1IP-ROUTE(8) Linux IP-ROUTE(8)
2
6 ip-route - routing table management
7
9 ip [ ip-OPTIONS ] route { COMMAND | help }
10 ip route { show | flush } SELECTOR
13 ip route save SELECTOR
15 ip route restore
17 ip route get ADDRESS [ from ADDRESS iif STRING ] [ oif STRING ] [ tos
19 TOS ] [ vrf NAME ]
20 ip route { add | del | change | append | replace } ROUTE
22 SELECTOR := [ root PREFIX ] [ match PREFIX ] [ exact PREFIX ] [ table
24 TABLE_ID ] [ vrf NAME ] [ proto RTPROTO ] [ type TYPE ] [ scope
25 SCOPE ]
26 ROUTE := NODE_SPEC [ INFO_SPEC ]
28 NODE_SPEC := [ TYPE ] PREFIX [ tos TOS ] [ table TABLE_ID ] [ proto
30 RTPROTO ] [ scope SCOPE ] [ metric METRIC ]
31 INFO_SPEC := NH OPTIONS FLAGS [ nexthop NH ] ...
33 NH := [ encap ENCAP ] [ via [ FAMILY ] ADDRESS ] [ dev STRING ] [
35 weight NUMBER ] NHFLAGS
36 FAMILY := [ inet | inet6 | ipx | dnet | mpls | bridge | link ]
38 OPTIONS := FLAGS [ mtu NUMBER ] [ advmss NUMBER ] [ as [ to ] ADDRESS ]
40 rtt TIME ] [ rttvar TIME ] [ reordering NUMBER ] [ window NUM‐
41 BER ] [ cwnd NUMBER ] [ ssthresh REALM ] [ realms REALM ] [
42 rto_min TIME ] [ initcwnd NUMBER ] [ initrwnd NUMBER ] [ fea‐
43 tures FEATURES ] [ quickack BOOL ] [ congctl NAME ] [ pref PREF
44 ] [ expires TIME ]
45 TYPE := [ unicast | local | broadcast | multicast | throw | unreachable
47 | prohibit | blackhole | nat ]
48 TABLE_ID := [ local| main | default | all | NUMBER ]
50 SCOPE := [ host | link | global | NUMBER ]
52 NHFLAGS := [ onlink | pervasive ]
54 RTPROTO := [ kernel | boot | static | NUMBER ]
56 FEATURES := [ ecn | ]
58 PREF := [ low | medium | high ]
60 ENCAP := [ MPLS | IP| BPF ]
62 ENCAP_MPLS := mpls [ LABEL ]
64 ENCAP_IP := ip id TUNNEL_ID dst REMOTE_IP [ tos TOS ] [ ttl TTL ]
66 ENCAP_BPF := bpf [ in PROG ] [ out PROG ] [ xmit PROG ] [ headroom SIZE
69 ]
70
73 ip route is used to manipulate entries in the kernel routing tables.
74 Route types:
76 unicast - the route entry describes real paths to the destina‐
78 tions covered by the route prefix.
79 unreachable - these destinations are unreachable. Packets are
82 discarded and the ICMP message host unreachable is generated.
83 The local senders get an EHOSTUNREACH error.
84 blackhole - these destinations are unreachable. Packets are
87 discarded silently. The local senders get an EINVAL error.
88 prohibit - these destinations are unreachable. Packets are dis‐
91 carded and the ICMP message communication administratively pro‐
92 hibited is generated. The local senders get an EACCES error.
93 local - the destinations are assigned to this host. The packets
96 are looped back and delivered locally.
97 broadcast - the destinations are broadcast addresses. The pack‐
100 ets are sent as link broadcasts.
101 throw - a special control route used together with policy
104 rules. If such a route is selected, lookup in this table is
105 terminated pretending that no route was found. Without policy
106 routing it is equivalent to the absence of the route in the
107 routing table. The packets are dropped and the ICMP message net
108 unreachable is generated. The local senders get an ENETUNREACH
109 error.
