLS handshake. --tls-auth can be strengthened by adding the --replay-persist option which will keep OpenVPN's replay protection state in a file so that it is not lost across restarts.
It should be emphasized that this feature is optional and that the passphrase/key file used with --tls-auth gives a peer nothing more than the power to initiate a TLS handshake. It is not used to encrypt or authenticate any tunnel data.
For the extremely security conscious, it is possible to protect your private key with a password. Of course this means that every time the OpenVPN daemon is started you must be there to type the password. The --askpass option allows you to start OpenVPN from the command line. It will query you for a password before it daemonizes. To protect a private key with a password you should omit the -nodes option when you use the openssl command line tool to manage certificates and private keys.
If file is specified, read the password from the first line of file. Keep in mind that storing your password in a file to a certain extent invalidates the extra security provided by using an encrypted key (Note: OpenVPN will only read passwords from a file if it has been built with the --enable-password-save configure option, or on Windows by defining ENABLE_PASSWORD_SAVE in config-win32.h).
If specified, this directive will cause OpenVPN to immediately forget username/password inputs after they are used. As a result, when OpenVPN needs a username/password, it will prompt for input from stdin, which may be multiple times during the duration of an OpenVPN session.
This directive does not affect the --http-proxy username/password. It is always cached.
cmd should return 0 to allow the TLS handshake to proceed, or 1 to fail. cmd is executed as
cmd certificate_depth X509_NAME_oneline
This feature is useful if the peer you want to trust has a certificate which was signed by a certificate authority who also signed many other certificates, where you don't necessarily want to trust all of them, but rather be selective about which peer certificate you will accept. This feature allows you to write a script which will test the X509 name on a certificate and decide whether or not it should be accepted. For a simple perl script which will test the common name field on the certificate, see the file verify-cn in the OpenVPN distribution.
See the "Environmental Variables" section below for additional parameters passed as environmental variables.
Note that cmd can be a shell command with multiple arguments, in which case all OpenVPN-generated arguments will be appended to cmd to build a command line which will be passed to the script.
Name can also be a common name prefix, for example if you want a client to only accept connections to "Server-1", "Server-2", etc., you can simply use --tls-remote Server
Using a common name prefix is a useful alternative to managing a CRL (Certificate Revocation List) on the client, since it allows the client to refuse all certificates except for those associated with designated servers.
--tls-remote is a useful replacement for the --tls-verify option to verify the remote host, because --tls-remote works in a --chroot environment too.
This is a useful security option for clients, to ensure that the host they connect with is a designated server.
See the easy-rsa/build-key-server script for an example of how to generate a certificate with the nsCertType field set to "server".
If the server certificate's nsCertType field is set to "server", then the clients can verify this with --ns-cert-type server.
This is an important security precaution to protect against a man-in-the-middle attack where an authorized client attempts to connect to another client by impersonating the server. The attack is easily prevented by having clients verify the server certificate using any one of --ns-cert-type, --tls-remote, or --tls-verify.
This is a useful security option for clients, to ensure that the host they connect to is a designated server.
The key usage should be encoded in hex, more than one key usage can be specified.
This is a useful security option for clients, to ensure that the host they connect to is a designated server.
The extended key usage should be encoded in oid notation, or OpenSSL symbolic representation.
This is a useful security option for clients, to ensure that the host they connect to is a designated server.
The --remote-cert-tls client option is equivalent to --remote-cert-ku 80 08 88 --remote-cert-eku "TLS Web Client Authentication"
The key usage is digitalSignature and/or keyAgreement.
The --remote-cert-tls server option is equivalent to --remote-cert-ku a0 88 --remote-cert-eku "TLS Web Server Authentication"
The key usage is digitalSignature and ( keyEncipherment or keyAgreement ).
This is an important security precaution to protect against a man-in-the-middle attack where an authorized client attempts to connect to another client by impersonating the server. The attack is easily prevented by having clients verify the server certificate using any one of --remote-cert-tls, --tls-remote, or --tls-verify.
A CRL (certificate revocation list) is used when a particular key is compromised but when the overall PKI is still intact.
Suppose you had a PKI consisting of a CA, root certificate, and a number of client certificates. Suppose a laptop computer containing a client key and certificate was stolen. By adding the stolen certificate to the CRL file, you could reject any connection which attempts to use it, while preserving the overall integrity of the PKI.
The only time when it would be necessary to rebuild the entire PKI from scratch would be if the root certificate key itself was compromised.
One of the advantages of persistent tunnels is that they eliminate the need for separate --up and --down scripts to run the appropriate ifconfig(8) and route(8) commands. These commands can be placed in the the same shell script which starts or terminates an OpenVPN session.
