-.Dd 2008-12-22
+.Dd 2010-01-16
.Dt TINC.CONF 5
.\" Manual page created by:
.\" Ivo Timmermans
sign, but doing so improves readability.
If you leave it out, remember to replace it with at least one space character.
+.Pp
+The server configuration is complemented with host specific configuration (see the next section).
+Although all configuration options for the local host listed in this document can also be put in
+.Pa @sysconfdir@/tinc/ Ns Ar NETNAME Ns Pa /tinc.conf ,
+it is recommended to put host specific configuration options in the host configuration file,
+as this makes it easy to exchange with other nodes.
+
.Pp
Here are all valid variables, listed in alphabetical order.
The default value is given between parentheses.
is selected, then depending on the operating system both IPv4 and IPv6 or just
IPv6 listening sockets will be created.
-.It Va BindToAddress Li = Ar address Bq experimental
+.It Va BindToAddress Li = Ar address Oo Ar port Oc Bq experimental
If your computer has more than one IPv4 or IPv6 address,
.Nm tinc
will by default listen on all of them for incoming connections.
-It is possible to bind only to a single address with this variable.
-
+Multiple
+.Va BindToAddress
+variables may be specified,
+in which case listening sockets for each specified address are made.
+.Pp
+If no
+.Ar port
+is specified, the socket will be bound to the port specified by the
+.Va Port
+option, or to port 655 if neither is given.
+To only bind to a specific port but not to a specific address, use
+.Li *
+for the
+.Ar address .
.Pp
This option may not work on all platforms.
.Pp
This option may not work on all platforms.
+.It Va Broadcast Li = yes | no Po yes Pc Bq experimental
+When disabled, tinc will drop all broadcast and multicast packets, in both router and switch mode.
+
.It Va ConnectTo Li = Ar name
Specifies which other tinc daemon to connect to on startup.
Multiple
won't try to connect to other daemons at all,
and will instead just listen for incoming connections.
+.It Va DecrementTTL Li = yes | no Po no Pc Bq experimental
+When enabled,
+.Nm tinc
+will decrement the Time To Live field in IPv4 packets, or the Hop Limit field in IPv6 packets,
+before forwarding a received packet to the virtual network device or to another node,
+and will drop packets that have a TTL value of zero,
+in which case it will send an ICMP Time Exceeded packet back.
+.Pp
+Do not use this option if you use switch mode and want to use IPv6.
+
.It Va Device Li = Ar device Po Pa /dev/tap0 , Pa /dev/net/tun No or other depending on platform Pc
The virtual network device to use.
.Nm tinc
The info pages of the tinc package contain more information
about configuring the virtual network device.
-.It Va DeviceType Li = tun | tunnohead | tunifhead | tap Po only supported on BSD platforms Pc
+.It Va DeviceType Li = Ar type Pq platform dependent
The type of the virtual network device.
-Tinc will normally automatically select the right type, and this option should not be used.
-However, in case tinc does not seem to correctly interpret packets received from the virtual network device,
-using this option might help.
+Tinc will normally automatically select the right type of tun/tap interface, and this option should not be used.
+However, this option can be used to select one of the special interface types, if support for them is compiled in.
+.Bl -tag -width indent
+
+.It dummy
+Use a dummy interface.
+No packets are ever read or written to a virtual network device.
+Useful for testing, or when setting up a node that only forwards packets for other nodes.
+
+.It raw_socket
+Open a raw socket, and bind it to a pre-existing
+.Va Interface
+(eth0 by default).
+All packets are read from this interface.
+Packets received for the local node are written to the raw socket.
+However, at least on Linux, the operating system does not process IP packets destined for the local host.
+
+.It multicast
+Open a multicast UDP socket and bind it to the address and port (separated by spaces) and optionally a TTL value specified using
+.Va Device .
+Packets are read from and written to this multicast socket.
+This can be used to connect to UML, QEMU or KVM instances listening on the same multicast address.
+Do NOT connect multiple
+.Nm tinc
+daemons to the same multicast address, this will very likely cause routing loops.
+Also note that this can cause decrypted VPN packets to be sent out on a real network if misconfigured.
+
+.It uml Pq not compiled in by default
+Create a UNIX socket with the filename specified by
+.Va Device ,
+or
+.Pa @localstatedir@/run/ Ns Ar NETNAME Ns Pa .umlsocket
+if not specified.