110 nat - a special NAT route. Destinations covered by the prefix
113 are considered to be dummy (or external) addresses which
114 require translation to real (or internal) ones before forward‐
115 ing. The addresses to translate to are selected with the
116 attribute via. Warning: Route NAT is no longer supported in
117 Linux 2.6.
118 anycast - not implemented the destinations are anycast
121 addresses assigned to this host. They are mainly equivalent to
122 local with one difference: such addresses are invalid when used
123 as the source address of any packet.
124 multicast - a special type used for multicast routing. It is
127 not present in normal routing tables.
128 Route tables: Linux-2.x can pack routes into several routing tables
131 identified by a number in the range from 1 to 2^31 or by name from the
132 file /etc/iproute2/rt_tables By default all normal routes are inserted
133 into the main table (ID 254) and the kernel only uses this table when
134 calculating routes. Values (0, 253, 254, and 255) are reserved for
135 built-in use.
136 Actually, one other table always exists, which is invisible but even
139 more important. It is the local table (ID 255). This table consists of
140 routes for local and broadcast addresses. The kernel maintains this ta‐
141 ble automatically and the administrator usually need not modify it or
142 even look at it.
143 The multiple routing tables enter the game when policy routing is used.
145 ip route add
148 add new route
149 ip route change
151 change route
152 ip route replace
154 change or add new one
155 to TYPE PREFIX (default)
157 the destination prefix of the route. If TYPE is omitted,
158 ip assumes type unicast. Other values of TYPE are listed
159 above. PREFIX is an IP or IPv6 address optionally fol‐
160 lowed by a slash and the prefix length. If the length of
161 the prefix is missing, ip assumes a full-length host
162 route. There is also a special PREFIX default - which is
163 equivalent to IP 0/0 or to IPv6 ::/0.
164 tos TOS
167 dsfield TOS
169 the Type Of Service (TOS) key. This key has no associated
170 mask and the longest match is understood as: First, com‐
171 pare the TOS of the route and of the packet. If they are
172 not equal, then the packet may still match a route with a
173 zero TOS. TOS is either an 8 bit hexadecimal number or
174 an identifier from /etc/iproute2/rt_dsfield.
175 metric NUMBER
178 preference NUMBER
180 the preference value of the route. NUMBER is an arbi‐
181 trary 32bit number.
182 table TABLEID
185 the table to add this route to. TABLEID may be a number
186 or a string from the file /etc/iproute2/rt_tables. If
187 this parameter is omitted, ip assumes the main table,
188 with the exception of local, broadcast and nat routes,
189 which are put into the local table by default.
190 vrf NAME
193 the vrf name to add this route to. Implicitly means the
194 table associated with the VRF.
195 dev NAME
198 the output device name.
199 via [ FAMILY ] ADDRESS
202 the address of the nexthop router, in the address family
203 FAMILY. Actually, the sense of this field depends on the
204 route type. For normal unicast routes it is either the
205 true next hop router or, if it is a direct route
206 installed in BSD compatibility mode, it can be a local
207 address of the interface. For NAT routes it is the first
208 address of the block of translated IP destinations.
209 src ADDRESS
212 the source address to prefer when sending to the destina‐
213 tions covered by the route prefix.
214 realm REALMID
217 the realm to which this route is assigned. REALMID may
218 be a number or a string from the file
219 /etc/iproute2/rt_realms.
220 mtu MTU
223 mtu lock MTU
225 the MTU along the path to the destination. If the modi‐
226 fier lock is not used, the MTU may be updated by the ker‐
227 nel due to Path MTU Discovery. If the modifier lock is
228 used, no path MTU discovery will be tried, all packets
229 will be sent without the DF bit in IPv4 case or frag‐
230 mented to MTU for IPv6.
231 window NUMBER
234 the maximal window for TCP to advertise to these destina‐
235 tions, measured in bytes. It limits maximal data bursts
236 that our TCP peers are allowed to send to us.
