1iptables-extensions(8) iptables 1.8.9 iptables-extensions(8)
2
6 iptables-extensions — list of extensions in the standard iptables dis‐
7 tribution
8
10 ip6tables [-m name [module-options...]] [-j target-name [target-op‐
11 tions...]
12 iptables [-m name [module-options...]] [-j target-name [target-op‐
14 tions...]
15
17 iptables can use extended packet matching modules with the -m or
18 --match options, followed by the matching module name; after these,
19 various extra command line options become available, depending on the
20 specific module. You can specify multiple extended match modules in
21 one line, and you can use the -h or --help options after the module has
22 been specified to receive help specific to that module. The extended
23 match modules are evaluated in the order they are specified in the
24 rule.
25 If the -p or --protocol was specified and if and only if an unknown op‐
27 tion is encountered, iptables will try load a match module of the same
28 name as the protocol, to try making the option available.
29 addrtype
31 This module matches packets based on their address type. Address types
32 are used within the kernel networking stack and categorize addresses
33 into various groups. The exact definition of that group depends on the
34 specific layer three protocol.
35 The following address types are possible:
37 UNSPEC an unspecified address (i.e. 0.0.0.0)
39 UNICAST
41 an unicast address
42 LOCAL a local address
44 BROADCAST
46 a broadcast address
47 ANYCAST
49 an anycast packet
50 MULTICAST
52 a multicast address
53 BLACKHOLE
55 a blackhole address
56 UNREACHABLE
58 an unreachable address
59 PROHIBIT
61 a prohibited address
62 THROW FIXME
64 NAT FIXME
66 XRESOLVE
68 [!] --src-type type
70 Matches if the source address is of given type
71 [!] --dst-type type
73 Matches if the destination address is of given type
74 --limit-iface-in
76 The address type checking can be limited to the interface the
77 packet is coming in. This option is only valid in the PREROUT‐
78 ING, INPUT and FORWARD chains. It cannot be specified with the
79 --limit-iface-out option.
80 --limit-iface-out
82 The address type checking can be limited to the interface the
83 packet is going out. This option is only valid in the POSTROUT‐
84 ING, OUTPUT and FORWARD chains. It cannot be specified with the
85 --limit-iface-in option.
86 ah (IPv6-specific)
88 This module matches the parameters in Authentication header of IPsec
89 packets.
90 [!] --ahspi spi[:spi]
92 Matches SPI.
93 [!] --ahlen length
95 Total length of this header in octets.
96 --ahres
98 Matches if the reserved field is filled with zero.
99 ah (IPv4-specific)
101 This module matches the SPIs in Authentication header of IPsec packets.
102 [!] --ahspi spi[:spi]
104 bpf
106 Match using Linux Socket Filter. Expects a path to an eBPF object or a
107 cBPF program in decimal format.
108 --object-pinned path
110 Pass a path to a pinned eBPF object.
111 Applications load eBPF programs into the kernel with the bpf() system
113 call and BPF_PROG_LOAD command and can pin them in a virtual filesystem
114 with BPF_OBJ_PIN. To use a pinned object in iptables, mount the bpf
115 filesystem using
116 mount -t bpf bpf ${BPF_MOUNT}
118 then insert the filter in iptables by path:
120 iptables -A OUTPUT -m bpf --object-pinned
122 ${BPF_MOUNT}/{PINNED_PATH} -j ACCEPT
123 --bytecode code
125 Pass the BPF byte code format as generated by the nfbpf_compile
126 utility.
127 The code format is similar to the output of the tcpdump -ddd command:
129 one line that stores the number of instructions, followed by one line
130 for each instruction. Instruction lines follow the pattern 'u16 u8 u8
131 u32' in decimal notation. Fields encode the operation, jump offset if
132 true, jump offset if false and generic multiuse field 'K'. Comments are
133 not supported.
134 For example, to read only packets matching 'ip proto 6', insert the
136 following, without the comments or trailing whitespace:
137 4 # number of instructions
139 48 0 0 9 # load byte ip->proto
140 21 0 1 6 # jump equal IPPROTO_TCP
141 6 0 0 1 # return pass (non-zero)
142 6 0 0 0 # return fail (zero)
143 You can pass this filter to the bpf match with the following command:
145 iptables -A OUTPUT -m bpf --bytecode '4,48 0 0 9,21 0 1 6,6 0 0
147 1,6 0 0 0' -j ACCEPT
148 Or instead, you can invoke the nfbpf_compile utility.
150 iptables -A OUTPUT -m bpf --bytecode "`nfbpf_compile RAW 'ip
152 proto 6'`" -j ACCEPT
153 Or use tcpdump -ddd. In that case, generate BPF targeting a device with
155 the same data link type as the xtables match. Iptables passes packets
156 from the network layer up, without mac layer. Select a device with data
157 link type RAW, such as a tun device:
158 ip tuntap add tun0 mode tun
160 ip link set tun0 up
161 tcpdump -ddd -i tun0 ip proto 6
162 See tcpdump -L -i $dev for a list of known data link types for a given
164 device.
165 You may want to learn more about BPF from FreeBSD's bpf(4) manpage.
167 cgroup
169 [!] --path path
170 Match cgroup2 membership.
171 Each socket is associated with the v2 cgroup of the creating
173 process. This matches packets coming from or going to all sock‐
174 ets in the sub-hierarchy of the specified path. The path should
175 be relative to the root of the cgroup2 hierarchy.
176 [!] --cgroup classid
178 Match cgroup net_cls classid.
179 classid is the marker set through the cgroup net_cls controller.
181 This option and --path can't be used together.
182 Example:
184 iptables -A OUTPUT -p tcp --sport 80 -m cgroup ! --path ser‐
186 vice/http-server -j DROP
187 iptables -A OUTPUT -p tcp --sport 80 -m cgroup ! --cgroup 1 -j
189 DROP
190 IMPORTANT: when being used in the INPUT chain, the cgroup matcher is
192 currently only of limited functionality, meaning it will only match on
193 packets that are processed for local sockets through early socket de‐
194 muxing. Therefore, general usage on the INPUT chain is not advised un‐
195 less the implications are well understood.
196 Available since Linux 3.14.
198 cluster
200 Allows you to deploy gateway and back-end load-sharing clusters without
201 the need of load-balancers.
202 This match requires that all the nodes see the same packets. Thus, the
204 cluster match decides if this node has to handle a packet given the
205 following options:
206 --cluster-total-nodes num
208 Set number of total nodes in cluster.
209 [!] --cluster-local-node num
211 Set the local node number ID.
212 [!] --cluster-local-nodemask mask
214 Set the local node number ID mask. You can use this option in‐
215 stead of --cluster-local-node.
216 --cluster-hash-seed value
218 Set seed value of the Jenkins hash.
219 Example:
221 iptables -A PREROUTING -t mangle -i eth1 -m cluster --clus‐
223 ter-total-nodes 2 --cluster-local-node 1 --cluster-hash-seed
224 0xdeadbeef -j MARK --set-mark 0xffff
225 iptables -A PREROUTING -t mangle -i eth2 -m cluster --clus‐
227 ter-total-nodes 2 --cluster-local-node 1 --cluster-hash-seed
228 0xdeadbeef -j MARK --set-mark 0xffff
229 iptables -A PREROUTING -t mangle -i eth1 -m mark ! --mark 0xffff
231 -j DROP
232 iptables -A PREROUTING -t mangle -i eth2 -m mark ! --mark 0xffff
234 -j DROP
235 And the following commands to make all nodes see the same packets:
237 ip maddr add 01:00:5e:00:01:01 dev eth1
239 ip maddr add 01:00:5e:00:01:02 dev eth2
241 arptables -A OUTPUT -o eth1 --h-length 6 -j mangle --mangle-mac-
243 s 01:00:5e:00:01:01
244 arptables -A INPUT -i eth1 --h-length 6 --destination-mac
246 01:00:5e:00:01:01 -j mangle --mangle-mac-d 00:zz:yy:xx:5a:27
247 arptables -A OUTPUT -o eth2 --h-length 6 -j mangle --man‐
249 gle-mac-s 01:00:5e:00:01:02
250 arptables -A INPUT -i eth2 --h-length 6 --destination-mac
252 01:00:5e:00:01:02 -j mangle --mangle-mac-d 00:zz:yy:xx:5a:27
253 NOTE: the arptables commands above use mainstream syntax. If you are
255 using arptables-jf included in some RedHat, CentOS and Fedora versions,
256 you will hit syntax errors. Therefore, you'll have to adapt these to
257 the arptables-jf syntax to get them working.
258 In the case of TCP connections, pickup facility has to be disabled to
260 avoid marking TCP ACK packets coming in the reply direction as valid.
261 echo 0 > /proc/sys/net/netfilter/nf_conntrack_tcp_loose
263 comment
265 Allows you to add comments (up to 256 characters) to any rule.
266 --comment comment
268 Example:
270 iptables -A INPUT -i eth1 -m comment --comment "my local LAN"
271 connbytes
273 Match by how many bytes or packets a connection (or one of the two
274 flows constituting the connection) has transferred so far, or by aver‐
275 age bytes per packet.
276 The counters are 64-bit and are thus not expected to overflow ;)
278 The primary use is to detect long-lived downloads and mark them to be
280 scheduled using a lower priority band in traffic control.
281 The transferred bytes per connection can also be viewed through `con‐
283 ntrack -L` and accessed via ctnetlink.
284 NOTE that for connections which have no accounting information, the
286 match will always return false. The "net.netfilter.nf_conntrack_acct"
287 sysctl flag controls whether new connections will be byte/packet
288 counted. Existing connection flows will not be gaining/losing a/the ac‐
289 counting structure when be sysctl flag is flipped.
290 [!] --connbytes from[:to]
292 match packets from a connection whose packets/bytes/average
293 packet size is more than FROM and less than TO bytes/packets. if
294 TO is omitted only FROM check is done. "!" is used to match
295 packets not falling in the range.
296 --connbytes-dir {original|reply|both}
298 which packets to consider
299 --connbytes-mode {packets|bytes|avgpkt}
301 whether to check the amount of packets, number of bytes trans‐
302 ferred or the average size (in bytes) of all packets received so
303 far. Note that when "both" is used together with "avgpkt", and
304 data is going (mainly) only in one direction (for example HTTP),
305 the average packet size will be about half of the actual data
306 packets.
307 Example:
309 iptables .. -m connbytes --connbytes 10000:100000
310 --connbytes-dir both --connbytes-mode bytes ...
311 connlabel
313 Module matches or adds connlabels to a connection. connlabels are sim‐
314 ilar to connmarks, except labels are bit-based; i.e. all labels may be
315 attached to a flow at the same time. Up to 128 unique labels are cur‐
316 rently supported.
317 [!] --label name
319 matches if label name has been set on a connection. Instead of
320 a name (which will be translated to a number, see EXAMPLE be‐
321 low), a number may be used instead. Using a number always over‐
322 rides connlabel.conf.
323 --set if the label has not been set on the connection, set it. Note
325 that setting a label can fail. This is because the kernel allo‐
326 cates the conntrack label storage area when the connection is
327 created, and it only reserves the amount of memory required by
328 the ruleset that exists at the time the connection is created.
329 In this case, the match will fail (or succeed, in case --label
330 option was negated).
331 This match depends on libnetfilter_conntrack 1.0.4 or later. Label
333 translation is done via the /etc/xtables/connlabel.conf configuration
334 file.
335 Example:
337 0 eth0-in
339 1 eth0-out
340 2 ppp-in
341 3 ppp-out
342 4 bulk-traffic
343 5 interactive
344 connlimit
346 Allows you to restrict the number of parallel connections to a server
347 per client IP address (or client address block).
348 --connlimit-upto n
350 Match if the number of existing connections is below or equal n.
351 --connlimit-above n
353 Match if the number of existing connections is above n.
354 --connlimit-mask prefix_length
356 Group hosts using the prefix length. For IPv4, this must be a
357 number between (including) 0 and 32. For IPv6, between 0 and
358 128. If not specified, the maximum prefix length for the appli‐
359 cable protocol is used.
360 --connlimit-saddr
362 Apply the limit onto the source group. This is the default if
363 --connlimit-daddr is not specified.
364 --connlimit-daddr
366 Apply the limit onto the destination group.
367 Examples:
369 # allow 2 telnet connections per client host
371 iptables -A INPUT -p tcp --syn --dport 23 -m connlimit
372 --connlimit-above 2 -j REJECT
373 # you can also match the other way around:
375 iptables -A INPUT -p tcp --syn --dport 23 -m connlimit
376 --connlimit-upto 2 -j ACCEPT
377 # limit the number of parallel HTTP requests to 16 per class C sized
379 source network (24 bit netmask)
380 iptables -p tcp --syn --dport 80 -m connlimit --connlimit-above
381 16 --connlimit-mask 24 -j REJECT
382 # limit the number of parallel HTTP requests to 16 for the link local
384 network
385 (ipv6) ip6tables -p tcp --syn --dport 80 -s fe80::/64 -m
386 connlimit --connlimit-above 16 --connlimit-mask 64 -j REJECT
387 # Limit the number of connections to a particular host:
389 ip6tables -p tcp --syn --dport 49152:65535 -d 2001:db8::1 -m
390 connlimit --connlimit-above 100 -j REJECT
391 connmark
393 This module matches the netfilter mark field associated with a connec‐
394 tion (which can be set using the CONNMARK target below).
395 [!] --mark value[/mask]
397 Matches packets in connections with the given mark value (if a
398 mask is specified, this is logically ANDed with the mark before
399 the comparison).
400 conntrack
402 This module, when combined with connection tracking, allows access to
403 the connection tracking state for this packet/connection.