+.Nm tinc
+will wait for a User Mode Linux instance to connect to this socket.
+
+.It vde Pq not compiled in by default
+Uses the libvdeplug library to connect to a Virtual Distributed Ethernet switch,
+using the UNIX socket specified by
+.Va Device ,
+or
+.Pa @localstatedir@/run/vde.ctl
+if not specified.
+.El
+
+Also, in case tinc does not seem to correctly interpret packets received from the virtual network device,
+it can be used to change the way packets are interpreted:
+
.Bl -tag -width indent
-.It tun
+.It tun Pq BSD and Linux
Set type to tun.
Depending on the platform, this can either be with or without an address family header (see below).
-.It tunnohead
+.It tunnohead Pq BSD
Set type to tun without an address family header.
Tinc will expect packets read from the virtual network device to start with an IP header.
On some platforms IPv6 packets cannot be read from or written to the device in this mode.
-.It tunifhead
+.It tunifhead Pq BSD
Set type to tun with an address family header.
Tinc will expect packets read from the virtual network device
to start with a four byte header containing the address family,
followed by an IP header.
This mode should support both IPv4 and IPv6 packets.
-.It tap
+.It tap Pq BSD and Linux
Set type to tap.
Tinc will expect packets read from the virtual network device
to start with an Ethernet header.
.El
+.It Va DirectOnly Li = yes | no Po no Pc Bq experimental
+When this option is enabled, packets that cannot be sent directly to the destination node,
+but which would have to be forwarded by an intermediate node, are dropped instead.
+When combined with the IndirectData option,
+packets for nodes for which we do not have a meta connection with are also dropped.
+
+.It Va Forwarding Li = off | internal | kernel Po internal Pc Bq experimental
+This option selects the way indirect packets are forwarded.
+.Bl -tag -width indent
+
+.It off
+Incoming packets that are not meant for the local node,
+but which should be forwarded to another node, are dropped.
+
+.It internal
+Incoming packets that are meant for another node are forwarded by tinc internally.
+
+.Pp
+This is the default mode, and unless you really know you need another forwarding mode, don't change it.
+
+.It kernel
+Incoming packets are always sent to the TUN/TAP device, even if the packets are not for the local node.
+This is less efficient, but allows the kernel to apply its routing and firewall rules on them,
+and can also help debugging.
+.El
+
.It Va GraphDumpFile Li = Ar filename Bq experimental
If this option is present,
.Nm tinc
This does not affect resolving hostnames to IP addresses from the
host configuration files.
+.It Va IffOneQueue Li = yes | no Po no Pc Bq experimental
+(Linux only) Set IFF_ONE_QUEUE flag on TUN/TAP devices.
+
.It Va Interface Li = Ar interface
Defines the name of the interface corresponding to the virtual network device.
Depending on the operating system and the type of device this may or may not actually set the name of the interface.
It is common practice to change keys at regular intervals to make it even harder for crackers,
even though it is thought to be nearly impossible to crack a single key.
+.It Va LocalDiscovery Li = yes | no Po no Pc Bq experimental
+When enabled,
+.Nm tinc
+will try to detect peers that are on the same local network.
+This will allow direct communication using LAN addresses, even if both peers are behind a NAT
+and they only ConnectTo a third node outside the NAT,
+which normally would prevent the peers from learning each other's LAN address.
+
+.Pp
+Currently, local discovery is implemented by sending broadcast packets to the LAN during path MTU discovery.
+This feature may not work in all possible situations.
+
.It Va MACExpire Li = Ar seconds Pq 600
This option controls the amount of time MAC addresses are kept before they are removed.
This only has effect when
.Va PrivateKeyFile
specified in the configuration file.
+.It Va ProcessPriority Li = low | normal | high
+When this option is used the priority of the tincd process will be adjusted.
+Increasing the priority may help to reduce latency and packet loss on the VPN.
+
+.It Va ReplayWindow Li = Ar bytes Pq 16
+This is the size of the replay tracking window for each remote node, in bytes.
+The window is a bitfield which tracks 1 packet per bit, so for example
+the default setting of 16 will track up to 128 packets in the window. In high
+bandwidth scenarios, setting this to a higher value can reduce packet loss from
+the interaction of replay tracking with underlying real packet loss and/or
+reordering. Setting this to zero will disable replay tracking completely and
+pass all traffic, but leaves tinc vulnerable to replay-based attacks on your
+traffic.
