| NATD(8) | MidnightBSD System Manager's Manual | NATD(8) |
natd — Network
Address Translation daemon
natd |
[-unregistered_only
| -u]
[-log | -l]
[-proxy_only] [-reverse]
[-deny_incoming | -d]
[-use_sockets | -s]
[-same_ports | -m]
[-verbose | -v]
[-dynamic]
[-in_port | -i port]
[-out_port | -o port]
[-port | -p port]
[-alias_address | -a address]
[-target_address | -t address]
[-interface | -n interface]
[-proxy_rule proxyspec]
[-redirect_port linkspec]
[-redirect_proto linkspec]
[-redirect_address linkspec]
[-config | -f configfile]
[-instance instancename]
[-globalport port]
[-log_denied]
[-log_facility facility_name]
[-punch_fw firewall_range]
[-skinny_port port]
[-log_ipfw_denied]
[-pid_file | -P pidfile]
[-exit_delay | -P ms] |
The natd utility provides a Network
Address Translation facility for use with
divert(4) sockets under
FreeBSD.
(If you need NAT on a PPP link,
ppp(8) provides the
-nat option that gives most of the
natd functionality, and uses the same
libalias(3)
library.)
The natd utility normally runs in the
background as a daemon. It is passed raw IP packets as they travel into and
out of the machine, and will possibly change these before re-injecting them
back into the IP packet stream.
It changes all packets destined for another host so that their source IP address is that of the current machine. For each packet changed in this manner, an internal table entry is created to record this fact. The source port number is also changed to indicate the table entry applying to the packet. Packets that are received with a target IP of the current host are checked against this internal table. If an entry is found, it is used to determine the correct target IP address and port to place in the packet.
The following command line options are available:
-log
| -lnatd is started.-deny_incoming
| -dIf this option is not used, then such a packet will be altered
using the rules in -target_address below, and
the entry will be made in the internal translation table.
-log_denied-log_facility).-log_facility
facility_name-use_sockets
| -s-same_ports
| -m-verbose
| -v-unregistered_only
| -u-redirect_port
proto
targetIP:targetPORT[-targetPORT]
[aliasIP:]aliasPORT[-aliasPORT]
[remoteIP[:remotePORT[-remotePORT]]]Arguments targetIP, aliasIP and remoteIP can be given as IP addresses or as hostnames. The targetPORT, aliasPORT and remotePORT ranges need not be the same numerically, but must have the same size. When targetPORT, aliasPORT or remotePORT specifies a singular value (not a range), it can be given as a service name that is searched for in the services(5) database.
For example, the argument
tcp
inside1:telnet 6666means that incoming TCP packets destined for port 6666 on this machine will be sent to the telnet port on the inside1 machine.
tcp
inside2:2300-2399 3300-3399will redirect incoming connections on ports 3300-3399 to host inside2, ports 2300-2399. The mapping is 1:1 meaning port 3300 maps to 2300, 3301 maps to 2301, etc.
-redirect_proto
proto localIP [publicIP
[remoteIP]]If publicIP is not specified, then the default aliasing address is used. If remoteIP is specified, then only packets coming from/to remoteIP will match the rule.
-redirect_address
localIP publicIPredirect_address
10.0.0.8 0.0.0.0The above command would redirect all incoming traffic to machine 10.0.0.8.
If several address aliases specify the same public address as follows
redirect_address 192.168.0.2 public_addr redirect_address 192.168.0.3 public_addr redirect_address 192.168.0.4 public_addr
the incoming traffic will be directed to the last translated local address (192.168.0.4), but outgoing traffic from the first two addresses will still be aliased to appear from the specified public_addr.
-redirect_port
proto
targetIP:targetPORT[,targetIP:targetPORT[,...]]
[aliasIP:]aliasPORT
[remoteIP[:remotePORT]]-redirect_address
localIP[,localIP[,...]]
publicIP-redirect_port and
-redirect_address are used to transparently
offload network load on a single server and distribute the load across a
pool of servers. This function is known as
LSNAT
(RFC 2391). For example, the argument
tcp
www1:http,www2:http,www3:http www:httpmeans that incoming HTTP requests for host www will be transparently redirected to one of the www1, www2 or www3, where a host is selected simply on a round-robin basis, without regard to load on the net.