404 [!] --ctstate statelist
406 statelist is a comma separated list of the connection states to
407 match. Possible states are listed below.
408 [!] --ctproto l4proto
410 Layer-4 protocol to match (by number or name)
411 [!] --ctorigsrc address[/mask]
413 [!] --ctorigdst address[/mask]
415 [!] --ctreplsrc address[/mask]
417 [!] --ctrepldst address[/mask]
419 Match against original/reply source/destination address
420 [!] --ctorigsrcport port[:port]
422 [!] --ctorigdstport port[:port]
424 [!] --ctreplsrcport port[:port]
426 [!] --ctrepldstport port[:port]
428 Match against original/reply source/destination port
429 (TCP/UDP/etc.) or GRE key. Matching against port ranges is only
430 supported in kernel versions above 2.6.38.
431 [!] --ctstatus statelist
433 statuslist is a comma separated list of the connection statuses
434 to match. Possible statuses are listed below.
435 [!] --ctexpire time[:time]
437 Match remaining lifetime in seconds against given value or range
438 of values (inclusive)
439 --ctdir {ORIGINAL|REPLY}
441 Match packets that are flowing in the specified direction. If
442 this flag is not specified at all, matches packets in both di‐
443 rections.
444 States for --ctstate:
446 INVALID
448 The packet is associated with no known connection.
449 NEW The packet has started a new connection or otherwise associated
451 with a connection which has not seen packets in both directions.
452 ESTABLISHED
454 The packet is associated with a connection which has seen pack‐
455 ets in both directions.
456 RELATED
458 The packet is starting a new connection, but is associated with
459 an existing connection, such as an FTP data transfer or an ICMP
460 error.
461 UNTRACKED
463 The packet is not tracked at all, which happens if you explic‐
464 itly untrack it by using -j CT --notrack in the raw table.
465 SNAT A virtual state, matching if the original source address differs
467 from the reply destination.
468 DNAT A virtual state, matching if the original destination differs
470 from the reply source.
471 Statuses for --ctstatus:
473 NONE None of the below.
475 EXPECTED
477 This is an expected connection (i.e. a conntrack helper set it
478 up).
479 SEEN_REPLY
481 Conntrack has seen packets in both directions.
482 ASSURED
484 Conntrack entry should never be early-expired.
485 CONFIRMED
487 Connection is confirmed: originating packet has left box.
488 cpu
490 [!] --cpu number
491 Match cpu handling this packet. cpus are numbered from 0 to
492 NR_CPUS-1 Can be used in combination with RPS (Remote Packet
493 Steering) or multiqueue NICs to spread network traffic on dif‐
494 ferent queues.
495 Example:
497 iptables -t nat -A PREROUTING -p tcp --dport 80 -m cpu --cpu 0 -j REDI‐
499 RECT --to-port 8080
500 iptables -t nat -A PREROUTING -p tcp --dport 80 -m cpu --cpu 1 -j REDI‐
502 RECT --to-port 8081
503 Available since Linux 2.6.36.
505 dccp
507 [!] --source-port,--sport port[:port]
508 [!] --destination-port,--dport port[:port]
510 [!] --dccp-types mask
512 Match when the DCCP packet type is one of 'mask'. 'mask' is a
513 comma-separated list of packet types. Packet types are: REQUEST
514 RESPONSE DATA ACK DATAACK CLOSEREQ CLOSE RESET SYNC SYNCACK IN‐
515 VALID.
516 [!] --dccp-option number
518 Match if DCCP option set.
519 devgroup
521 Match device group of a packet's incoming/outgoing interface.
522 [!] --src-group name
524 Match device group of incoming device
525 [!] --dst-group name
527 Match device group of outgoing device
528 dscp
530 This module matches the 6 bit DSCP field within the TOS field in the IP
531 header. DSCP has superseded TOS within the IETF.
532 [!] --dscp value
534 Match against a numeric (decimal or hex) value [0-63].
535 [!] --dscp-class class
537 Match the DiffServ class. This value may be any of the BE, EF,
538 AFxx or CSx classes. It will then be converted into its accord‐
539 ing numeric value.
540 dst (IPv6-specific)
542 This module matches the parameters in Destination Options header
543 [!] --dst-len length
545 Total length of this header in octets.
546 --dst-opts type[:length][,type[:length]...]
548 numeric type of option and the length of the option data in
549 octets.
550 ecn
552 This allows you to match the ECN bits of the IPv4/IPv6 and TCP header.
553 ECN is the Explicit Congestion Notification mechanism as specified in
554 RFC3168
555 [!] --ecn-tcp-cwr
557 This matches if the TCP ECN CWR (Congestion Window Received) bit
558 is set.
559 [!] --ecn-tcp-ece
561 This matches if the TCP ECN ECE (ECN Echo) bit is set.
562 [!] --ecn-ip-ect num
564 This matches a particular IPv4/IPv6 ECT (ECN-Capable Transport).
565 You have to specify a number between `0' and `3'.
566 esp
568 This module matches the SPIs in ESP header of IPsec packets.
569 [!] --espspi spi[:spi]
571 eui64 (IPv6-specific)
573 This module matches the EUI-64 part of a stateless autoconfigured IPv6
574 address. It compares the EUI-64 derived from the source MAC address in
575 Ethernet frame with the lower 64 bits of the IPv6 source address. But
576 "Universal/Local" bit is not compared. This module doesn't match other
577 link layer frame, and is only valid in the PREROUTING, INPUT and FOR‐
578 WARD chains.
579 frag (IPv6-specific)
581 This module matches the parameters in Fragment header.
582 [!] --fragid id[:id]
584 Matches the given Identification or range of it.
585 [!] --fraglen length
587 This option cannot be used with kernel version 2.6.10 or later.
588 The length of Fragment header is static and this option doesn't
589 make sense.
590 --fragres
592 Matches if the reserved fields are filled with zero.
593 --fragfirst
595 Matches on the first fragment.
596 --fragmore
598 Matches if there are more fragments.
599 --fraglast
601 Matches if this is the last fragment.
602 hashlimit
604 hashlimit uses hash buckets to express a rate limiting match (like the
605 limit match) for a group of connections using a single iptables rule.
606 Grouping can be done per-hostgroup (source and/or destination address)
607 and/or per-port. It gives you the ability to express "N packets per
608 time quantum per group" or "N bytes per seconds" (see below for some
609 examples).
610 A hash limit option (--hashlimit-upto, --hashlimit-above) and --hash‐
612 limit-name are required.
613 --hashlimit-upto amount[/second|/minute|/hour|/day]
615 Match if the rate is below or equal to amount/quantum. It is
616 specified either as a number, with an optional time quantum suf‐
617 fix (the default is 3/hour), or as amountb/second (number of
618 bytes per second).
619 --hashlimit-above amount[/second|/minute|/hour|/day]
621 Match if the rate is above amount/quantum.
622 --hashlimit-burst amount
624 Maximum initial number of packets to match: this number gets
625 recharged by one every time the limit specified above is not
626 reached, up to this number; the default is 5. When byte-based
627 rate matching is requested, this option specifies the amount of
628 bytes that can exceed the given rate. This option should be
629 used with caution -- if the entry expires, the burst value is
630 reset too.
631 --hashlimit-mode {srcip|srcport|dstip|dstport},...
633 A comma-separated list of objects to take into consideration. If
634 no --hashlimit-mode option is given, hashlimit acts like limit,
635 but at the expensive of doing the hash housekeeping.
636 --hashlimit-srcmask prefix
638 When --hashlimit-mode srcip is used, all source addresses en‐
639 countered will be grouped according to the given prefix length
640 and the so-created subnet will be subject to hashlimit. prefix
641 must be between (inclusive) 0 and 32. Note that --hashlimit-src‐
642 mask 0 is basically doing the same thing as not specifying srcip
643 for --hashlimit-mode, but is technically more expensive.
644 --hashlimit-dstmask prefix
646 Like --hashlimit-srcmask, but for destination addresses.
647 --hashlimit-name foo
649 The name for the /proc/net/ipt_hashlimit/foo entry.
650 --hashlimit-htable-size buckets
652 The number of buckets of the hash table
653 --hashlimit-htable-max entries
655 Maximum entries in the hash.
656 --hashlimit-htable-expire msec
658 After how many milliseconds do hash entries expire.
659 --hashlimit-htable-gcinterval msec
661 How many milliseconds between garbage collection intervals.
662 --hashlimit-rate-match
664 Classify the flow instead of rate-limiting it. This acts like a
665 true/false match on whether the rate is above/below a certain
666 number
667 --hashlimit-rate-interval sec
669 Can be used with --hashlimit-rate-match to specify the interval
670 at which the rate should be sampled
671 Examples:
673 matching on source host
675 "1000 packets per second for every host in 192.168.0.0/16" => -s
676 192.168.0.0/16 --hashlimit-mode srcip --hashlimit-upto 1000/sec
677 matching on source port
679 "100 packets per second for every service of 192.168.1.1" => -s
680 192.168.1.1 --hashlimit-mode srcport --hashlimit-upto 100/sec
681 matching on subnet
683 "10000 packets per minute for every /28 subnet (groups of 8 ad‐
684 dresses) in 10.0.0.0/8" => -s 10.0.0.0/8 --hashlimit-mask 28
685 --hashlimit-upto 10000/min
686 matching bytes per second
688 "flows exceeding 512kbyte/s" => --hashlimit-mode srcip,dstip,sr‐
689 cport,dstport --hashlimit-above 512kb/s
690 matching bytes per second
692 "hosts that exceed 512kbyte/s, but permit up to 1Megabytes with‐
693 out matching" --hashlimit-mode dstip --hashlimit-above 512kb/s
694 --hashlimit-burst 1mb
695 hbh (IPv6-specific)
697 This module matches the parameters in Hop-by-Hop Options header
698 [!] --hbh-len length
700 Total length of this header in octets.
701 --hbh-opts type[:length][,type[:length]...]
703 numeric type of option and the length of the option data in
704 octets.
705 helper
707 This module matches packets related to a specific conntrack-helper.
708 [!] --helper string
710 Matches packets related to the specified conntrack-helper.
711 string can be "ftp" for packets related to a ftp-session on de‐
713 fault port. For other ports append -portnr to the value, ie.
714 "ftp-2121".
715 Same rules apply for other conntrack-helpers.
717 hl (IPv6-specific)
719 This module matches the Hop Limit field in the IPv6 header.
720 [!] --hl-eq value
722 Matches if Hop Limit equals value.
723 --hl-lt value
725 Matches if Hop Limit is less than value.
726 --hl-gt value
728 Matches if Hop Limit is greater than value.
729 icmp (IPv4-specific)
731 This extension can be used if `--protocol icmp' is specified. It pro‐
732 vides the following option:
733 [!] --icmp-type {type[/code]|typename}
735 This allows specification of the ICMP type, which can be a nu‐
736 meric ICMP type, type/code pair, or one of the ICMP type names
737 shown by the command
738 iptables -p icmp -h
739 icmp6 (IPv6-specific)
741 This extension can be used if `--protocol ipv6-icmp' or `--protocol
742 icmpv6' is specified. It provides the following option:
743 [!] --icmpv6-type type[/code]|typename
745 This allows specification of the ICMPv6 type, which can be a nu‐
746 meric ICMPv6 type, type and code, or one of the ICMPv6 type
747 names shown by the command
748 ip6tables -p ipv6-icmp -h
749 iprange
751 This matches on a given arbitrary range of IP addresses.
752 [!] --src-range from[-to]
754 Match source IP in the specified range.
755 [!] --dst-range from[-to]
757 Match destination IP in the specified range.
758 ipv6header (IPv6-specific)
760 This module matches IPv6 extension headers and/or upper layer header.
761 --soft Matches if the packet includes any of the headers specified with
763 --header.
764 [!] --header header[,header...]
766 Matches the packet which EXACTLY includes all specified headers.
767 The headers encapsulated with ESP header are out of scope. Pos‐
768 sible header types can be:
769 hop|hop-by-hop
771 Hop-by-Hop Options header
772 dst Destination Options header
774 route Routing header
776 frag Fragment header
778 auth Authentication header
780 esp Encapsulating Security Payload header
782 none No Next header which matches 59 in the 'Next Header field' of
784 IPv6 header or any IPv6 extension headers
785 prot which matches any upper layer protocol header. A protocol name
787 from /etc/protocols and numeric value also allowed. The number
788 255 is equivalent to prot.
789 ipvs
791 Match IPVS connection properties.
792 [!] --ipvs
794 packet belongs to an IPVS connection
795 Any of the following options implies --ipvs (even negated)
797 [!] --vproto protocol
799 VIP protocol to match; by number or name, e.g. "tcp"
800 [!] --vaddr address[/mask]
802 VIP address to match
803 [!] --vport port
805 VIP port to match; by number or name, e.g. "http"
806 --vdir {ORIGINAL|REPLY}
808 flow direction of packet
809 [!] --vmethod {GATE|IPIP|MASQ}
811 IPVS forwarding method used
812 [!] --vportctl port
814 VIP port of the controlling connection to match, e.g. 21 for FTP
815 length
817 This module matches the length of the layer-3 payload (e.g. layer-4
818 packet) of a packet against a specific value or range of values.
819 [!] --length length[:length]
821 limit
823 This module matches at a limited rate using a token bucket filter. A
824 rule using this extension will match until this limit is reached. It
825 can be used in combination with the LOG target to give limited logging,
826 for example.
827 xt_limit has no negation support - you will have to use -m hashlimit !
829 --hashlimit rate in this case whilst omitting --hashlimit-mode.
830 --limit rate[/second|/minute|/hour|/day]
832 Maximum average matching rate: specified as a number, with an
833 optional `/second', `/minute', `/hour', or `/day' suffix; the
834 default is 3/hour.