Another advantage is that open connections through the TUN/TAP-based tunnel will not be reset if the OpenVPN peer restarts. This can be useful to provide uninterrupted connectivity through the tunnel in the event of a DHCP reset of the peer's public IP address (see the --ipchange option above).
One disadvantage of persistent tunnels is that it is harder to automatically configure their MTU value (see --link-mtu and --tun-mtu above).
On some platforms such as Windows, TAP-Win32 tunnels are persistent by default.
The special string 'env' indicates that the pathname should be read from the SystemRoot environmental variable.
manual -- Don't set the IP address or netmask automatically. Instead output a message to the console telling the user to configure the adapter manually and indicating the IP/netmask which OpenVPN expects the adapter to be set to.
dynamic [offset] [lease-time] -- Automatically set the IP address and netmask by replying to DHCP query messages generated by the kernel. This mode is probably the "cleanest" solution for setting the TCP/IP properties since it uses the well-known DHCP protocol. There are, however, two prerequisites for using this mode: (1) The TCP/IP properties for the TAP-Win32 adapter must be set to "Obtain an IP address automatically," and (2) OpenVPN needs to claim an IP address in the subnet for use as the virtual DHCP server address. By default in --dev tap mode, OpenVPN will take the normally unused first address in the subnet. For example, if your subnet is 192.168.4.0 netmask 255.255.255.0, then OpenVPN will take the IP address 192.168.4.0 to use as the virtual DHCP server address. In --dev tun mode, OpenVPN will cause the DHCP server to masquerade as if it were coming from the remote endpoint. The optional offset parameter is an integer which is > -256 and < 256 and which defaults to 0. If offset is positive, the DHCP server will masquerade as the IP address at network address + offset. If offset is negative, the DHCP server will masquerade as the IP address at broadcast address + offset. The Windows ipconfig /all command can be used to show what Windows thinks the DHCP server address is. OpenVPN will "claim" this address, so make sure to use a free address. Having said that, different OpenVPN instantiations, including different ends of the same connection, can share the same virtual DHCP server address. The lease-time parameter controls the lease time of the DHCP assignment given to the TAP-Win32 adapter, and is denoted in seconds. Normally a very long lease time is preferred because it prevents routes involving the TAP-Win32 adapter from being lost when the system goes to sleep. The default lease time is one year.
netsh -- Automatically set the IP address and netmask using the Windows command-line "netsh" command. This method appears to work correctly on Windows XP but not Windows 2000.
ipapi -- Automatically set the IP address and netmask using the Windows IP Helper API. This approach does not have ideal semantics, though testing has indicated that it works okay in practice. If you use this option, it is best to leave the TCP/IP properties for the TAP-Win32 adapter in their default state, i.e. "Obtain an IP address automatically."
adaptive -- (Default) Try dynamic method initially and fail over to netsh if the DHCP negotiation with the TAP-Win32 adapter does not succeed in 20 seconds. Such failures have been known to occur when certain third-party firewall packages installed on the client machine block the DHCP negotiation used by the TAP-Win32 adapter. Note that if the netsh failover occurs, the TAP-Win32 adapter TCP/IP properties will be reset from DHCP to static, and this will cause future OpenVPN startups using the adaptive mode to use netsh immediately, rather than trying dynamic first. To "unstick" the adaptive mode from using netsh, run OpenVPN at least once using the dynamic mode to restore the TAP-Win32 adapter TCP/IP properties to a DHCP configuration.
adaptive
(default) -- Try IP helper API first. If that fails, fall
back to the route.exe shell command.
ipapi
-- Use IP helper API.
exe
-- Call the route.exe shell command.
DOMAIN name -- Set Connection-specific DNS Suffix.
DNS addr -- Set primary domain name server address. Repeat this option to set secondary DNS server addresses.
WINS addr -- Set primary WINS server address (NetBIOS over TCP/IP Name Server). Repeat this option to set secondary WINS server addresses.
NBDD addr -- Set primary NBDD server address (NetBIOS over TCP/IP Datagram Distribution Server) Repeat this option to set secondary NBDD server addresses.
NTP addr -- Set primary NTP server address (Network Time Protocol). Repeat this option to set secondary NTP server addresses.
NBT type -- Set NetBIOS over TCP/IP Node type. Possible options: 1 = b-node (broadcasts), 2 = p-node (point-to-point name queries to a WINS server), 4 = m-node (broadcast then query name server), and 8 = h-node (query name server, then broadcast).