-dynamic-n or -interface
option is used, natd will monitor the routing
socket for alterations to the interface passed. If
the interface's IP address is changed, natd will
dynamically alter its concept of the alias address.-in_port
| -i
port-out_port
| -o
port-port
| -p
port-alias_address
| -a
address-interface option must be used (but not both),
if the -proxy_only option is not specified. The
specified address is usually the address assigned to the
“public” network interface.
All data passing
out will be
rewritten with a source address equal to address.
All data coming
in will be
checked to see if it matches any already-aliased outgoing connection. If
it does, the packet is altered accordingly. If not, all
-redirect_port,
-redirect_proto and
-redirect_address assignments are checked and
actioned. If no other action can be made and if
-deny_incoming is not specified, the packet is
delivered to the local machine using the rules specified in
-target_address option below.
-t |
-target_address
addressThe target address may be set to
255.255.255.255, in which case all new incoming
packets go to the alias address set by
-alias_address or
-interface.
If this option is not used, or called with the argument 0.0.0.0, then all new incoming packets go to the address specified in the packet. This allows external machines to talk directly to internal machines if they can route packets to the machine in question.
-interface
| -n
interface-dynamic option should also be used. If this
option is not specified, the -alias_address option
must be used.
The specified interface is usually the “public” (or “external”) network interface.
-config
| -f
filealias_address
158.152.17.1would specify an alias address of 158.152.17.1. Options that do not take an argument are specified with an argument of yes or no in the configuration file. For example, the line
log yesis synonymous with -log.
Options can be divided to several sections. Each section
applies to own natd instance. This ability
allows the configuration of one natd process for
several NAT instances. The first instance that always exists is a
"default" instance. Each another instance should begin
with
instance
instance_nameAt the next should be placed a configuration option. Example:
# default instanceport 8668alias_address
158.152.17.1# second instanceinstance dsl1port 8888alias_address
192.168.0.1Trailing spaces and empty lines are ignored. A
‘#’ sign will mark the rest of the
line as a comment.
-instance
instancename-instance option or end of command line. It
is easier to set up multiple instances in the configuration file specified
with the -config option rather than on a command
line.-globalport
port-reversenatd reverse the way it handles
“incoming” and “outgoing” packets, allowing it
to operate on the “internal” network interface rather than
the “external” one.
This can be useful in some transparent proxying situations
when outgoing traffic is redirected to the local machine and
natd is running on the internal interface (it
usually runs on the external interface).
-proxy_onlynatd to perform transparent proxying only.
Normal address translation is not performed.-proxy_rule
[type encode_ip_hdr |
encode_tcp_stream] port xxxx
server a.b.c.d:yyyy-punch_fw
basenumber:countnatd to “punch
holes” in an
ipfirewall(4) based
firewall for FTP/IRC DCC connections. This is done dynamically by
installing temporary firewall rules which allow a particular connection
(and only that connection) to go through the firewall. The rules are
removed once the corresponding connection terminates.
A maximum of count rules starting from the rule number basenumber will be used for punching firewall holes. The range will be cleared for all rules on startup. This option has no effect when the kernel is in security level 3, see init(8) for more information.
-skinny_port
port-log_ipfw_denied-verbose.-pid_file
| -P
file-exit_delay
msThe following steps are necessary before attempting to run
natd:
options IPFIREWALL options IPDIVERT
Refer to the handbook for detailed instructions on building a custom kernel.
gateway_enable=YESin the /etc/rc.conf file or using the command
sysctl
net.inet.ip.forwarding=1-interface option, make sure that
your interface is already configured. If, for example, you wish to specify
‘tun0’ as your
interface, and you are using
ppp(8) on that interface,
you must make sure that you start ppp prior to
starting natd.Running natd is fairly straight forward.
The line
natd -interface ed0should suffice in most cases (substituting the correct interface
name). Please check
rc.conf(5) on how to
configure it to be started automatically during boot. Once
natd is running, you must ensure that traffic is
diverted to natd:
/sbin/ipfw -f flush /sbin/ipfw add divert natd all from any to any via ed0 /sbin/ipfw add pass all from any to any
The second line depends on your interface (change
‘ed0’ as appropriate).
You should be aware of the fact that, with these firewall settings, everyone on your local network can fake his source-address using your host as gateway. If there are other hosts on your local network, you are strongly encouraged to create firewall rules that only allow traffic to and from trusted hosts.
If you specify real firewall rules, it is best to specify line
2 at the start of the script so that natd sees
all packets before they are dropped by the firewall.