835 --limit-burst number
837 Maximum initial number of packets to match: this number gets
838 recharged by one every time the limit specified above is not
839 reached, up to this number; the default is 5.
840 mac
842 [!] --mac-source address
843 Match source MAC address. It must be of the form
844 XX:XX:XX:XX:XX:XX. Note that this only makes sense for packets
845 coming from an Ethernet device and entering the PREROUTING, FOR‐
846 WARD or INPUT chains.
847 mark
849 This module matches the netfilter mark field associated with a packet
850 (which can be set using the MARK target below).
851 [!] --mark value[/mask]
853 Matches packets with the given unsigned mark value (if a mask is
854 specified, this is logically ANDed with the mask before the com‐
855 parison).
856 mh (IPv6-specific)
858 This extension is loaded if `--protocol ipv6-mh' or `--protocol mh' is
859 specified. It provides the following option:
860 [!] --mh-type type[:type]
862 This allows specification of the Mobility Header(MH) type, which
863 can be a numeric MH type, type or one of the MH type names shown
864 by the command
865 ip6tables -p mh -h
866 multiport
868 This module matches a set of source or destination ports. Up to 15
869 ports can be specified. A port range (port:port) counts as two ports.
870 It can only be used in conjunction with one of the following protocols:
871 tcp, udp, udplite, dccp and sctp.
872 [!] --source-ports,--sports port[,port|,port:port]...
874 Match if the source port is one of the given ports. The flag
875 --sports is a convenient alias for this option. Multiple ports
876 or port ranges are separated using a comma, and a port range is
877 specified using a colon. 53,1024:65535 would therefore match
878 ports 53 and all from 1024 through 65535.
879 [!] --destination-ports,--dports port[,port|,port:port]...
881 Match if the destination port is one of the given ports. The
882 flag --dports is a convenient alias for this option.
883 [!] --ports port[,port|,port:port]...
885 Match if either the source or destination ports are equal to one
886 of the given ports.
887 nfacct
889 The nfacct match provides the extended accounting infrastructure for
890 iptables. You have to use this match together with the standalone
891 user-space utility nfacct(8)
892 The only option available for this match is the following:
894 --nfacct-name name
896 This allows you to specify the existing object name that will be
897 use for accounting the traffic that this rule-set is matching.
898 To use this extension, you have to create an accounting object:
900 nfacct add http-traffic
902 Then, you have to attach it to the accounting object via iptables:
904 iptables -I INPUT -p tcp --sport 80 -m nfacct --nfacct-name
906 http-traffic
907 iptables -I OUTPUT -p tcp --dport 80 -m nfacct --nfacct-name
909 http-traffic
910 Then, you can check for the amount of traffic that the rules match:
912 nfacct get http-traffic
914 { pkts = 00000000000000000156, bytes = 00000000000000151786 } =
916 http-traffic;
917 You can obtain nfacct(8) from http://www.netfilter.org or, alterna‐
919 tively, from the git.netfilter.org repository.
920 osf
922 The osf module does passive operating system fingerprinting. This mod‐
923 ule compares some data (Window Size, MSS, options and their order, TTL,
924 DF, and others) from packets with the SYN bit set.
925 [!] --genre string
927 Match an operating system genre by using a passive fingerprint‐
928 ing.
929 --ttl level
931 Do additional TTL checks on the packet to determine the operat‐
932 ing system. level can be one of the following values:
933 • 0 - True IP address and fingerprint TTL comparison. This generally
935 works for LANs.
936 • 1 - Check if the IP header's TTL is less than the fingerprint one.
938 Works for globally-routable addresses.
939 • 2 - Do not compare the TTL at all.
941 --log level
943 Log determined genres into dmesg even if they do not match the de‐
944 sired one. level can be one of the following values:
945 • 0 - Log all matched or unknown signatures
947 • 1 - Log only the first one
949 • 2 - Log all known matched signatures
951 You may find something like this in syslog:
953 Windows [2000:SP3:Windows XP Pro SP1, 2000 SP3]: 11.22.33.55:4024 ->
955 11.22.33.44:139 hops=3 Linux [2.5-2.6:] : 1.2.3.4:42624 -> 1.2.3.5:22
956 hops=4
957 OS fingerprints are loadable using the nfnl_osf program. To load fin‐
959 gerprints from a file, use:
960 nfnl_osf -f /usr/share/xtables/pf.os
962 To remove them again,
964 nfnl_osf -f /usr/share/xtables/pf.os -d
966 The fingerprint database can be downloaded from http://www.open‐
968 bsd.org/cgi-bin/cvsweb/src/etc/pf.os .
969 owner
971 This module attempts to match various characteristics of the packet
972 creator, for locally generated packets. This match is only valid in the
973 OUTPUT and POSTROUTING chains. Forwarded packets do not have any socket
974 associated with them. Packets from kernel threads do have a socket, but
975 usually no owner.
976 [!] --uid-owner username
978 [!] --uid-owner userid[-userid]
980 Matches if the packet socket's file structure (if it has one) is
981 owned by the given user. You may also specify a numerical UID,
982 or an UID range.
983 [!] --gid-owner groupname
985 [!] --gid-owner groupid[-groupid]
987 Matches if the packet socket's file structure is owned by the
988 given group. You may also specify a numerical GID, or a GID
989 range.
990 --suppl-groups
992 Causes group(s) specified with --gid-owner to be also checked in
993 the supplementary groups of a process.
994 [!] --socket-exists
996 Matches if the packet is associated with a socket.
997 physdev
999 This module matches on the bridge port input and output devices en‐
1000 slaved to a bridge device. This module is a part of the infrastructure
1001 that enables a transparent bridging IP firewall and is only useful for
1002 kernel versions above version 2.5.44.
1003 [!] --physdev-in name
1005 Name of a bridge port via which a packet is received (only for
1006 packets entering the INPUT, FORWARD and PREROUTING chains). If
1007 the interface name ends in a "+", then any interface which be‐
1008 gins with this name will match. If the packet didn't arrive
1009 through a bridge device, this packet won't match this option,
1010 unless '!' is used.
1011 [!] --physdev-out name
1013 Name of a bridge port via which a packet is going to be sent
1014 (for bridged packets entering the FORWARD and POSTROUTING
1015 chains). If the interface name ends in a "+", then any inter‐
1016 face which begins with this name will match.
1017 [!] --physdev-is-in
1019 Matches if the packet has entered through a bridge interface.
1020 [!] --physdev-is-out
1022 Matches if the packet will leave through a bridge interface.
1023 [!] --physdev-is-bridged
1025 Matches if the packet is being bridged and therefore is not be‐
1026 ing routed. This is only useful in the FORWARD and POSTROUTING
1027 chains.
1028 pkttype
1030 This module matches the link-layer packet type.
1031 [!] --pkt-type {unicast|broadcast|multicast}
1033 policy
1035 This module matches the policy used by IPsec for handling a packet.
1036 --dir {in|out}
1038 Used to select whether to match the policy used for decapsula‐
1039 tion or the policy that will be used for encapsulation. in is
1040 valid in the PREROUTING, INPUT and FORWARD chains, out is valid
1041 in the POSTROUTING, OUTPUT and FORWARD chains.
1042 --pol {none|ipsec}
1044 Matches if the packet is subject to IPsec processing. --pol none
1045 cannot be combined with --strict.
1046 --strict
1048 Selects whether to match the exact policy or match if any rule
1049 of the policy matches the given policy.
1050 For each policy element that is to be described, one can use one or
1052 more of the following options. When --strict is in effect, at least one
1053 must be used per element.
1054 [!] --reqid id
1056 Matches the reqid of the policy rule. The reqid can be specified
1057 with setkey(8) using unique:id as level.
1058 [!] --spi spi
1060 Matches the SPI of the SA.
1061 [!] --proto {ah|esp|ipcomp}
1063 Matches the encapsulation protocol.
1064 [!] --mode {tunnel|transport}
1066 Matches the encapsulation mode.
1067 [!] --tunnel-src addr[/mask]
1069 Matches the source end-point address of a tunnel mode SA. Only
1070 valid with --mode tunnel.
1071 [!] --tunnel-dst addr[/mask]
1073 Matches the destination end-point address of a tunnel mode SA.
1074 Only valid with --mode tunnel.
1075 --next Start the next element in the policy specification. Can only be
1077 used with --strict.
1078 quota
1080 Implements network quotas by decrementing a byte counter with each
1081 packet. The condition matches until the byte counter reaches zero. Be‐
1082 havior is reversed with negation (i.e. the condition does not match un‐
1083 til the byte counter reaches zero).
1084 [!] --quota bytes
1086 The quota in bytes.
1087 rateest
1089 The rate estimator can match on estimated rates as collected by the RA‐
1090 TEEST target. It supports matching on absolute bps/pps values, compar‐
1091 ing two rate estimators and matching on the difference between two rate
1092 estimators.
1093 For a better understanding of the available options, these are all pos‐
1095 sible combinations:
1096 • rateest operator rateest-bps
1098 • rateest operator rateest-pps
1100 • (rateest minus rateest-bps1) operator rateest-bps2
1102 • (rateest minus rateest-pps1) operator rateest-pps2
1104 • rateest1 operator rateest2 rateest-bps(without rate!)
1106 • rateest1 operator rateest2 rateest-pps(without rate!)
1108 • (rateest1 minus rateest-bps1) operator (rateest2 minus rateest-
1110 bps2)
1111 • (rateest1 minus rateest-pps1) operator (rateest2 minus rateest-
1113 pps2)
1114 --rateest-delta
1116 For each estimator (either absolute or relative mode), calculate
1117 the difference between the estimator-determined flow rate and the
1118 static value chosen with the BPS/PPS options. If the flow rate is
1119 higher than the specified BPS/PPS, 0 will be used instead of a neg‐
1120 ative value. In other words, "max(0, rateest#_rate - rateest#_bps)"
1121 is used.
1122 [!] --rateest-lt
1124 Match if rate is less than given rate/estimator.
1125 [!] --rateest-gt
1127 Match if rate is greater than given rate/estimator.
1128 [!] --rateest-eq
1130 Match if rate is equal to given rate/estimator.
1131 In the so-called "absolute mode", only one rate estimator is used and
1133 compared against a static value, while in "relative mode", two rate es‐
1134 timators are compared against another.
1135 --rateest name
1137 Name of the one rate estimator for absolute mode.
1138 --rateest1 name
1140 --rateest2 name
1142 The names of the two rate estimators for relative mode.
1143 --rateest-bps [value]
1145 --rateest-pps [value]
1147 --rateest-bps1 [value]
1149 --rateest-bps2 [value]
1151 --rateest-pps1 [value]
1153 --rateest-pps2 [value]
1155 Compare the estimator(s) by bytes or packets per second, and
1156 compare against the chosen value. See the above bullet list for
1157 which option is to be used in which case. A unit suffix may be
1158 used - available ones are: bit, [kmgt]bit, [KMGT]ibit, Bps,
1159 [KMGT]Bps, [KMGT]iBps.
1160 Example: This is what can be used to route outgoing data connections
1162 from an FTP server over two lines based on the available bandwidth at
1163 the time the data connection was started:
1164 # Estimate outgoing rates
1166 iptables -t mangle -A POSTROUTING -o eth0 -j RATEEST --rateest-name
1168 eth0 --rateest-interval 250ms --rateest-ewma 0.5s
1169 iptables -t mangle -A POSTROUTING -o ppp0 -j RATEEST --rateest-name
1171 ppp0 --rateest-interval 250ms --rateest-ewma 0.5s
1172 # Mark based on available bandwidth
1174 iptables -t mangle -A balance -m conntrack --ctstate NEW -m helper
1176 --helper ftp -m rateest --rateest-delta --rateest1 eth0 --rateest-bps1
1177 2.5mbit --rateest-gt --rateest2 ppp0 --rateest-bps2 2mbit -j CONNMARK
1178 --set-mark 1
1179 iptables -t mangle -A balance -m conntrack --ctstate NEW -m helper
1181 --helper ftp -m rateest --rateest-delta --rateest1 ppp0 --rateest-bps1
1182 2mbit --rateest-gt --rateest2 eth0 --rateest-bps2 2.5mbit -j CONNMARK
1183 --set-mark 2
1184 iptables -t mangle -A balance -j CONNMARK --restore-mark
1186 realm (IPv4-specific)
1188 This matches the routing realm. Routing realms are used in complex
1189 routing setups involving dynamic routing protocols like BGP.
1190 [!] --realm value[/mask]
1192 Matches a given realm number (and optionally mask). If not a
1193 number, value can be a named realm from /etc/iproute2/rt_realms
1194 (mask can not be used in that case). Both value and mask are
1195 four byte unsigned integers and may be specified in decimal, hex
1196 (by prefixing with "0x") or octal (if a leading zero is given).
1197 recent
1199 Allows you to dynamically create a list of IP addresses and then match
1200 against that list in a few different ways.
1201 For example, you can create a "badguy" list out of people attempting to
1203 connect to port 139 on your firewall and then DROP all future packets
1204 from them without considering them.
1205 --set, --rcheck, --update and --remove are mutually exclusive.
1207 --name name
1209 Specify the list to use for the commands. If no name is given
1210 then DEFAULT will be used.
1211 [!] --set
1213 This will add the source address of the packet to the list. If
1214 the source address is already in the list, this will update the
1215 existing entry. This will always return success (or failure if !
1216 is passed in).
1217 --rsource
1219 Match/save the source address of each packet in the recent list
1220 table. This is the default.
1221 --rdest
1223 Match/save the destination address of each packet in the recent
1224 list table.
1225 --mask netmask
1227 Netmask that will be applied to this recent list.
1228 [!] --rcheck
1230 Check if the source address of the packet is currently in the
1231 list.