+
+.It Va StrictSubnets Li = yes | no Po no Pc Bq experimental
+When this option is enabled tinc will only use Subnet statements which are
+present in the host config files in the local
+.Pa @sysconfdir@/tinc/ Ns Ar NETNAME Ns Pa /hosts/
+directory.
+
.It Va TunnelServer Li = yes | no Po no Pc Bq experimental
When this option is enabled tinc will no longer forward information between other tinc daemons,
-and will only allow nodes and subnets on the VPN which are present in the
+and will only allow connections with nodes for which host config files are present in the local
.Pa @sysconfdir@/tinc/ Ns Ar NETNAME Ns Pa /hosts/
directory.
+Setting this options also implicitly sets StrictSubnets.
+
+.It Va UDPRcvBuf Li = Ar bytes Pq OS default
+Sets the socket receive buffer size for the UDP socket, in bytes.
+If unset, the default buffer size will be used by the operating system.
+
+.It Va UDPSndBuf Li = Ar bytes Pq OS default
+Sets the socket send buffer size for the UDP socket, in bytes.
+If unset, the default buffer size will be used by the operating system.
.El
.Sh HOST CONFIGURATION FILES
no secrets are revealed by sending out this information.
.Bl -tag -width indent
-.It Va Address Li = Ar address Bq recommended
+.It Va Address Li = Ar address Oo Ar port Oc Bq recommended
The IP address or hostname of this tinc daemon on the real network.
This will only be used when trying to make an outgoing connection to this tinc daemon.
+Optionally, a port can be specified to use for this address.
Multiple
.Va Address
variables can be specified, in which case each address will be tried until a working
will turn off packet encryption.
It is best to use only those ciphers which support CBC mode.
+.It Va ClampMSS Li = yes | no Pq yes
+This option specifies whether tinc should clamp the maximum segment size (MSS)
+of TCP packets to the path MTU. This helps in situations where ICMP
+Fragmentation Needed or Packet too Big messages are dropped by firewalls.
+
.It Va Compression Li = Ar level Pq 0
This option sets the level of compression used for UDP packets.
Possible values are 0 (off), 1 (fast zlib) and any integer up to 9 (best zlib),
After the path MTU has been discovered, it will be enforced on the VPN.
.It Va Port Li = Ar port Pq 655
-The port number on which this tinc daemon is listening for incoming connections.
+The port number on which this tinc daemon is listening for incoming connections,
+which is used if no port number is specified in an
+.Va Address
+statement.
.It Va PublicKey Li = Ar key Bq obsolete
The public RSA key of this tinc daemon.
in each host configuration file,
if you want to be able to establish a connection with that host.
-.It Va Subnet Li = Ar address Ns Op Li / Ns Ar prefixlength
+.It Va Subnet Li = Ar address Ns Op Li / Ns Ar prefixlength Ns Op Li # Ns Ar weight
The subnet which this tinc daemon will serve.
.Nm tinc
tries to look up which other daemon it should send a packet to by searching the appropriate subnet.
IPv6 subnets are notated like fec0:0:0:1:0:0:0:0/64.
MAC addresses are notated like 0:1a:2b:3c:4d:5e.
-.It Va TCPOnly Li = yes | no Pq no
+.Pp
+A Subnet can be given a weight to indicate its priority over identical Subnets
+owned by different nodes. The default weight is 10. Lower values indicate
+higher priority. Packets will be sent to the node with the highest priority,
+unless that node is not reachable, in which case the node with the next highest
+priority will be tried, and so on.
+
+.It Va TCPOnly Li = yes | no Pq no Bq obsolete
If this variable is set to yes,
then the packets are tunnelled over the TCP connection instead of a UDP connection.
This is especially useful for those who want to run a tinc daemon
from behind a masquerading firewall,
or if UDP packet routing is disabled somehow.
Setting this options also implicitly sets IndirectData.
+
+.Pp
+Since version 1.0.10, tinc will automatically detect whether communication via
+UDP is possible or not.
.El
.Sh SCRIPTS
.It Ev SUBNET
When a subnet becomes (un)reachable, this is set to the subnet.
+
+.It Ev WEIGHT
+When a subnet becomes (un)reachable, this is set to the subnet weight.
.El
.Sh FILES
projects / tinc / blobdiff