NBS scope-id -- Set NetBIOS over TCP/IP Scope. A NetBIOS Scope ID provides an extended naming service for the NetBIOS over TCP/IP (Known as NBT) module. The primary purpose of a NetBIOS scope ID is to isolate NetBIOS traffic on a single network to only those nodes with the same NetBIOS scope ID. The NetBIOS scope ID is a character string that is appended to the NetBIOS name. The NetBIOS scope ID on two hosts must match, or the two hosts will not be able to communicate. The NetBIOS Scope ID also allows computers to use the same computer name, as they have different scope IDs. The Scope ID becomes a part of the NetBIOS name, making the name unique. (This description of NetBIOS scopes courtesy of NeonSurge@abyss.com)
DISABLE-NBT -- Disable Netbios-over-TCP/IP.
Note that if --dhcp-option is pushed via --push to a non-windows client, the option will be saved in the client's environment before the up script is called, under the name "foreign_option_{n}".
This option is intended to be used to troubleshoot problems with the --ifconfig and --ip-win32 options, and is used to give the TAP-Win32 adapter time to come up before Windows IP Helper API operations are applied to it.
exit-event is the name of a Windows global event object, and OpenVPN will continuously monitor the state of this event object and exit when it becomes signaled.
The second parameter indicates the initial state of exit-event and normally defaults to 0.
Multiple OpenVPN processes can be simultaneously executed with the same exit-event parameter. In any case, the controlling process can signal exit-event, causing all such OpenVPN processes to exit.
When executing an OpenVPN process using the --service directive, OpenVPN will probably not have a console window to output status/error messages, therefore it is useful to use --log or --log-append to write these messages to a file.
Namely, the point-to-point endpoints used in TUN device emulation must be the middle two addresses of a /30 subnet (netmask 255.255.255.252).
--verb option can be used BEFORE this option to produce debugging information.
Q: Why is string remapping necessary?
A: It's an important security feature to prevent the malicious coding of strings from untrusted sources to be passed as parameters to scripts, saved in the environment, used as a common name, translated to a filename, etc.
Q: Can string remapping be disabled?
A: Yes, by using the --no-name-remapping option, however this should be considered an advanced option.
Here is a brief rundown of OpenVPN's current string types and the permitted character class for each string:
X509 Names: Alphanumeric, underbar ('_'), dash ('-'), dot ('.'), at ('@'), colon (':'), slash ('/'), and equal ('='). Alphanumeric is defined as a character which will cause the C library isalnum() function to return true.
Common Names: Alphanumeric, underbar ('_'), dash ('-'), dot ('.'), and at ('@').
--auth-user-pass username: Same as Common Name, with one exception: starting with OpenVPN 2.0.1, the username is passed to the OPENVPN_PLUGIN_AUTH_USER_PASS_VERIFY plugin in its raw form, without string remapping.
--auth-user-pass password: Any "printable" character except CR or LF. Printable is defined to be a character which will cause the C library isprint() function to return true.
--client-config-dir filename as derived from common name or username: Alphanumeric, underbar ('_'), dash ('-'), and dot ('.') except for "." or ".." as standalone strings. As of 2.0.1-rc6, the at ('@') character has been added as well for compatibility with the common name character class.
Environmental variable names: Alphanumeric or underbar ('_').
Environmental variable values: Any printable character.
For all cases, characters in a string which are not members of the legal character class for that string type will be remapped to underbar ('_').
As of OpenVPN 2.0-beta12, in server mode, environmental variables set by OpenVPN are scoped according to the client objects they are associated with, so there should not be any issues with scripts having access to stale, previously set variables which refer to different client instances.
parm will be one of "network", "netmask", "gateway", or "metric".
n is the OpenVPN route number, starting from 1.
If the network or gateway are resolvable DNS names, their IP address translations will be recorded rather than their names as denoted on the command line or configuration file.
X509_0_emailAddress=me@myhost.mydomain X509_0_CN=Test-Client X509_0_O=OpenVPN-TEST X509_0_ST=NA X509_0_C=KG X509_1_emailAddress=me@myhost.mydomain X509_1_O=OpenVPN-TEST X509_1_L=BISHKEK X509_1_ST=NA X509_1_C=KG
This signal may also be internally generated by a timeout condition, governed by the --ping-restart option.
This signal, when combined with --persist-remote-ip, may be sent when the underlying parameters of the host's network interface change such as when the host is a DHCP client and is assigned a new IP address. See --ipchange above for more information.
Make device: mknod /dev/net/tun c 10 200
Load driver: modprobe tun
If you have Linux 2.2 or earlier, you should obtain version 1.1 of the TUN/TAP driver from http://vtun.sourceforge.net/tun/ and follow the installation instructions.
If you installed from RPM, the mknod step may be omitted, because the RPM install does that for you.
If you have Linux 2.2, you should obtain version 1.1 of the TUN/TAP driver from http://vtun.sourceforge.net/tun/ and follow the installation instructions.
For other platforms, consult the INSTALL file at http://openvpn.net/install.html for more information.