1232 [!] --update
1234 Like --rcheck, except it will update the "last seen" timestamp
1235 if it matches.
1236 [!] --remove
1238 Check if the source address of the packet is currently in the
1239 list and if so that address will be removed from the list and
1240 the rule will return true. If the address is not found, false is
1241 returned.
1242 --seconds seconds
1244 This option must be used in conjunction with one of --rcheck or
1245 --update. When used, this will narrow the match to only happen
1246 when the address is in the list and was seen within the last
1247 given number of seconds.
1248 --reap This option can only be used in conjunction with --seconds.
1250 When used, this will cause entries older than the last given
1251 number of seconds to be purged.
1252 --hitcount hits
1254 This option must be used in conjunction with one of --rcheck or
1255 --update. When used, this will narrow the match to only happen
1256 when the address is in the list and packets had been received
1257 greater than or equal to the given value. This option may be
1258 used along with --seconds to create an even narrower match re‐
1259 quiring a certain number of hits within a specific time frame.
1260 The maximum value for the hitcount parameter is given by the
1261 "ip_pkt_list_tot" parameter of the xt_recent kernel module. Ex‐
1262 ceeding this value on the command line will cause the rule to be
1263 rejected.
1264 --rttl This option may only be used in conjunction with one of --rcheck
1266 or --update. When used, this will narrow the match to only hap‐
1267 pen when the address is in the list and the TTL of the current
1268 packet matches that of the packet which hit the --set rule. This
1269 may be useful if you have problems with people faking their
1270 source address in order to DoS you via this module by disallow‐
1271 ing others access to your site by sending bogus packets to you.
1272 Examples:
1274 iptables -A FORWARD -m recent --name badguy --rcheck --seconds
1276 60 -j DROP
1277 iptables -A FORWARD -p tcp -i eth0 --dport 139 -m recent --name
1279 badguy --set -j DROP
1280 /proc/net/xt_recent/* are the current lists of addresses and informa‐
1282 tion about each entry of each list.
1283 Each file in /proc/net/xt_recent/ can be read from to see the current
1285 list or written two using the following commands to modify the list:
1286 echo +addr >/proc/net/xt_recent/DEFAULT
1288 to add addr to the DEFAULT list
1289 echo -addr >/proc/net/xt_recent/DEFAULT
1291 to remove addr from the DEFAULT list
1292 echo / >/proc/net/xt_recent/DEFAULT
1294 to flush the DEFAULT list (remove all entries).
1295 The module itself accepts parameters, defaults shown:
1297 ip_list_tot=100
1299 Number of addresses remembered per table.
1300 ip_pkt_list_tot=20
1302 Number of packets per address remembered.
1303 ip_list_hash_size=0
1305 Hash table size. 0 means to calculate it based on ip_list_tot,
1306 default: 512.
1307 ip_list_perms=0644
1309 Permissions for /proc/net/xt_recent/* files.
1310 ip_list_uid=0
1312 Numerical UID for ownership of /proc/net/xt_recent/* files.
1313 ip_list_gid=0
1315 Numerical GID for ownership of /proc/net/xt_recent/* files.
1316 rpfilter
1318 Performs a reverse path filter test on a packet. If a reply to the
1319 packet would be sent via the same interface that the packet arrived on,
1320 the packet will match. Note that, unlike the in-kernel rp_filter,
1321 packets protected by IPSec are not treated specially. Combine this
1322 match with the policy match if you want this. Also, packets arriving
1323 via the loopback interface are always permitted. This match can only
1324 be used in the PREROUTING chain of the raw or mangle table.
1325 --loose
1327 Used to specify that the reverse path filter test should match
1328 even if the selected output device is not the expected one.
1329 --validmark
1331 Also use the packets' nfmark value when performing the reverse
1332 path route lookup.
1333 --accept-local
1335 This will permit packets arriving from the network with a source
1336 address that is also assigned to the local machine.
1337 --invert
1339 This will invert the sense of the match. Instead of matching
1340 packets that passed the reverse path filter test, match those
1341 that have failed it.
1342 Example to log and drop packets failing the reverse path filter test:
1344 iptables -t raw -N RPFILTER
1346 iptables -t raw -A RPFILTER -m rpfilter -j RETURN
1348 iptables -t raw -A RPFILTER -m limit --limit 10/minute -j NFLOG
1350 --nflog-prefix "rpfilter drop"
1351 iptables -t raw -A RPFILTER -j DROP
1353 iptables -t raw -A PREROUTING -j RPFILTER
1355 Example to drop failed packets, without logging:
1357 iptables -t raw -A RPFILTER -m rpfilter --invert -j DROP
1359 rt (IPv6-specific)
1361 Match on IPv6 routing header
1362 [!] --rt-type type
1364 Match the type (numeric).
1365 [!] --rt-segsleft num[:num]
1367 Match the `segments left' field (range).
1368 [!] --rt-len length
1370 Match the length of this header.
1371 --rt-0-res
1373 Match the reserved field, too (type=0)
1374 --rt-0-addrs addr[,addr...]
1376 Match type=0 addresses (list).
1377 --rt-0-not-strict
1379 List of type=0 addresses is not a strict list.
1380 sctp
1382 This module matches Stream Control Transmission Protocol headers.
1383 [!] --source-port,--sport port[:port]
1385 [!] --destination-port,--dport port[:port]
1387 [!] --chunk-types {all|any|only} chunktype[:flags] [...]
1389 The flag letter in upper case indicates that the flag is to
1390 match if set, in the lower case indicates to match if unset.
1391 Match types:
1393 all Match if all given chunk types are present and flags match.
1395 any Match if any of the given chunk types is present with given
1397 flags.
1398 only Match if only the given chunk types are present with given flags
1400 and none are missing.
1401 Chunk types: DATA INIT INIT_ACK SACK HEARTBEAT HEARTBEAT_ACK
1403 ABORT SHUTDOWN SHUTDOWN_ACK ERROR COOKIE_ECHO COOKIE_ACK
1404 ECN_ECNE ECN_CWR SHUTDOWN_COMPLETE I_DATA RE_CONFIG PAD ASCONF
1405 ASCONF_ACK FORWARD_TSN I_FORWARD_TSN
1406 chunk type available flags
1408 DATA I U B E i u b e
1409 I_DATA I U B E i u b e
1410 ABORT T t
1411 SHUTDOWN_COMPLETE T t
1412 (lowercase means flag should be "off", uppercase means "on")
1414 Examples:
1416 iptables -A INPUT -p sctp --dport 80 -j DROP
1418 iptables -A INPUT -p sctp --chunk-types any DATA,INIT -j DROP
1420 iptables -A INPUT -p sctp --chunk-types any DATA:Be -j ACCEPT
1422 set
1424 This module matches IP sets which can be defined by ipset(8).
1425 [!] --match-set setname flag[,flag]...
1427 where flags are the comma separated list of src and/or dst spec‐
1428 ifications and there can be no more than six of them. Hence the
1429 command
1430 iptables -A FORWARD -m set --match-set test src,dst
1432 will match packets, for which (if the set type is ipportmap) the
1434 source address and destination port pair can be found in the
1435 specified set. If the set type of the specified set is single
1436 dimension (for example ipmap), then the command will match pack‐
1437 ets for which the source address can be found in the specified
1438 set.
1439 --return-nomatch
1441 If the --return-nomatch option is specified and the set type
1442 supports the nomatch flag, then the matching is reversed: a
1443 match with an element flagged with nomatch returns true, while a
1444 match with a plain element returns false.
1445 ! --update-counters
1447 If the --update-counters flag is negated, then the packet and
1448 byte counters of the matching element in the set won't be up‐
1449 dated. Default the packet and byte counters are updated.
1450 ! --update-subcounters
1452 If the --update-subcounters flag is negated, then the packet and
1453 byte counters of the matching element in the member set of a
1454 list type of set won't be updated. Default the packet and byte
1455 counters are updated.
1456 [!] --packets-eq value
1458 If the packet is matched an element in the set, match only if
1459 the packet counter of the element matches the given value too.
1460 --packets-lt value
1462 If the packet is matched an element in the set, match only if
1463 the packet counter of the element is less than the given value
1464 as well.
1465 --packets-gt value
1467 If the packet is matched an element in the set, match only if
1468 the packet counter of the element is greater than the given
1469 value as well.
1470 [!] --bytes-eq value
1472 If the packet is matched an element in the set, match only if
1473 the byte counter of the element matches the given value too.
1474 --bytes-lt value
1476 If the packet is matched an element in the set, match only if
1477 the byte counter of the element is less than the given value as
1478 well.
1479 --bytes-gt value
1481 If the packet is matched an element in the set, match only if
1482 the byte counter of the element is greater than the given value
1483 as well.
1484 The packet and byte counters related options and flags are ignored when
1486 the set was defined without counter support.
1487 The option --match-set can be replaced by --set if that does not clash
1489 with an option of other extensions.
1490 Use of -m set requires that ipset kernel support is provided, which,
1492 for standard kernels, is the case since Linux 2.6.39.
1493 socket
1495 This matches if an open TCP/UDP socket can be found by doing a socket
1496 lookup on the packet. It matches if there is an established or non-zero
1497 bound listening socket (possibly with a non-local address). The lookup
1498 is performed using the packet tuple of TCP/UDP packets, or the original
1499 TCP/UDP header embedded in an ICMP/ICPMv6 error packet.
1500 --transparent
1502 Ignore non-transparent sockets.
1503 --nowildcard
1505 Do not ignore sockets bound to 'any' address. The socket match
1506 won't accept zero-bound listeners by default, since then local
1507 services could intercept traffic that would otherwise be for‐
1508 warded. This option therefore has security implications when
1509 used to match traffic being forwarded to redirect such packets
1510 to local machine with policy routing. When using the socket
1511 match to implement fully transparent proxies bound to non-local
1512 addresses it is recommended to use the --transparent option in‐
1513 stead.
1514 Example (assuming packets with mark 1 are delivered locally):
1516 -t mangle -A PREROUTING -m socket --transparent -j MARK
1518 --set-mark 1
1519 --restore-skmark
1521 Set the packet mark to the matching socket's mark. Can be com‐
1522 bined with the --transparent and --nowildcard options to re‐
1523 strict the sockets to be matched when restoring the packet mark.
1524 Example: An application opens 2 transparent (IP_TRANSPARENT) sockets
1526 and sets a mark on them with SO_MARK socket option. We can filter
1527 matching packets:
1528 -t mangle -I PREROUTING -m socket --transparent --restore-skmark
1530 -j action
1531 -t mangle -A action -m mark --mark 10 -j action2
1533 -t mangle -A action -m mark --mark 11 -j action3
1535 state
1537 The "state" extension is a subset of the "conntrack" module. "state"
1538 allows access to the connection tracking state for this packet.
1539 [!] --state state
1541 Where state is a comma separated list of the connection states
1542 to match. Only a subset of the states unterstood by "conntrack"
1543 are recognized: INVALID, ESTABLISHED, NEW, RELATED or UNTRACKED.
1544 For their description, see the "conntrack" heading in this man‐
1545 page.
1546 statistic
1548 This module matches packets based on some statistic condition. It sup‐
1549 ports two distinct modes settable with the --mode option.
1550 Supported options:
1552 --mode mode
1554 Set the matching mode of the matching rule, supported modes are
1555 random and nth.
1556 [!] --probability p
1558 Set the probability for a packet to be randomly matched. It only
1559 works with the random mode. p must be within 0.0 and 1.0. The
1560 supported granularity is in 1/2147483648th increments.
1561 [!] --every n
1563 Match one packet every nth packet. It works only with the nth
1564 mode (see also the --packet option).
1565 --packet p
1567 Set the initial counter value (0 <= p <= n-1, default 0) for the
1568 nth mode.
1569 string
1571 This module matches a given string by using some pattern matching
1572 strategy. It requires a linux kernel >= 2.6.14.
1573 --algo {bm|kmp}
1575 Select the pattern matching strategy. (bm = Boyer-Moore, kmp =
1576 Knuth-Pratt-Morris)
1577 --from offset
1579 Set the offset from which it starts looking for any matching. If
1580 not passed, default is 0.
1581 --to offset
1583 Set the offset up to which should be scanned. That is, byte off‐
1584 set-1 (counting from 0) is the last one that is scanned. If not
1585 passed, default is the packet size.
1586 [!] --string pattern
1588 Matches the given pattern.
1589 [!] --hex-string pattern
1591 Matches the given pattern in hex notation.
1592 --icase
1594 Ignore case when searching.
1595 Examples:
1597 # The string pattern can be used for simple text characters.
1599 iptables -A INPUT -p tcp --dport 80 -m string --algo bm --string
1600 'GET /index.html' -j LOG
1601 # The hex string pattern can be used for non-printable charac‐
1603 ters, like |0D 0A| or |0D0A|.
1604 iptables -p udp --dport 53 -m string --algo bm --from 40 --to 57
1605 --hex-string '|03|www|09|netfilter|03|org|00|'
1606 tcp
1608 These extensions can be used if `--protocol tcp' is specified. It pro‐
1609 vides the following options:
1610 [!] --source-port,--sport port[:port]
1612 Source port or port range specification. This can either be a
1613 service name or a port number. An inclusive range can also be
1614 specified, using the format first:last. If the first port is
1615 omitted, "0" is assumed; if the last is omitted, "65535" is as‐
1616 sumed. The flag --sport is a convenient alias for this option.
1617 [!] --destination-port,--dport port[:port]
1619 Destination port or port range specification. The flag --dport
1620 is a convenient alias for this option.