237 rtt TIME
240 the initial RTT ('Round Trip Time') estimate. If no suf‐
241 fix is specified the units are raw values passed directly
242 to the routing code to maintain compatibility with previ‐
243 ous releases. Otherwise if a suffix of s, sec or secs is
244 used to specify seconds and ms, msec or msecs to specify
245 milliseconds.
246 rttvar TIME (2.3.15+ only)
250 the initial RTT variance estimate. Values are specified
251 as with rtt above.
252 rto_min TIME (2.6.23+ only)
255 the minimum TCP Retransmission TimeOut to use when commu‐
256 nicating with this destination. Values are specified as
257 with rtt above.
258 ssthresh NUMBER (2.3.15+ only)
261 an estimate for the initial slow start threshold.
262 cwnd NUMBER (2.3.15+ only)
265 the clamp for congestion window. It is ignored if the
266 lock flag is not used.
267 initcwnd NUMBER (2.5.70+ only)
270 the initial congestion window size for connections to
271 this destination. Actual window size is this value mul‐
272 tiplied by the MSS (``Maximal Segment Size'') for same
273 connection. The default is zero, meaning to use the val‐
274 ues specified in RFC2414.
275 initrwnd NUMBER (2.6.33+ only)
278 the initial receive window size for connections to this
279 destination. Actual window size is this value multiplied
280 by the MSS of the connection. The default value is zero,
281 meaning to use Slow Start value.
282 features FEATURES (3.18+only)
285 Enable or disable per-route features. Only available fea‐
286 ture at this time is ecn to enable explicit congestion
287 notification when initiating connections to the given
288 destination network. When responding to a connection
289 request from the given network, ecn will also be used
290 even if the net.ipv4.tcp_ecn sysctl is set to 0.
291 quickack BOOL (3.11+ only)
294 Enable or disable quick ack for connections to this des‐
295 tination.
296 congctl NAME (3.20+ only)
299 congctl lock NAME (3.20+ only)
301 Sets a specific TCP congestion control algorithm only for
302 a given destination. If not specified, Linux keeps the
303 current global default TCP congestion control algorithm,
304 or the one set from the application. If the modifier lock
305 is not used, an application may nevertheless overwrite
306 the suggested congestion control algorithm for that des‐
307 tination. If the modifier lock is used, then an applica‐
308 tion is not allowed to overwrite the specified congestion
309 control algorithm for that destination, thus it will be
310 enforced/guaranteed to use the proposed algorithm.
311 advmss NUMBER (2.3.15+ only)
314 the MSS ('Maximal Segment Size') to advertise to these
315 destinations when establishing TCP connections. If it is
316 not given, Linux uses a default value calculated from the
317 first hop device MTU. (If the path to these destination
318 is asymmetric, this guess may be wrong.)
319 reordering NUMBER (2.3.15+ only)
322 Maximal reordering on the path to this destination. If
323 it is not given, Linux uses the value selected with
324 sysctl variable net/ipv4/tcp_reordering.
325 nexthop NEXTHOP
328 the nexthop of a multipath route. NEXTHOP is a complex
329 value with its own syntax similar to the top level argu‐
330 ment lists:
331 via [ FAMILY ] ADDRESS - is the nexthop router.
333 dev NAME - is the output device.
336 weight NUMBER - is a weight for this element of a
339 multipath route reflecting its relative bandwidth
340 or quality.
341 The internal buffer used in iproute2 limits the maximum
343 number of nexthops that may be specified in one go. If
344 only ADDRESS is given, the current buffer size allows for
345 144 IPv6 nexthops and 253 IPv4 ones. For IPv4, this
346 effectively limits the number of nexthops possible per
347 route. With IPv6, further nexthops may be appended to the
348 same route via ip route append command.
349 scope SCOPE_VAL
352 the scope of the destinations covered by the route pre‐
353 fix. SCOPE_VAL may be a number or a string from the file
354 /etc/iproute2/rt_scopes. If this parameter is omitted,
355 ip assumes scope global for all gatewayed unicast routes,
356 scope link for direct unicast and broadcast routes and
357 scope host for local routes.