If you are using a Linux iptables-based firewall, you may need to enter the following command to allow incoming packets on the TUN device:
See the firewalls section below for more information on configuring firewalls for use with OpenVPN.
Now we will choose the tunnel endpoints. Tunnel endpoints are private IP addresses that only have meaning in the context of the VPN. Each machine will use the tunnel endpoint of the other machine to access it over the VPN. In our example, the tunnel endpoint for may.kg will be 10.4.0.1 and for june.kg, 10.4.0.2.
Once the VPN is established, you have essentially created a secure alternate path between the two hosts which is addressed by using the tunnel endpoints. You can control which network traffic passes between the hosts (a) over the VPN or (b) independently of the VPN, by choosing whether to use (a) the VPN endpoint address or (b) the public internet address, to access the remote host. For example if you are on may.kg and you wish to connect to june.kg via ssh without using the VPN (since ssh has its own built-in security) you would use the command ssh june.kg. However in the same scenario, you could also use the command telnet 10.4.0.2 to create a telnet session with june.kg over the VPN, that would use the VPN to secure the session rather than ssh.
You can use any address you wish for the tunnel endpoints but make sure that they are private addresses (such as those that begin with 10 or 192.168) and that they are not part of any existing subnet on the networks of either peer, unless you are bridging. If you use an address that is part of your local subnet for either of the tunnel endpoints, you will get a weird feedback loop.
On may:
On june:
Now verify the tunnel is working by pinging across the tunnel.
On may:
On june:
The --verb 9 option will produce verbose output, similar to the tcpdump(8) program. Omit the --verb 9 option to have OpenVPN run quietly.
This command will build a random key file called key (in ascii format). Now copy key to june over a secure medium such as by using the scp(1) program.
On may:
On june:
Now verify the tunnel is working by pinging across the tunnel.
On may:
On june:
First, build a separate certificate/key pair for both may and june (see above where --cert is discussed for more info). Then construct Diffie Hellman parameters (see above where --dh is discussed for more info). You can also use the included test files client.crt, client.key, server.crt, server.key and ca.crt. The .crt files are certificates/public-keys, the .key files are private keys, and ca.crt is a certification authority who has signed both client.crt and server.crt. For Diffie Hellman parameters you can use the included file dh1024.pem. Note that all client, server, and certificate authority certificates and keys included in the OpenVPN distribution are totally insecure and should be used for testing only.
On may:
On june:
Now verify the tunnel is working by pinging across the tunnel.
On may:
On june:
Notice the --reneg-sec 60 option we used above. That tells OpenVPN to renegotiate the data channel keys every minute. Since we used --verb 5 above, you will see status information on each new key negotiation.
For production operations, a key renegotiation interval of 60 seconds is probably too frequent. Omit the --reneg-sec 60 option to use OpenVPN's default key renegotiation interval of one hour.
First, ensure that IP forwarding is enabled on both peers. On Linux, enable routing:
and enable TUN packet forwarding through the firewall:
On may:
On june:
Now any machine on the 10.0.0.0/24 subnet can access any machine on the 10.0.1.0/24 subnet over the secure tunnel (or vice versa).
In a production environment, you could put the route command(s) in a shell script and execute with the --up option.
This will allow incoming packets on UDP port 1194 (OpenVPN's default UDP port) from an OpenVPN peer at 1.2.3.4.
If you are using HMAC-based packet authentication (the default in any of OpenVPN's secure modes), having the firewall filter on source address can be considered optional, since HMAC packet authentication is a much more secure method of verifying the authenticity of a packet source. In that case:
would be adequate and would not render the host inflexible with respect to its peer having a dynamic IP address.
OpenVPN also works well on stateful firewalls. In some cases, you may not need to add any static rules to the firewall list if you are using a stateful firewall that knows how to track UDP connections. If you specify --ping n, OpenVPN will be guaranteed to send a packet to its peer at least once every n seconds. If n is less than the stateful firewall connection timeout, you can maintain an OpenVPN connection indefinitely without explicit firewall rules.
You should also add firewall rules to allow incoming IP traffic on TUN or TAP devices such as:
to allow input packets from tun devices,
to allow input packets from tun devices to be forwarded to other hosts on the local network,
to allow input packets from tap devices, and
to allow input packets from tap devices to be forwarded to other hosts on the local network.
These rules are secure if you use packet authentication, since no incoming packets will arrive on a TUN or TAP virtual device unless they first pass an HMAC authentication test.
Go here to download the latest version of OpenVPN, subscribe to the mailing lists, read the mailing list archives, or browse the SVN repository.
This product includes software developed by the OpenSSL Project ( http://www.openssl.org/ )
For more information on the TLS protocol, see http://www.ietf.org/rfc/rfc2246.txt
For more information on the LZO real-time compression library see http://www.oberhumer.com/opensource/lzo/