1621 [!] --tcp-flags mask comp
1623 Match when the TCP flags are as specified. The first argument
1624 mask is the flags which we should examine, written as a comma-
1625 separated list, and the second argument comp is a comma-sepa‐
1626 rated list of flags which must be set. Flags are: SYN ACK FIN
1627 RST URG PSH ALL NONE. Hence the command
1628 iptables -A FORWARD -p tcp --tcp-flags SYN,ACK,FIN,RST SYN
1629 will only match packets with the SYN flag set, and the ACK, FIN
1630 and RST flags unset.
1631 [!] --syn
1633 Only match TCP packets with the SYN bit set and the ACK,RST and
1634 FIN bits cleared. Such packets are used to request TCP connec‐
1635 tion initiation; for example, blocking such packets coming in an
1636 interface will prevent incoming TCP connections, but outgoing
1637 TCP connections will be unaffected. It is equivalent to
1638 --tcp-flags SYN,RST,ACK,FIN SYN. If the "!" flag precedes the
1639 "--syn", the sense of the option is inverted.
1640 [!] --tcp-option number
1642 Match if TCP option set.
1643 tcpmss
1645 This matches the TCP MSS (maximum segment size) field of the TCP
1646 header. You can only use this on TCP SYN or SYN/ACK packets, since the
1647 MSS is only negotiated during the TCP handshake at connection startup
1648 time.
1649 [!] --mss value[:value]
1651 Match a given TCP MSS value or range. If a range is given, the
1652 second value must be greater than or equal to the first value.
1653 time
1655 This matches if the packet arrival time/date is within a given range.
1656 All options are optional, but are ANDed when specified. All times are
1657 interpreted as UTC by default.
1658 --datestart YYYY[-MM[-DD[Thh[:mm[:ss]]]]]
1660 --datestop YYYY[-MM[-DD[Thh[:mm[:ss]]]]]
1662 Only match during the given time, which must be in ISO 8601 "T"
1663 notation. The possible time range is 1970-01-01T00:00:00 to
1664 2038-01-19T04:17:07.
1665 If --datestart or --datestop are not specified, it will default
1667 to 1970-01-01 and 2038-01-19, respectively.
1668 --timestart hh:mm[:ss]
1670 --timestop hh:mm[:ss]
1672 Only match during the given daytime. The possible time range is
1673 00:00:00 to 23:59:59. Leading zeroes are allowed (e.g. "06:03")
1674 and correctly interpreted as base-10.
1675 [!] --monthdays day[,day...]
1677 Only match on the given days of the month. Possible values are 1
1678 to 31. Note that specifying 31 will of course not match on
1679 months which do not have a 31st day; the same goes for 28- or
1680 29-day February.
1681 [!] --weekdays day[,day...]
1683 Only match on the given weekdays. Possible values are Mon, Tue,
1684 Wed, Thu, Fri, Sat, Sun, or values from 1 to 7, respectively.
1685 You may also use two-character variants (Mo, Tu, etc.).
1686 --contiguous
1688 When --timestop is smaller than --timestart value, match this as
1689 a single time period instead distinct intervals. See EXAMPLES.
1690 --kerneltz
1692 Use the kernel timezone instead of UTC to determine whether a
1693 packet meets the time regulations.
1694 About kernel timezones: Linux keeps the system time in UTC, and always
1696 does so. On boot, system time is initialized from a referential time
1697 source. Where this time source has no timezone information, such as the
1698 x86 CMOS RTC, UTC will be assumed. If the time source is however not in
1699 UTC, userspace should provide the correct system time and timezone to
1700 the kernel once it has the information.
1701 Local time is a feature on top of the (timezone independent) system
1703 time. Each process has its own idea of local time, specified via the TZ
1704 environment variable. The kernel also has its own timezone offset vari‐
1705 able. The TZ userspace environment variable specifies how the UTC-based
1706 system time is displayed, e.g. when you run date(1), or what you see on
1707 your desktop clock. The TZ string may resolve to different offsets at
1708 different dates, which is what enables the automatic time-jumping in
1709 userspace. when DST changes. The kernel's timezone offset variable is
1710 used when it has to convert between non-UTC sources, such as FAT
1711 filesystems, to UTC (since the latter is what the rest of the system
1712 uses).
1713 The caveat with the kernel timezone is that Linux distributions may ig‐
1715 nore to set the kernel timezone, and instead only set the system time.
1716 Even if a particular distribution does set the timezone at boot, it is
1717 usually does not keep the kernel timezone offset - which is what
1718 changes on DST - up to date. ntpd will not touch the kernel timezone,
1719 so running it will not resolve the issue. As such, one may encounter a
1720 timezone that is always +0000, or one that is wrong half of the time of
1721 the year. As such, using --kerneltz is highly discouraged.
1722 EXAMPLES. To match on weekends, use:
1724 -m time --weekdays Sa,Su
1726 Or, to match (once) on a national holiday block:
1728 -m time --datestart 2007-12-24 --datestop 2007-12-27
1730 Since the stop time is actually inclusive, you would need the following
1732 stop time to not match the first second of the new day:
1733 -m time --datestart 2007-01-01T17:00 --datestop
1735 2007-01-01T23:59:59
1736 During lunch hour:
1738 -m time --timestart 12:30 --timestop 13:30
1740 The fourth Friday in the month:
1742 -m time --weekdays Fr --monthdays 22,23,24,25,26,27,28
1744 (Note that this exploits a certain mathematical property. It is not
1746 possible to say "fourth Thursday OR fourth Friday" in one rule. It is
1747 possible with multiple rules, though.)
1748 Matching across days might not do what is expected. For instance,
1750 -m time --weekdays Mo --timestart 23:00 --timestop 01:00 Will
1752 match Monday, for one hour from midnight to 1 a.m., and then
1753 again for another hour from 23:00 onwards. If this is unwanted,
1754 e.g. if you would like 'match for two hours from Montay 23:00
1755 onwards' you need to also specify the --contiguous option in the
1756 example above.
1757 tos
1759 This module matches the 8-bit Type of Service field in the IPv4 header
1760 (i.e. including the "Precedence" bits) or the (also 8-bit) Priority
1761 field in the IPv6 header.
1762 [!] --tos value[/mask]
1764 Matches packets with the given TOS mark value. If a mask is
1765 specified, it is logically ANDed with the TOS mark before the
1766 comparison.
1767 [!] --tos symbol
1769 You can specify a symbolic name when using the tos match for
1770 IPv4. The list of recognized TOS names can be obtained by call‐
1771 ing iptables with -m tos -h. Note that this implies a mask of
1772 0x3F, i.e. all but the ECN bits.
1773 ttl (IPv4-specific)
1775 This module matches the time to live field in the IP header.
1776 [!] --ttl-eq ttl
1778 Matches the given TTL value.
1779 --ttl-gt ttl
1781 Matches if TTL is greater than the given TTL value.
1782 --ttl-lt ttl
1784 Matches if TTL is less than the given TTL value.
1785 u32
1787 U32 tests whether quantities of up to 4 bytes extracted from a packet
1788 have specified values. The specification of what to extract is general
1789 enough to find data at given offsets from tcp headers or payloads.
1790 [!] --u32 tests
1792 The argument amounts to a program in a small language described
1793 below.
1794 tests := location "=" value | tests "&&" location "=" value
1796 value := range | value "," range
1798 range := number | number ":" number
1800 a single number, n, is interpreted the same as n:n. n:m is interpreted
1802 as the range of numbers >=n and <=m.
1803 location := number | location operator number
1805 operator := "&" | "<<" | ">>" | "@"
1807 The operators &, <<, >> and && mean the same as in C. The = is really
1809 a set membership operator and the value syntax describes a set. The @
1810 operator is what allows moving to the next header and is described fur‐
1811 ther below.
1812 There are currently some artificial implementation limits on the size
1814 of the tests:
1815 * no more than 10 of "=" (and 9 "&&"s) in the u32 argument
1817 * no more than 10 ranges (and 9 commas) per value
1819 * no more than 10 numbers (and 9 operators) per location
1821 To describe the meaning of location, imagine the following machine that
1823 interprets it. There are three registers:
1824 A is of type char *, initially the address of the IP header
1826 B and C are unsigned 32 bit integers, initially zero
1828 The instructions are:
1830 number B = number;
1832 C = (*(A+B)<<24) + (*(A+B+1)<<16) + (*(A+B+2)<<8) + *(A+B+3)
1834 &number
1836 C = C & number
1837 << number
1839 C = C << number
1840 >> number
1842 C = C >> number
1843 @number
1845 A = A + C; then do the instruction number
1846 Any access of memory outside [skb->data,skb->end] causes the match to
1848 fail. Otherwise the result of the computation is the final value of C.
1849 Whitespace is allowed but not required in the tests. However, the char‐
1851 acters that do occur there are likely to require shell quoting, so it
1852 is a good idea to enclose the arguments in quotes.
1853 Example:
1855 match IP packets with total length >= 256
1857 The IP header contains a total length field in bytes 2-3.
1859 --u32 "0 & 0xFFFF = 0x100:0xFFFF"
1861 read bytes 0-3
1863 AND that with 0xFFFF (giving bytes 2-3), and test whether that
1865 is in the range [0x100:0xFFFF]
1866 Example: (more realistic, hence more complicated)
1868 match ICMP packets with icmp type 0
1870 First test that it is an ICMP packet, true iff byte 9 (protocol)
1872 = 1
1873 --u32 "6 & 0xFF = 1 && ...
1875 read bytes 6-9, use & to throw away bytes 6-8 and compare the
1877 result to 1. Next test that it is not a fragment. (If so, it
1878 might be part of such a packet but we cannot always tell.) N.B.:
1879 This test is generally needed if you want to match anything be‐
1880 yond the IP header. The last 6 bits of byte 6 and all of byte 7
1881 are 0 iff this is a complete packet (not a fragment). Alterna‐
1882 tively, you can allow first fragments by only testing the last 5
1883 bits of byte 6.
1884 ... 4 & 0x3FFF = 0 && ...
1886 Last test: the first byte past the IP header (the type) is 0.
1888 This is where we have to use the @syntax. The length of the IP
1889 header (IHL) in 32 bit words is stored in the right half of byte
1890 0 of the IP header itself.
1891 ... 0 >> 22 & 0x3C @ 0 >> 24 = 0"
1893 The first 0 means read bytes 0-3, >>22 means shift that 22 bits
1895 to the right. Shifting 24 bits would give the first byte, so
1896 only 22 bits is four times that plus a few more bits. &3C then
1897 eliminates the two extra bits on the right and the first four
1898 bits of the first byte. For instance, if IHL=5, then the IP
1899 header is 20 (4 x 5) bytes long. In this case, bytes 0-1 are (in
1900 binary) xxxx0101 yyzzzzzz, >>22 gives the 10 bit value
1901 xxxx0101yy and &3C gives 010100. @ means to use this number as a
1902 new offset into the packet, and read four bytes starting from
1903 there. This is the first 4 bytes of the ICMP payload, of which
1904 byte 0 is the ICMP type. Therefore, we simply shift the value 24
1905 to the right to throw out all but the first byte and compare the
1906 result with 0.
1907 Example:
1909 TCP payload bytes 8-12 is any of 1, 2, 5 or 8
1911 First we test that the packet is a tcp packet (similar to ICMP).
1913 --u32 "6 & 0xFF = 6 && ...
1915 Next, test that it is not a fragment (same as above).
1917 ... 0 >> 22 & 0x3C @ 12 >> 26 & 0x3C @ 8 = 1,2,5,8"
1919 0>>22&3C as above computes the number of bytes in the IP header.
1921 @ makes this the new offset into the packet, which is the start
1922 of the TCP header. The length of the TCP header (again in 32 bit
1923 words) is the left half of byte 12 of the TCP header. The
1924 12>>26&3C computes this length in bytes (similar to the IP
1925 header before). "@" makes this the new offset, which is the
1926 start of the TCP payload. Finally, 8 reads bytes 8-12 of the
1927 payload and = checks whether the result is any of 1, 2, 5 or 8.
1928 udp
1930 These extensions can be used if `--protocol udp' is specified. It pro‐
1931 vides the following options:
1932 [!] --source-port,--sport port[:port]
1934 Source port or port range specification. See the description of
1935 the --source-port option of the TCP extension for details.
1936 [!] --destination-port,--dport port[:port]
1938 Destination port or port range specification. See the descrip‐
1939 tion of the --destination-port option of the TCP extension for
1940 details.
1941
1943 iptables can use extended target modules: the following are included in
1944 the standard distribution.
1945 AUDIT
1947 This target creates audit records for packets hitting the target. It
1948 can be used to record accepted, dropped, and rejected packets. See au‐
1949 ditd(8) for additional details.
1950 --type {accept|drop|reject}
1952 Set type of audit record. Starting with linux-4.12, this option
1953 has no effect on generated audit messages anymore. It is still
1954 accepted by iptables for compatibility reasons, but ignored.
1955 Example:
1957 iptables -N AUDIT_DROP
1959 iptables -A AUDIT_DROP -j AUDIT
1961 iptables -A AUDIT_DROP -j DROP
1963 CHECKSUM
1965 This target selectively works around broken/old applications. It can
1966 only be used in the mangle table.
1967 --checksum-fill
1969 Compute and fill in the checksum in a packet that lacks a check‐
1970 sum. This is particularly useful, if you need to work around
1971 old applications such as dhcp clients, that do not work well
1972 with checksum offloads, but don't want to disable checksum off‐
1973 load in your device.
1974 CLASSIFY
1976 This module allows you to set the skb->priority value (and thus clas‐
1977 sify the packet into a specific CBQ class).
1978 --set-class major:minor
1980 Set the major and minor class value. The values are always in‐
1981 terpreted as hexadecimal even if no 0x prefix is given.