358 protocol RTPROTO
361 the routing protocol identifier of this route. RTPROTO
362 may be a number or a string from the file
363 /etc/iproute2/rt_protos. If the routing protocol ID is
364 not given, ip assumes protocol boot (i.e. it assumes the
365 route was added by someone who doesn't understand what
366 they are doing). Several protocol values have a fixed
367 interpretation. Namely:
368 redirect - the route was installed due to an ICMP
370 redirect.
371 kernel - the route was installed by the kernel
374 during autoconfiguration.
375 boot - the route was installed during the bootup
378 sequence. If a routing daemon starts, it will
379 purge all of them.
380 static - the route was installed by the adminis‐
383 trator to override dynamic routing. Routing dae‐
384 mon will respect them and, probably, even adver‐
385 tise them to its peers.
386 ra - the route was installed by Router Discovery
389 protocol.
390 The rest of the values are not reserved and the adminis‐
393 trator is free to assign (or not to assign) protocol
394 tags.
395 onlink pretend that the nexthop is directly attached to this
398 link, even if it does not match any interface prefix.
399 pref PREF
402 the IPv6 route preference. PREF is a string specifying
403 the route preference as defined in RFC4191 for Router
404 Discovery messages. Namely:
405 low - the route has a lowest priority
407 medium - the route has a default priority
410 high - the route has a highest priority
413 encap ENCAPTYPE ENCAPHDR
417 attach tunnel encapsulation attributes to this route.
418 ENCAPTYPE is a string specifying the supported encapsula‐
420 tion type. Namely:
421 mpls - encapsulation type MPLS
423 ip - IP encapsulation (Geneve, GRE, VXLAN, ...)
425 bpf - Execution of BPF program
427 ENCAPHDR is a set of encapsulation attributes specific to
430 the ENCAPTYPE.
431 mpls
433 MPLSLABEL - mpls label stack with labels sepa‐
434 rated by /
435 ip
438 id TUNNEL_ID dst REMOTE_IP [ tos TOS ] [ ttl
439 TTL ]
440 bpf
443 in PROG - BPF program to execute for incoming
444 packets
445 out PROG - BPF program to execute for outgoing
448 packets
449 xmit PROG - BPF program to execute for trans‐
452 mitted packets
453 headroom SIZE - Size of header BPF program will
456 attach (xmit)
457 expires TIME (4.4+ only)
462 the route will be deleted after the expires time. Only
463 support IPv6 at present.
464 ip route delete
467 delete route
468 ip route del has the same arguments as ip route add, but their
469 semantics are a bit different.
470 Key values (to, tos, preference and table) select the route to
472 delete. If optional attributes are present, ip verifies that
473 they coincide with the attributes of the route to delete. If no
474 route with the given key and attributes was found, ip route del
475 fails.
476 ip route show
479 list routes
480 the command displays the contents of the routing tables or the
481 route(s) selected by some criteria.
482 to SELECTOR (default)
485 only select routes from the given range of destinations.
486 SELECTOR consists of an optional modifier (root, match or
487 exact) and a prefix. root PREFIX selects routes with
488 prefixes not shorter than PREFIX. F.e. root 0/0 selects
489 the entire routing table. match PREFIX selects routes
490 with prefixes not longer than PREFIX. F.e. match
491 10.0/16 selects 10.0/16, 10/8 and 0/0, but it does not
492 select 10.1/16 and 10.0.0/24. And exact PREFIX (or just
493 PREFIX) selects routes with this exact prefix. If neither
494 of these options are present, ip assumes root 0/0 i.e. it
495 lists the entire table.
496 tos TOS
499 dsfield TOS
501 only select routes with the given TOS.
502 table TABLEID
505 show the routes from this table(s). The default setting
506 is to show table main. TABLEID may either be the ID of a
507 real table or one of the special values:
508 all - list all of the tables.
510 cache - dump the routing cache.