After translation by natd, packets
re-enter the firewall at the rule number following the rule number that
caused the diversion (not the next rule if there are several at the same
number).
firewall_enable=YESin /etc/rc.conf. This tells the system startup scripts to run the /etc/rc.firewall script. If you do not wish to reboot now, just run this by hand from the console. NEVER run this from a remote session unless you put it into the background. If you do, you will lock yourself out after the flush takes place, and execution of /etc/rc.firewall will stop at this point - blocking all accesses permanently. Running the script in the background should be enough to prevent this disaster.
It is not so uncommon to have a need of aliasing to several
external IP addresses. While this traditionally was achieved by running
several natd processes with independent
configurations, natd can have multiple aliasing
instances in a single process, also allowing them to be not so independent
of each other. For example, let us see a common task of load balancing two
channels to different providers on a machine with two external interfaces
‘sis0’ (with IP 1.2.3.4) and
‘sis2’ (with IP 2.3.4.5):
net 1.2.3.0/24
1.2.3.1 ------------------ sis0
(router) (1.2.3.4)
net 10.0.0.0/24
sis1 ------------------- 10.0.0.2
(10.0.0.1)
net 2.3.4.0/24
2.3.4.1 ------------------ sis2
(router) (2.3.4.5)
Default route is out via
‘sis0’.
Interior machine (10.0.0.2) is accessible on TCP port 122 through
both exterior IPs, and outgoing connections choose a path randomly between
‘sis0’ and
‘sis2’.
The way this works is that natd.conf builds two instances of the aliasing engine.
In addition to these instances' private
divert(4) sockets, a
third socket called the “globalport” is created; packets sent
to natd via this one will be matched against all
instances and translated if an existing entry is found, and unchanged if no
entry is found. The following lines are placed into
/etc/natd.conf:
log deny_incoming verbose instance default interface sis0 port 1000 redirect_port tcp 10.0.0.2:122 122 instance sis2 interface sis2 port 2000 redirect_port tcp 10.0.0.2:122 122 globalport 3000
And the following ipfw(8) rules are used:
ipfw -f flush ipfw add allow ip from any to any via sis1 ipfw add skipto 1000 ip from any to any in via sis0 ipfw add skipto 2000 ip from any to any out via sis0 ipfw add skipto 3000 ip from any to any in via sis2 ipfw add skipto 4000 ip from any to any out via sis2 ipfw add 1000 count ip from any to any ipfw add divert 1000 ip from any to any ipfw add allow ip from any to any ipfw add 2000 count ip from any to any ipfw add divert 3000 ip from any to any ipfw add allow ip from 1.2.3.4 to any ipfw add skipto 5000 ip from 2.3.4.5 to any ipfw add prob .5 skipto 4000 ip from any to any ipfw add divert 1000 ip from any to any ipfw add allow ip from any to any ipfw add 3000 count ip from any to any ipfw add divert 2000 ip from any to any ipfw add allow ip from any to any ipfw add 4000 count ip from any to any ipfw add divert 2000 ip from any to any ipfw add 5000 fwd 2.3.4.1 ip from 2.3.4.5 to not 2.3.4.0/24 ipfw add allow ip from any to any
Here the packet from internal network to Internet goes out via
‘sis0’ (rule number 2000) and gets
caught by the globalport socket (3000). After that,
either a match is found in a translation table of one of the two instances,
or the packet is passed to one of the two other
divert(4) ports (1000 or
2000), with equal probability. This ensures that load balancing is done on a
per-flow basis (i.e., packets from a single TCP connection always flow
through the same interface). Translated packets with source IP of a
non-default interface (‘sis2’) are
forwarded to the appropriate router on that interface.
libalias(3), divert(4), protocols(5), rc.conf(5), services(5), syslog.conf(5), init(8), ipfw(8), ppp(8)
The natd utility appeared in
FreeBSD 3.0.
This program is the result of the efforts of many people at different times:
Archie Cobbs
<archie@FreeBSD.org>
(divert sockets)
Charles Mott
<cm@linktel.net>
(packet aliasing)
Eivind Eklund
<perhaps@yes.no> (IRC
support & misc additions)
Ari Suutari
<suutari@iki.fi> (natd)
Dru Nelson
<dnelson@redwoodsoft.com>
(early PPTP support)
Brian Somers
<brian@awfulhak.org>
(glue)
Ruslan Ermilov
<ru@FreeBSD.org> (natd,
packet aliasing, glue)
Poul-Henning Kamp
<phk@FreeBSD.org>
(multiple instances)
| October 5, 2016 | midnightbsd-3.1 |