1982 CLUSTERIP (IPv4-specific)
1984 This module allows you to configure a simple cluster of nodes that
1985 share a certain IP and MAC address without an explicit load balancer in
1986 front of them. Connections are statically distributed between the
1987 nodes in this cluster.
1988 Please note that CLUSTERIP target is considered deprecated in favour of
1990 cluster match which is more flexible and not limited to IPv4.
1991 --new Create a new ClusterIP. You always have to set this on the
1993 first rule for a given ClusterIP.
1994 --hashmode mode
1996 Specify the hashing mode. Has to be one of sourceip, sour‐
1997 ceip-sourceport, sourceip-sourceport-destport.
1998 --clustermac mac
2000 Specify the ClusterIP MAC address. Has to be a link-layer multi‐
2001 cast address
2002 --total-nodes num
2004 Number of total nodes within this cluster.
2005 --local-node num
2007 Local node number within this cluster.
2008 --hash-init rnd
2010 Specify the random seed used for hash initialization.
2011 CONNMARK
2013 This module sets the netfilter mark value associated with a connection.
2014 The mark is 32 bits wide.
2015 --set-xmark value[/mask]
2017 Zero out the bits given by mask and XOR value into the ctmark.
2018 --save-mark [--nfmask nfmask] [--ctmask ctmask]
2020 Copy the packet mark (nfmark) to the connection mark (ctmark)
2021 using the given masks. The new nfmark value is determined as
2022 follows:
2023 ctmark = (ctmark & ~ctmask) ^ (nfmark & nfmask)
2025 i.e. ctmask defines what bits to clear and nfmask what bits of
2027 the nfmark to XOR into the ctmark. ctmask and nfmask default to
2028 0xFFFFFFFF.
2029 --restore-mark [--nfmask nfmask] [--ctmask ctmask]
2031 Copy the connection mark (ctmark) to the packet mark (nfmark)
2032 using the given masks. The new ctmark value is determined as
2033 follows:
2034 nfmark = (nfmark & ~nfmask) ^ (ctmark & ctmask);
2036 i.e. nfmask defines what bits to clear and ctmask what bits of
2038 the ctmark to XOR into the nfmark. ctmask and nfmask default to
2039 0xFFFFFFFF.
2040 --restore-mark is only valid in the mangle table.
2042 The following mnemonics are available for --set-xmark:
2044 --and-mark bits
2046 Binary AND the ctmark with bits. (Mnemonic for --set-xmark 0/in‐
2047 vbits, where invbits is the binary negation of bits.)
2048 --or-mark bits
2050 Binary OR the ctmark with bits. (Mnemonic for --set-xmark
2051 bits/bits.)
2052 --xor-mark bits
2054 Binary XOR the ctmark with bits. (Mnemonic for --set-xmark
2055 bits/0.)
2056 --set-mark value[/mask]
2058 Set the connection mark. If a mask is specified then only those
2059 bits set in the mask are modified.
2060 --save-mark [--mask mask]
2062 Copy the nfmark to the ctmark. If a mask is specified, only
2063 those bits are copied.
2064 --restore-mark [--mask mask]
2066 Copy the ctmark to the nfmark. If a mask is specified, only
2067 those bits are copied. This is only valid in the mangle table.
2068 CONNSECMARK
2070 This module copies security markings from packets to connections (if
2071 unlabeled), and from connections back to packets (also only if unla‐
2072 beled). Typically used in conjunction with SECMARK, it is valid in the
2073 security table (for backwards compatibility with older kernels, it is
2074 also valid in the mangle table).
2075 --save If the packet has a security marking, copy it to the connection
2077 if the connection is not marked.
2078 --restore
2080 If the packet does not have a security marking, and the connec‐
2081 tion does, copy the security marking from the connection to the
2082 packet.
2083 CT
2086 The CT target sets parameters for a packet or its associated connec‐
2087 tion. The target attaches a "template" connection tracking entry to the
2088 packet, which is then used by the conntrack core when initializing a
2089 new ct entry. This target is thus only valid in the "raw" table.
2090 --notrack
2092 Disables connection tracking for this packet.
2093 --helper name
2095 Use the helper identified by name for the connection. This is
2096 more flexible than loading the conntrack helper modules with
2097 preset ports.
2098 --ctevents event[,...]
2100 Only generate the specified conntrack events for this connec‐
2101 tion. Possible event types are: new, related, destroy, reply,
2102 assured, protoinfo, helper, mark (this refers to the ctmark, not
2103 nfmark), natseqinfo, secmark (ctsecmark).
2104 --expevents event[,...]
2106 Only generate the specified expectation events for this connec‐
2107 tion. Possible event types are: new.
2108 --zone-orig {id|mark}
2110 For traffic coming from ORIGINAL direction, assign this packet
2111 to zone id and only have lookups done in that zone. If mark is
2112 used instead of id, the zone is derived from the packet nfmark.
2113 --zone-reply {id|mark}
2115 For traffic coming from REPLY direction, assign this packet to
2116 zone id and only have lookups done in that zone. If mark is used
2117 instead of id, the zone is derived from the packet nfmark.
2118 --zone {id|mark}
2120 Assign this packet to zone id and only have lookups done in that
2121 zone. If mark is used instead of id, the zone is derived from
2122 the packet nfmark. By default, packets have zone 0. This option
2123 applies to both directions.
2124 --timeout name
2126 Use the timeout policy identified by name for the connection.
2127 This is provides more flexible timeout policy definition than
2128 global timeout values available at /proc/sys/net/netfil‐
2129 ter/nf_conntrack_*_timeout_*.
2130 DNAT
2132 This target is only valid in the nat table, in the PREROUTING and OUT‐
2133 PUT chains, and user-defined chains which are only called from those
2134 chains. It specifies that the destination address of the packet should
2135 be modified (and all future packets in this connection will also be
2136 mangled), and rules should cease being examined. It takes the follow‐
2137 ing options:
2138 --to-destination [ipaddr[-ipaddr]][:port[-port[/baseport]]]
2140 which can specify a single new destination IP address, an inclu‐
2141 sive range of IP addresses. Optionally a port range, if the rule
2142 also specifies one of the following protocols: tcp, udp, dccp or
2143 sctp. If no port range is specified, then the destination port
2144 will never be modified. If no IP address is specified then only
2145 the destination port will be modified. If baseport is given,
2146 the difference of the original destination port and its value is
2147 used as offset into the mapping port range. This allows to cre‐
2148 ate shifted portmap ranges and is available since kernel version
2149 4.18. For a single port or baseport, a service name as listed
2150 in /etc/services may be used.
2151 --random
2153 Randomize source port mapping (kernel >= 2.6.22).
2154 --persistent
2156 Gives a client the same source-/destination-address for each
2157 connection. This supersedes the SAME target. Support for per‐
2158 sistent mappings is available from 2.6.29-rc2.
2159 IPv6 support available since Linux kernels >= 3.7.
2161 DNPT (IPv6-specific)
2163 Provides stateless destination IPv6-to-IPv6 Network Prefix Translation
2164 (as described by RFC 6296).
2165 You have to use this target in the mangle table, not in the nat table.
2167 It takes the following options:
2168 --src-pfx [prefix/length]
2170 Set source prefix that you want to translate and length
2171 --dst-pfx [prefix/length]
2173 Set destination prefix that you want to use in the translation
2174 and length
2175 You have to use the SNPT target to undo the translation. Example:
2177 ip6tables -t mangle -I POSTROUTING -s fd00::/64 -o vboxnet0 -j
2179 SNPT --src-pfx fd00::/64 --dst-pfx 2001:e20:2000:40f::/64
2180 ip6tables -t mangle -I PREROUTING -i wlan0 -d
2182 2001:e20:2000:40f::/64 -j DNPT --src-pfx 2001:e20:2000:40f::/64
2183 --dst-pfx fd00::/64
2184 You may need to enable IPv6 neighbor proxy:
2186 sysctl -w net.ipv6.conf.all.proxy_ndp=1
2188 You also have to use the NOTRACK target to disable connection tracking
2190 for translated flows.
2191 DSCP
2193 This target alters the value of the DSCP bits within the TOS header of
2194 the IPv4 packet. As this manipulates a packet, it can only be used in
2195 the mangle table.
2196 --set-dscp value
2198 Set the DSCP field to a numerical value (can be decimal or hex)
2199 --set-dscp-class class
2201 Set the DSCP field to a DiffServ class.
2202 ECN (IPv4-specific)
2204 This target selectively works around known ECN blackholes. It can only
2205 be used in the mangle table.
2206 --ecn-tcp-remove
2208 Remove all ECN bits from the TCP header. Of course, it can only
2209 be used in conjunction with -p tcp.
2210 HL (IPv6-specific)
2212 This is used to modify the Hop Limit field in IPv6 header. The Hop
2213 Limit field is similar to what is known as TTL value in IPv4. Setting
2214 or incrementing the Hop Limit field can potentially be very dangerous,
2215 so it should be avoided at any cost. This target is only valid in man‐
2216 gle table.
2217 Don't ever set or increment the value on packets that leave your local
2219 network!
2220 --hl-set value
2222 Set the Hop Limit to `value'.
2223 --hl-dec value
2225 Decrement the Hop Limit `value' times.
2226 --hl-inc value
2228 Increment the Hop Limit `value' times.
2229 HMARK
2231 Like MARK, i.e. set the fwmark, but the mark is calculated from hashing
2232 packet selector at choice. You have also to specify the mark range and,
2233 optionally, the offset to start from. ICMP error messages are inspected
2234 and used to calculate the hashing.
2235 Existing options are:
2237 --hmark-tuple tuple
2239 Possible tuple members are: src meaning source address (IPv4,
2240 IPv6 address), dst meaning destination address (IPv4, IPv6 ad‐
2241 dress), sport meaning source port (TCP, UDP, UDPlite, SCTP,
2242 DCCP), dport meaning destination port (TCP, UDP, UDPlite, SCTP,
2243 DCCP), spi meaning Security Parameter Index (AH, ESP), and ct
2244 meaning the usage of the conntrack tuple instead of the packet
2245 selectors.
2246 --hmark-mod value (must be > 0)
2248 Modulus for hash calculation (to limit the range of possible
2249 marks)
2250 --hmark-offset value
2252 Offset to start marks from.
2253 For advanced usage, instead of using --hmark-tuple, you can specify
2255 custom
2256 prefixes and masks:
2257 --hmark-src-prefix cidr
2259 The source address mask in CIDR notation.
2260 --hmark-dst-prefix cidr
2262 The destination address mask in CIDR notation.
2263 --hmark-sport-mask value
2265 A 16 bit source port mask in hexadecimal.
2266 --hmark-dport-mask value
2268 A 16 bit destination port mask in hexadecimal.
2269 --hmark-spi-mask value
2271 A 32 bit field with spi mask.
2272 --hmark-proto-mask value
2274 An 8 bit field with layer 4 protocol number.
2275 --hmark-rnd value
2277 A 32 bit random custom value to feed hash calculation.
2278 Examples:
2280 iptables -t mangle -A PREROUTING -m conntrack --ctstate NEW
2282 -j HMARK --hmark-tuple ct,src,dst,proto --hmark-offset 10000
2283 --hmark-mod 10 --hmark-rnd 0xfeedcafe
2284 iptables -t mangle -A PREROUTING -j HMARK --hmark-offset 10000 --hmark-
2286 tuple src,dst,proto --hmark-mod 10 --hmark-rnd 0xdeafbeef
2287 IDLETIMER
2289 This target can be used to identify when interfaces have been idle for
2290 a certain period of time. Timers are identified by labels and are cre‐
2291 ated when a rule is set with a new label. The rules also take a time‐
2292 out value (in seconds) as an option. If more than one rule uses the
2293 same timer label, the timer will be restarted whenever any of the rules
2294 get a hit. One entry for each timer is created in sysfs. This attri‐
2295 bute contains the timer remaining for the timer to expire. The at‐
2296 tributes are located under the xt_idletimer class:
2297 /sys/class/xt_idletimer/timers/<label>
2299 When the timer expires, the target module sends a sysfs notification to
2301 the userspace, which can then decide what to do (eg. disconnect to save
2302 power).
2303 --timeout amount
2305 This is the time in seconds that will trigger the notification.
2306 --label string
2308 This is a unique identifier for the timer. The maximum length
2309 for the label string is 27 characters.
2310 LED
2312 This creates an LED-trigger that can then be attached to system indica‐
2313 tor lights, to blink or illuminate them when certain packets pass
2314 through the system. One example might be to light up an LED for a few
2315 minutes every time an SSH connection is made to the local machine. The
2316 following options control the trigger behavior:
2317 --led-trigger-id name
2319 This is the name given to the LED trigger. The actual name of
2320 the trigger will be prefixed with "netfilter-".
2321 --led-delay ms
2323 This indicates how long (in milliseconds) the LED should be left
2324 illuminated when a packet arrives before being switched off
2325 again. The default is 0 (blink as fast as possible.) The special
2326 value inf can be given to leave the LED on permanently once ac‐
2327 tivated. (In this case the trigger will need to be manually de‐
2328 tached and reattached to the LED device to switch it off again.)
2329 --led-always-blink
2331 Always make the LED blink on packet arrival, even if the LED is
2332 already on. This allows notification of new packets even with
2333 long delay values (which otherwise would result in a silent pro‐
2334 longing of the delay time.)
2335 Example:
2337 Create an LED trigger for incoming SSH traffic:
2339 iptables -A INPUT -p tcp --dport 22 -j LED --led-trigger-id ssh
2340 Then attach the new trigger to an LED:
2342 echo netfilter-ssh >/sys/class/leds/ledname/trigger
2343 LOG
2345 Turn on kernel logging of matching packets. When this option is set
2346 for a rule, the Linux kernel will print some information on all match‐
2347 ing packets (like most IP/IPv6 header fields) via the kernel log (where
2348 it can be read with dmesg(1) or read in the syslog).