512 vrf NAME
515 show the routes for the table associated with the vrf
516 name
517 cloned
520 cached list cloned routes i.e. routes which were dynamically
522 forked from other routes because some route attribute
523 (f.e. MTU) was updated. Actually, it is equivalent to
524 table cache.
525 from SELECTOR
528 the same syntax as for to, but it binds the source
529 address range rather than destinations. Note that the
530 from option only works with cloned routes.
531 protocol RTPROTO
534 only list routes of this protocol.
535 scope SCOPE_VAL
538 only list routes with this scope.
539 type TYPE
542 only list routes of this type.
543 dev NAME
546 only list routes going via this device.
547 via [ FAMILY ] PREFIX
550 only list routes going via the nexthop routers selected
551 by PREFIX.
552 src PREFIX
555 only list routes with preferred source addresses selected
556 by PREFIX.
557 realm REALMID
560 realms FROMREALM/TOREALM
562 only list routes with these realms.
563 ip route flush
566 flush routing tables
567 this command flushes routes selected by some criteria.
568 The arguments have the same syntax and semantics as the argu‐
571 ments of ip route show, but routing tables are not listed but
572 purged. The only difference is the default action: show dumps
573 all the IP main routing table but flush prints the helper page.
574 With the -statistics option, the command becomes verbose. It
577 prints out the number of deleted routes and the number of rounds
578 made to flush the routing table. If the option is given twice,
579 ip route flush also dumps all the deleted routes in the format
580 described in the previous subsection.
581 ip route get
584 get a single route
585 this command gets a single route to a destination and prints its
586 contents exactly as the kernel sees it.
587 to ADDRESS (default)
590 the destination address.
591 from ADDRESS
594 the source address.
595 tos TOS
598 dsfield TOS
600 the Type Of Service.
601 iif NAME
604 the device from which this packet is expected to arrive.
605 oif NAME
608 force the output device on which this packet will be
609 routed.
610 vrf NAME
613 force the vrf device on which this packet will be routed.
614 connected
617 if no source address (option from) was given, relookup
618 the route with the source set to the preferred address
619 received from the first lookup. If policy routing is
620 used, it may be a different route.
621 Note that this operation is not equivalent to ip route show.
624 show shows existing routes. get resolves them and creates new
625 clones if necessary. Essentially, get is equivalent to sending a
626 packet along this path. If the iif argument is not given, the
627 kernel creates a route to output packets towards the requested
628 destination. This is equivalent to pinging the destination with
629 a subsequent ip route ls cache, however, no packets are actually
630 sent. With the iif argument, the kernel pretends that a packet
631 arrived from this interface and searches for a path to forward
632 the packet.
633 ip route save
636 save routing table information to stdout
637 This command behaves like ip route show except that the output
638 is raw data suitable for passing to ip route restore.
639 ip route restore
642 restore routing table information from stdin
643 This command expects to read a data stream as returned from ip
644 route save. It will attempt to restore the routing table infor‐
645 mation exactly as it was at the time of the save, so any trans‐
646 lation of information in the stream (such as device indexes)
647 must be done first. Any existing routes are left unchanged. Any
648 routes specified in the data stream that already exist in the
649 table will be ignored.
650
653 Starting with Linux kernel version 3.6, there is no routing cache for
654 IPv4 anymore. Hence ip route show cached will never print any entries
655 on systems with this or newer kernel versions.
656
659 ip ro
660 Show all route entries in the kernel.
661 ip route add default via 192.168.1.1 dev eth0
663 Adds a default route (for all addresses) via the local gateway
664 192.168.1.1 that can be reached on device eth0.
665 ip route add 10.1.1.0/30 encap mpls 200/300 via 10.1.1.1 dev eth0
667 Adds an ipv4 route with mpls encapsulation attributes attached to
668 it.
669
671 ip(8)
672
675 Original Manpage by Michail Litvak <mci@owl.openwall.com>
676iproute2 13 Dec 2012 IP-ROUTE(8)