IPNAT(8) OpenBSD System Manager's Manual IPNAT(8) NAME ipnat - manage IP network address translation rules SYNOPSIS ipnat [-CFlnrsv] [-f filename] DESCRIPTION The ipnat utility provides control over the kernel's network address translation (NAT). The NAT facility remaps IP addresses from one range the another. It can be used to provide internal networks with Internet connectivity by mapping several private IP addresses to a single route- able (i.e., ``real'') Internet address. In other words, when properly configured on a gateway, the NAT provides Internet access to connected computers lacking officially assigned IP ad- dresses. It is discussed in RFC 1631. The options are as follows: -f filename File from which rules are read. -C Delete all entries in the NAT list. -F Flush all active mappings from the NAT table. -l Display the current rule list and active mappings. -n Do not alter the NAT table. -r Remove, rather than add, entries specified in the rule list. -s Display statistics. -v Verbosity. Displays detailed information pertaining to rule pro- cessing. Certain configuration requirements must be met before ipnat will work: 1. Network address translation requires packet forwarding capa- bilities. Ensure the /etc/sysctl.conf file contains the as- signment net.inet.ip.forwarding=1. 2. Packet filtering (see ipf(8)) must be enabled, even if it's not being used. Check the /etc/rc.conf file and make sure it contains the assignment ipfilter=YES. 3. The kernel must be configured with option IPFILTER (and option IPFILTER_LOG if ipmon(8) is needed). Both options are com- piled into the default (GENERIC) kernel that comes with the system. 4. Finally, enable NAT itself by setting ipnat=YES in /etc/rc.conf. This will cause /etc/netstart to run ipnat at boot-time with /etc/ipnat.rules as the rule list to install. The ipnat utility operates on a list of rules, specified by -f filename. This file is typically /etc/ipnat.rules; standard input is represented by a single dash (`-'). Each rule is parsed, then sequentially added to the kernel's internal NAT list. Like ipf(8), if an entry contradicts another previously added, the newer will take precedence. Comments (beginning with a `#') and blank lines are ignored as ipnat parses the file. Entries may be separated by spaces or tabs. Each rule must begin with either map, bimap, or rdr. See below for rule syntax, or refer to /usr/share/ipf/nat.1 for sample rule entries. Mapping rules map tells the NAT how a range of addresses should be translated. The en- tries use the following format: map ifname internal/mask -> external/mask options The ifname field is the interface to which packets are sent. A gateway with a PPP link would probably use ``ppp0'' or ``tun0'', while an Ether- net connection would instead have the name of its device. In the pres- ence of multiple network devices, you wish to use the device which is on the external side. As a quick example: map ep1 10.1.1.0/24 -> ep1/32 portmap tcp/udp 10000:20000 This rule would remap all connections originating from 10.1.1.0 through 10.1.1.254 to the externally-connected network. Note that ``ep1'' is the name of the outside interface on the gateway; that is, the interface with the external (i.e., ``real'') IP address. Do not specify internal inter- face names, use their addresses instead. The address range of the LAN goes in the internal field. This is usually one of the three blocks of address space the Internet Assigned Numbers Authority has allocated for private networks (RFC 1597): 10.0.0.0 - 10.255.255.255 (ie. 10/24) 172.16.0.0 - 172.31.255.255 (ie. 172.16/20) 192.168.0.0 - 192.168.255.255 (ie. 192.168/16) The external address is the offically assigned IP number of the gateway or network. mask is the netmask of the address. This mask is 32 bits long, and is divided into four 8-bit numbers. 11111111.0.0.0 Class A - 8 bits set. 11111111.11111111.0.0 Class B - 16 bits set. 11111111.11111111.11111111.0 Class C - 24 bits set. The number of bits set in the mask is placed following the IP address. Both internal and external may be an actual IP address, the name of an interface, or a hostname. If it is a network number, however, a problem may arise. For example: map ppp0 10.0.0.0/8 -> 209.1.2.0/24 16,000,000 IP addresses are being squeezed into an address space of only 254. This is solved by the portmap option, which remaps ports instead of IP addresses. The protocol is specified by following the option with ei- ther tcp, udp, tcp/udp, or tcpudp (the last two have the same effect). The syntax to assign a range of ports is ``portnumber:portnumber''. This looks like: map ppp0 10.0.0.0/8 -> 209.1.2.0/24 portmap tcp/udp 1025:65000 map ppp0 10.0.0.0/8 -> 209.1.2.0/24 That will cut the number down from ~16,000,000 addresses short to only 527,566. Bi-directional mapping rules bimap is used to create static, bidirectional NAT mappings. Standard map rules only create NAT mappings when the connection is initiated from the internal IP address. For example, using the following rule: map ppp0 10.0.0.3/32 -> 209.1.2.3/32 NAT mappings will only be created if the machine at 10.0.0.3 initiates the connection. To create a truly bidirectional NAT entry, bimap is nec- essary. Using the following rule, for example, clients on the ppp0 side of the NAT box can initiate requests to 209.1.2.3. This traffic will be mapped to 10.0.0.3 as expected: bimap ppp0 10.0.0.3/32 -> 209.1.2.3/32 To be genuinely useful, bimap should be used in conjunction with either proxy arp, or ifconfig(8) aliases. For example, if we create two bimap entries such as: bimap fxp0 10.0.0.3/32 -> 209.1.2.3/32 bimap fxp0 10.0.0.4/32 -> 209.1.2.4/32 It is necessary to do either: arp -s 209.1.2.3 00:40:aa:bb:cc:dd pub arp -s 209.1.2.4 00:40:aa:bb:cc:dd pub (where 00:40:aa:bb:cc:dd is the MAC address of fxp0) or ifconfig fxp0 alias 209.1.2.3 netmask 255.255.255.255 ifconfig fxp0 alias 209.1.2.4 netmask 255.255.255.255 Note that since ipnat(8) works on the principle of first match (as ap- posed to ipf(1) which is last match), it is customary to put all rdr rules before any and all (bi)map rules. This is particularly vital if the network ranges in question verlap. Otherwise the rdr rules simply will not work. Redirection rules rdr tells the NAT how to redirect incoming packets. It is useful if one wishes to redirect a connection through a proxy, or to another box on the private network. The format of this directive is: rdr ifname external/mask port service -> internal port service protocol This setup is best described by an example of an actual entry: rdr xl0 0.0.0.0/0 port 25 -> 204.213.176.10 port smtp This redirects all smtp packets received on xl0 to 204.213.176.10, port 25. A netmask is not needed on the internal address; it is always 32. The external and internal fields, similar to the map directive, may be actual addresses, hostnames, or interfaces. Likewise, the service field may be the name of a service, or a port number. The protocol of the ser- vice may be selected by appending tcp, udp, tcp/udp, or tcpudp (the last two have the same effect) to the end of the line. TCP is the default. FILES /etc/ipnat.rules default system rule list /usr/share/ipf/nat.1 example rules /usr/share/ipf/nat.2 system requirements for use of the NAT /dev/ipnat device file BUGS bimap should really only be used with single IP addresses (x.x.x.x/32). Bimapping other CIDR ranges will result in unexpected, and possibly ran- dom mappings into the destination address block. SEE ALSO ipnat(4), ipnat(5), ipf(8) http://coombs.anu.edu.au/~avalon OpenBSD 2.7 October 10, 1998 4