2349 This is a "non-terminating target", i.e. rule traversal continues at
2351 the next rule. So if you want to LOG the packets you refuse, use two
2352 separate rules with the same matching criteria, first using target LOG
2353 then DROP (or REJECT).
2354 --log-level level
2356 Level of logging, which can be (system-specific) numeric or a
2357 mnemonic. Possible values are (in decreasing order of prior‐
2358 ity): emerg, alert, crit, error, warning, notice, info or debug.
2359 --log-prefix prefix
2361 Prefix log messages with the specified prefix; up to 29 letters
2362 long, and useful for distinguishing messages in the logs.
2363 --log-tcp-sequence
2365 Log TCP sequence numbers. This is a security risk if the log is
2366 readable by users.
2367 --log-tcp-options
2369 Log options from the TCP packet header.
2370 --log-ip-options
2372 Log options from the IP/IPv6 packet header.
2373 --log-uid
2375 Log the userid of the process which generated the packet.
2376 --log-macdecode
2378 Log MAC addresses and protocol.
2379 MARK
2381 This target is used to set the Netfilter mark value associated with the
2382 packet. It can, for example, be used in conjunction with routing based
2383 on fwmark (needs iproute2). If you plan on doing so, note that the mark
2384 needs to be set in either the PREROUTING or the OUTPUT chain of the
2385 mangle table to affect routing. The mark field is 32 bits wide.
2386 --set-xmark value[/mask]
2388 Zeroes out the bits given by mask and XORs value into the packet
2389 mark ("nfmark"). If mask is omitted, 0xFFFFFFFF is assumed.
2390 --set-mark value[/mask]
2392 Zeroes out the bits given by mask and ORs value into the packet
2393 mark. If mask is omitted, 0xFFFFFFFF is assumed.
2394 The following mnemonics are available:
2396 --and-mark bits
2398 Binary AND the nfmark with bits. (Mnemonic for --set-xmark 0/in‐
2399 vbits, where invbits is the binary negation of bits.)
2400 --or-mark bits
2402 Binary OR the nfmark with bits. (Mnemonic for --set-xmark
2403 bits/bits.)
2404 --xor-mark bits
2406 Binary XOR the nfmark with bits. (Mnemonic for --set-xmark
2407 bits/0.)
2408 MASQUERADE
2410 This target is only valid in the nat table, in the POSTROUTING chain.
2411 It should only be used with dynamically assigned IP (dialup) connec‐
2412 tions: if you have a static IP address, you should use the SNAT target.
2413 Masquerading is equivalent to specifying a mapping to the IP address of
2414 the interface the packet is going out, but also has the effect that
2415 connections are forgotten when the interface goes down. This is the
2416 correct behavior when the next dialup is unlikely to have the same in‐
2417 terface address (and hence any established connections are lost any‐
2418 way).
2419 --to-ports port[-port]
2421 This specifies a range of source ports to use, overriding the
2422 default SNAT source port-selection heuristics (see above). This
2423 is only valid if the rule also specifies one of the following
2424 protocols: tcp, udp, dccp or sctp.
2425 --random
2427 Randomize source port mapping (kernel >= 2.6.21). Since kernel
2428 5.0, --random is identical to --random-fully.
2429 --random-fully
2431 Fully randomize source port mapping (kernel >= 3.13).
2432 IPv6 support available since Linux kernels >= 3.7.
2434 NETMAP
2436 This target allows you to statically map a whole network of addresses
2437 onto another network of addresses. It can only be used from rules in
2438 the nat table.
2439 --to address[/mask]
2441 Network address to map to. The resulting address will be con‐
2442 structed in the following way: All 'one' bits in the mask are
2443 filled in from the new `address'. All bits that are zero in the
2444 mask are filled in from the original address.
2445 IPv6 support available since Linux kernels >= 3.7.
2447 NFLOG
2449 This target provides logging of matching packets. When this target is
2450 set for a rule, the Linux kernel will pass the packet to the loaded
2451 logging backend to log the packet. This is usually used in combination
2452 with nfnetlink_log as logging backend, which will multicast the packet
2453 through a netlink socket to the specified multicast group. One or more
2454 userspace processes may subscribe to the group to receive the packets.
2455 Like LOG, this is a non-terminating target, i.e. rule traversal contin‐
2456 ues at the next rule.
2457 --nflog-group nlgroup
2459 The netlink group (0 - 2^16-1) to which packets are (only appli‐
2460 cable for nfnetlink_log). The default value is 0.
2461 --nflog-prefix prefix
2463 A prefix string to include in the log message, up to 64 charac‐
2464 ters long, useful for distinguishing messages in the logs.
2465 --nflog-range size
2467 This option has never worked, use --nflog-size instead
2468 --nflog-size size
2470 The number of bytes to be copied to userspace (only applicable
2471 for nfnetlink_log). nfnetlink_log instances may specify their
2472 own range, this option overrides it.
2473 --nflog-threshold size
2475 Number of packets to queue inside the kernel before sending them
2476 to userspace (only applicable for nfnetlink_log). Higher values
2477 result in less overhead per packet, but increase delay until the
2478 packets reach userspace. The default value is 1.
2479 NFQUEUE
2481 This target passes the packet to userspace using the nfnetlink_queue
2482 handler. The packet is put into the queue identified by its 16-bit
2483 queue number. Userspace can inspect and modify the packet if desired.
2484 Userspace must then drop or reinject the packet into the kernel.
2485 Please see libnetfilter_queue for details. nfnetlink_queue was added
2486 in Linux 2.6.14. The queue-balance option was added in Linux 2.6.31,
2487 queue-bypass in 2.6.39.
2488 --queue-num value
2490 This specifies the QUEUE number to use. Valid queue numbers are
2491 0 to 65535. The default value is 0.
2492 --queue-balance value:value
2494 This specifies a range of queues to use. Packets are then bal‐
2495 anced across the given queues. This is useful for multicore
2496 systems: start multiple instances of the userspace program on
2497 queues x, x+1, .. x+n and use "--queue-balance x:x+n". Packets
2498 belonging to the same connection are put into the same nfqueue.
2499 Due to implementation details, a lower range value of 0 limits
2500 the higher range value to 65534, i.e. one can only balance be‐
2501 tween at most 65535 queues.
2502 --queue-bypass
2504 By default, if no userspace program is listening on an NFQUEUE,
2505 then all packets that are to be queued are dropped. When this
2506 option is used, the NFQUEUE rule behaves like ACCEPT instead,
2507 and the packet will move on to the next table.
2508 --queue-cpu-fanout
2510 Available starting Linux kernel 3.10. When used together with
2511 --queue-balance this will use the CPU ID as an index to map
2512 packets to the queues. The idea is that you can improve perfor‐
2513 mance if there's a queue per CPU. This requires --queue-balance
2514 to be specified.
2515 NOTRACK
2517 This extension disables connection tracking for all packets matching
2518 that rule. It is equivalent with -j CT --notrack. Like CT, NOTRACK can
2519 only be used in the raw table.
2520 RATEEST
2522 The RATEEST target collects statistics, performs rate estimation calcu‐
2523 lation and saves the results for later evaluation using the rateest
2524 match.
2525 --rateest-name name
2527 Count matched packets into the pool referred to by name, which
2528 is freely choosable.
2529 --rateest-interval amount{s|ms|us}
2531 Rate measurement interval, in seconds, milliseconds or microsec‐
2532 onds.
2533 --rateest-ewmalog value
2535 Rate measurement averaging time constant.
2536 REDIRECT
2538 This target is only valid in the nat table, in the PREROUTING and OUT‐
2539 PUT chains, and user-defined chains which are only called from those
2540 chains. It redirects the packet to the machine itself by changing the
2541 destination IP to the primary address of the incoming interface (lo‐
2542 cally-generated packets are mapped to the localhost address, 127.0.0.1
2543 for IPv4 and ::1 for IPv6, and packets arriving on interfaces that
2544 don't have an IP address configured are dropped).
2545 --to-ports port[-port]
2547 This specifies a destination port or range of ports to use:
2548 without this, the destination port is never altered. This is
2549 only valid if the rule also specifies one of the following pro‐
2550 tocols: tcp, udp, dccp or sctp. For a single port, a service
2551 name as listed in /etc/services may be used.
2552 --random
2554 Randomize source port mapping (kernel >= 2.6.22).
2555 IPv6 support available starting Linux kernels >= 3.7.
2557 REJECT (IPv6-specific)
2559 This is used to send back an error packet in response to the matched
2560 packet: otherwise it is equivalent to DROP so it is a terminating TAR‐
2561 GET, ending rule traversal. This target is only valid in the INPUT,
2562 FORWARD and OUTPUT chains, and user-defined chains which are only
2563 called from those chains. The following option controls the nature of
2564 the error packet returned:
2565 --reject-with type
2567 The type given can be icmp6-no-route, no-route, icmp6-adm-pro‐
2568 hibited, adm-prohibited, icmp6-addr-unreachable, addr-unreach,
2569 or icmp6-port-unreachable, which return the appropriate ICMPv6
2570 error message (icmp6-port-unreachable is the default). Finally,
2571 the option tcp-reset can be used on rules which only match the
2572 TCP protocol: this causes a TCP RST packet to be sent back.
2573 This is mainly useful for blocking ident (113/tcp) probes which
2574 frequently occur when sending mail to broken mail hosts (which
2575 won't accept your mail otherwise). tcp-reset can only be used
2576 with kernel versions 2.6.14 or later.
2577 Warning: You should not indiscriminately apply the REJECT target to
2579 packets whose connection state is classified as INVALID; instead, you
2580 should only DROP these.
2581 Consider a source host transmitting a packet P, with P experiencing so
2583 much delay along its path that the source host issues a retransmission,
2584 P_2, with P_2 being successful in reaching its destination and advanc‐
2585 ing the connection state normally. It is conceivable that the late-ar‐
2586 riving P may be considered not to be associated with any connection
2587 tracking entry. Generating a reject response for a packet so classed
2588 would then terminate the healthy connection.
2589 So, instead of:
2591 -A INPUT ... -j REJECT
2593 do consider using:
2595 -A INPUT ... -m conntrack --ctstate INVALID -j DROP -A INPUT ... -j RE‐
2597 JECT
2598 REJECT (IPv4-specific)
2600 This is used to send back an error packet in response to the matched
2601 packet: otherwise it is equivalent to DROP so it is a terminating TAR‐
2602 GET, ending rule traversal. This target is only valid in the INPUT,
2603 FORWARD and OUTPUT chains, and user-defined chains which are only
2604 called from those chains. The following option controls the nature of
2605 the error packet returned:
2606 --reject-with type
2608 The type given can be icmp-net-unreachable, icmp-host-unreach‐
2609 able, icmp-port-unreachable, icmp-proto-unreachable,
2610 icmp-net-prohibited, icmp-host-prohibited, or icmp-admin-prohib‐
2611 ited (*), which return the appropriate ICMP error message
2612 (icmp-port-unreachable is the default). The option tcp-reset
2613 can be used on rules which only match the TCP protocol: this
2614 causes a TCP RST packet to be sent back. This is mainly useful
2615 for blocking ident (113/tcp) probes which frequently occur when
2616 sending mail to broken mail hosts (which won't accept your mail
2617 otherwise).
2618 (*) Using icmp-admin-prohibited with kernels that do not support
2620 it will result in a plain DROP instead of REJECT
2621 Warning: You should not indiscriminately apply the REJECT target to
2623 packets whose connection state is classified as INVALID; instead, you
2624 should only DROP these.
2625 Consider a source host transmitting a packet P, with P experiencing so
2627 much delay along its path that the source host issues a retransmission,
2628 P_2, with P_2 being successful in reaching its destination and advanc‐
2629 ing the connection state normally. It is conceivable that the late-ar‐
2630 riving P may be considered not to be associated with any connection
2631 tracking entry. Generating a reject response for a packet so classed
2632 would then terminate the healthy connection.
2633 So, instead of:
2635 -A INPUT ... -j REJECT
2637 do consider using:
2639 -A INPUT ... -m conntrack --ctstate INVALID -j DROP -A INPUT ... -j RE‐
2641 JECT
2642 SECMARK
2644 This is used to set the security mark value associated with the packet
2645 for use by security subsystems such as SELinux. It is valid in the se‐
2646 curity table (for backwards compatibility with older kernels, it is
2647 also valid in the mangle table). The mark is 32 bits wide.
2648 --selctx security_context
2650 SET
2652 This module adds and/or deletes entries from IP sets which can be de‐
2653 fined by ipset(8).
2654 --add-set setname flag[,flag...]
2656 add the address(es)/port(s) of the packet to the set
2657 --del-set setname flag[,flag...]
2659 delete the address(es)/port(s) of the packet from the set
2660 --map-set setname flag[,flag...]
2662 [--map-mark] [--map-prio] [--map-queue] map packet properties
2663 (firewall mark, tc priority, hardware queue)
2664 where flag(s) are src and/or dst specifications and there can be
2666 no more than six of them.
2667 --timeout value
2669 when adding an entry, the timeout value to use instead of the
2670 default one from the set definition
2671 --exist
2673 when adding an entry if it already exists, reset the timeout
2674 value to the specified one or to the default from the set defi‐
2675 nition
2676 --map-set set-name
2678 the set-name should be created with --skbinfo option --map-mark
2679 map firewall mark to packet by lookup of value in the set
2680 --map-prio map traffic control priority to packet by lookup of
2681 value in the set --map-queue map hardware NIC queue to packet by
2682 lookup of value in the set
2683 The --map-set option can be used from the mangle table only. The
2685 --map-prio and --map-queue flags can be used in the OUTPUT, FOR‐
2686 WARD and POSTROUTING chains.
2687 Use of -j SET requires that ipset kernel support is provided, which,
2689 for standard kernels, is the case since Linux 2.6.39.
2690 SNAT
2692 This target is only valid in the nat table, in the POSTROUTING and IN‐
2693 PUT chains, and user-defined chains which are only called from those
2694 chains. It specifies that the source address of the packet should be
2695 modified (and all future packets in this connection will also be man‐
2696 gled), and rules should cease being examined. It takes the following
2697 options:
2698 --to-source [ipaddr[-ipaddr]][:port[-port]]
2700 which can specify a single new source IP address, an inclusive
2701 range of IP addresses. Optionally a port range, if the rule also
2702 specifies one of the following protocols: tcp, udp, dccp or
2703 sctp. If no port range is specified, then source ports below
2704 512 will be mapped to other ports below 512: those between 512
2705 and 1023 inclusive will be mapped to ports below 1024, and other
2706 ports will be mapped to 1024 or above. Where possible, no port
2707 alteration will occur.
2708 --random
2710 Randomize source port mapping through a hash-based algorithm
2711 (kernel >= 2.6.21).
2712 --random-fully
2714 Fully randomize source port mapping through a PRNG (kernel >=
2715 3.14).
2716 --persistent
2718 Gives a client the same source-/destination-address for each
2719 connection. This supersedes the SAME target. Support for per‐
2720 sistent mappings is available from 2.6.29-rc2.
2721 Kernels prior to 2.6.36-rc1 don't have the ability to SNAT in the INPUT
2723 chain.
2724 IPv6 support available since Linux kernels >= 3.7.
2726 SNPT (IPv6-specific)
2728 Provides stateless source IPv6-to-IPv6 Network Prefix Translation (as
2729 described by RFC 6296).
2730 You have to use this target in the mangle table, not in the nat table.
2732 It takes the following options:
2733 --src-pfx [prefix/length]
2735 Set source prefix that you want to translate and length
2736 --dst-pfx [prefix/length]
2738 Set destination prefix that you want to use in the translation
2739 and length
2740 You have to use the DNPT target to undo the translation. Example:
2742 ip6tables -t mangle -I POSTROUTING -s fd00::/64 -o vboxnet0 -j
2744 SNPT --src-pfx fd00::/64 --dst-pfx 2001:e20:2000:40f::/64
2745 ip6tables -t mangle -I PREROUTING -i wlan0 -d
2747 2001:e20:2000:40f::/64 -j DNPT --src-pfx 2001:e20:2000:40f::/64
2748 --dst-pfx fd00::/64
2749 You may need to enable IPv6 neighbor proxy:
2751 sysctl -w net.ipv6.conf.all.proxy_ndp=1
2753 You also have to use the NOTRACK target to disable connection tracking
2755 for translated flows.
2756 SYNPROXY
2758 This target will process TCP three-way-handshake parallel in netfilter
2759 context to protect either local or backend system. This target requires
2760 connection tracking because sequence numbers need to be translated.
2761 The kernels ability to absorb SYNFLOOD was greatly improved starting
2762 with Linux 4.4, so this target should not be needed anymore to protect
2763 Linux servers.
2764 --mss maximum segment size
2766 Maximum segment size announced to clients. This must match the
2767 backend.
2768 --wscale window scale
2770 Window scale announced to clients. This must match the backend.
2771 --sack-perm
2773 Pass client selective acknowledgement option to backend (will be
2774 disabled if not present).
2775 --timestamps
2777 Pass client timestamp option to backend (will be disabled if not
2778 present, also needed for selective acknowledgement and window
2779 scaling).
2780 Example:
2782 Determine tcp options used by backend, from an external system
2784 tcpdump -pni eth0 -c 1 'tcp[tcpflags] == (tcp-syn|tcp-ack)'
2786 port 80 &
2787 telnet 192.0.2.42 80
2788 18:57:24.693307 IP 192.0.2.42.80 > 192.0.2.43.48757:
2789 Flags [S.], seq 360414582, ack 788841994, win 14480,
2790 options [mss 1460,sackOK,
2791 TS val 1409056151 ecr 9690221,
2792 nop,wscale 9],
2793 length 0
2794 Switch tcp_loose mode off, so conntrack will mark out-of-flow packets
2796 as state INVALID.
2797 echo 0 > /proc/sys/net/netfilter/nf_conntrack_tcp_loose
2799 Make SYN packets untracked
2801 iptables -t raw -A PREROUTING -i eth0 -p tcp --dport 80
2803 --syn -j CT --notrack
2804 Catch UNTRACKED (SYN packets) and INVALID (3WHS ACK packets) states and
2806 send them to SYNPROXY. This rule will respond to SYN packets with
2807 SYN+ACK syncookies, create ESTABLISHED for valid client response (3WHS
2808 ACK packets) and drop incorrect cookies. Flags combinations not ex‐
2809 pected during 3WHS will not match and continue (e.g. SYN+FIN, SYN+ACK).
2810 iptables -A INPUT -i eth0 -p tcp --dport 80
2812 -m state --state UNTRACKED,INVALID -j SYNPROXY
2813 --sack-perm --timestamp --mss 1460 --wscale 9
2814 Drop invalid packets, this will be out-of-flow packets that were not
2816 matched by SYNPROXY.
2817 iptables -A INPUT -i eth0 -p tcp --dport 80 -m state --state IN‐
2819 VALID -j DROP
2820 TCPMSS
2822 This target alters the MSS value of TCP SYN packets, to control the
2823 maximum size for that connection (usually limiting it to your outgoing
2824 interface's MTU minus 40 for IPv4 or 60 for IPv6, respectively). Of
2825 course, it can only be used in conjunction with -p tcp.
2826 This target is used to overcome criminally braindead ISPs or servers
2828 which block "ICMP Fragmentation Needed" or "ICMPv6 Packet Too Big"
2829 packets. The symptoms of this problem are that everything works fine
2830 from your Linux firewall/router, but machines behind it can never ex‐
2831 change large packets:
2832 1. Web browsers connect, then hang with no data received.
2834 2. Small mail works fine, but large emails hang.
2836 3. ssh works fine, but scp hangs after initial handshaking.
2838 Workaround: activate this option and add a rule to your firewall con‐
2840 figuration like:
2841 iptables -t mangle -A FORWARD -p tcp --tcp-flags SYN,RST SYN
2843 -j TCPMSS --clamp-mss-to-pmtu
2844 --set-mss value
2846 Explicitly sets MSS option to specified value. If the MSS of the
2847 packet is already lower than value, it will not be increased
2848 (from Linux 2.6.25 onwards) to avoid more problems with hosts
2849 relying on a proper MSS.
2850 --clamp-mss-to-pmtu
2852 Automatically clamp MSS value to (path_MTU - 40 for IPv4; -60
2853 for IPv6). This may not function as desired where asymmetric
2854 routes with differing path MTU exist — the kernel uses the path
2855 MTU which it would use to send packets from itself to the source
2856 and destination IP addresses. Prior to Linux 2.6.25, only the
2857 path MTU to the destination IP address was considered by this
2858 option; subsequent kernels also consider the path MTU to the
2859 source IP address.
2860 These options are mutually exclusive.
2862 TCPOPTSTRIP
2864 This target will strip TCP options off a TCP packet. (It will actually
2865 replace them by NO-OPs.) As such, you will need to add the -p tcp pa‐
2866 rameters.
2867 --strip-options option[,option...]
2869 Strip the given option(s). The options may be specified by TCP
2870 option number or by symbolic name. The list of recognized op‐
2871 tions can be obtained by calling iptables with -j TCPOPTSTRIP
2872 -h.
2873 TEE
2875 The TEE target will clone a packet and redirect this clone to another
2876 machine on the local network segment. In other words, the nexthop must
2877 be the target, or you will have to configure the nexthop to forward it
2878 further if so desired.
2879 --gateway ipaddr
2881 Send the cloned packet to the host reachable at the given IP ad‐
2882 dress. Use of 0.0.0.0 (for IPv4 packets) or :: (IPv6) is in‐
2883 valid.
2884 To forward all incoming traffic on eth0 to an Network Layer logging
2886 box:
2887 -t mangle -A PREROUTING -i eth0 -j TEE --gateway 2001:db8::1
2889 TOS
2891 This module sets the Type of Service field in the IPv4 header (includ‐
2892 ing the "precedence" bits) or the Priority field in the IPv6 header.
2893 Note that TOS shares the same bits as DSCP and ECN. The TOS target is
2894 only valid in the mangle table.
2895 --set-tos value[/mask]
2897 Zeroes out the bits given by mask (see NOTE below) and XORs
2898 value into the TOS/Priority field. If mask is omitted, 0xFF is
2899 assumed.
2900 --set-tos symbol
2902 You can specify a symbolic name when using the TOS target for
2903 IPv4. It implies a mask of 0xFF (see NOTE below). The list of
2904 recognized TOS names can be obtained by calling iptables with -j
2905 TOS -h.
2906 The following mnemonics are available:
2908 --and-tos bits
2910 Binary AND the TOS value with bits. (Mnemonic for --set-tos
2911 0/invbits, where invbits is the binary negation of bits. See
2912 NOTE below.)
2913 --or-tos bits
2915 Binary OR the TOS value with bits. (Mnemonic for --set-tos
2916 bits/bits. See NOTE below.)
2917 --xor-tos bits
2919 Binary XOR the TOS value with bits. (Mnemonic for --set-tos
2920 bits/0. See NOTE below.)
2921 NOTE: In Linux kernels up to and including 2.6.38, with the exception
2923 of longterm releases 2.6.32 (>=.42), 2.6.33 (>=.15), and 2.6.35
2924 (>=.14), there is a bug whereby IPv6 TOS mangling does not behave as
2925 documented and differs from the IPv4 version. The TOS mask indicates
2926 the bits one wants to zero out, so it needs to be inverted before ap‐
2927 plying it to the original TOS field. However, the aformentioned kernels
2928 forgo the inversion which breaks --set-tos and its mnemonics.
2929 TPROXY
2931 This target is only valid in the mangle table, in the PREROUTING chain
2932 and user-defined chains which are only called from this chain. It redi‐
2933 rects the packet to a local socket without changing the packet header
2934 in any way. It can also change the mark value which can then be used in
2935 advanced routing rules. It takes three options:
2936 --on-port port
2938 This specifies a destination port to use. It is a required op‐
2939 tion, 0 means the new destination port is the same as the origi‐
2940 nal. This is only valid if the rule also specifies -p tcp or -p
2941 udp.
2942 --on-ip address
2944 This specifies a destination address to use. By default the ad‐
2945 dress is the IP address of the incoming interface. This is only
2946 valid if the rule also specifies -p tcp or -p udp.
2947 --tproxy-mark value[/mask]
2949 Marks packets with the given value/mask. The fwmark value set
2950 here can be used by advanced routing. (Required for transparent
2951 proxying to work: otherwise these packets will get forwarded,
2952 which is probably not what you want.)
2953 TRACE
2955 This target marks packets so that the kernel will log every rule which
2956 match the packets as those traverse the tables, chains, rules. It can
2957 only be used in the raw table.
2958 With iptables-legacy, a logging backend, such as ip(6)t_LOG or
2960 nfnetlink_log, must be loaded for this to be visible. The packets are
2961 logged with the string prefix: "TRACE: tablename:chainname:type:rulenum
2962 " where type can be "rule" for plain rule, "return" for implicit rule
2963 at the end of a user defined chain and "policy" for the policy of the
2964 built in chains.
2965 With iptables-nft, the target is translated into nftables' meta nftrace
2967 expression. Hence the kernel sends trace events via netlink to
2968 userspace where they may be displayed using xtables-monitor --trace
2969 command. For details, refer to xtables-monitor(8).
2970 TTL (IPv4-specific)
2972 This is used to modify the IPv4 TTL header field. The TTL field deter‐
2973 mines how many hops (routers) a packet can traverse until it's time to
2974 live is exceeded.
2975 Setting or incrementing the TTL field can potentially be very danger‐
2977 ous, so it should be avoided at any cost. This target is only valid in
2978 mangle table.
2979 Don't ever set or increment the value on packets that leave your local
2981 network!
2982 --ttl-set value
2984 Set the TTL value to `value'.
2985 --ttl-dec value
2987 Decrement the TTL value `value' times.
2988 --ttl-inc value
2990 Increment the TTL value `value' times.
2991 ULOG (IPv4-specific)
2993 This is the deprecated ipv4-only predecessor of the NFLOG target. It
2994 provides userspace logging of matching packets. When this target is
2995 set for a rule, the Linux kernel will multicast this packet through a
2996 netlink socket. One or more userspace processes may then subscribe to
2997 various multicast groups and receive the packets. Like LOG, this is a
2998 "non-terminating target", i.e. rule traversal continues at the next
2999 rule.
3000 --ulog-nlgroup nlgroup
3002 This specifies the netlink group (1-32) to which the packet is
3003 sent. Default value is 1.
3004 --ulog-prefix prefix
3006 Prefix log messages with the specified prefix; up to 32 charac‐
3007 ters long, and useful for distinguishing messages in the logs.
3008 --ulog-cprange size
3010 Number of bytes to be copied to userspace. A value of 0 always
3011 copies the entire packet, regardless of its size. Default is 0.
3012 --ulog-qthreshold size
3014 Number of packet to queue inside kernel. Setting this value to,
3015 e.g. 10 accumulates ten packets inside the kernel and transmits
3016 them as one netlink multipart message to userspace. Default is
3017 1 (for backwards compatibility).
3018iptables 1.8.9 iptables-extensions(8)