1 .. SPDX-License-Identifier: GPL-2.0
7 /proc/sys/net/ipv4/* Variables
8 ==============================
11 - 0 - disabled (default)
14 Forward Packets between interfaces.
16 This variable is special, its change resets all configuration
17 parameters to their default state (RFC1122 for hosts, RFC1812
20 ip_default_ttl - INTEGER
21 Default value of TTL field (Time To Live) for outgoing (but not
22 forwarded) IP packets. Should be between 1 and 255 inclusive.
23 Default: 64 (as recommended by RFC1700)
25 ip_no_pmtu_disc - INTEGER
26 Disable Path MTU Discovery. If enabled in mode 1 and a
27 fragmentation-required ICMP is received, the PMTU to this
28 destination will be set to min_pmtu (see below). You will need
29 to raise min_pmtu to the smallest interface MTU on your system
30 manually if you want to avoid locally generated fragments.
32 In mode 2 incoming Path MTU Discovery messages will be
33 discarded. Outgoing frames are handled the same as in mode 1,
34 implicitly setting IP_PMTUDISC_DONT on every created socket.
36 Mode 3 is a hardened pmtu discover mode. The kernel will only
37 accept fragmentation-needed errors if the underlying protocol
38 can verify them besides a plain socket lookup. Current
39 protocols for which pmtu events will be honored are TCP, SCTP
40 and DCCP as they verify e.g. the sequence number or the
41 association. This mode should not be enabled globally but is
42 only intended to secure e.g. name servers in namespaces where
43 TCP path mtu must still work but path MTU information of other
44 protocols should be discarded. If enabled globally this mode
45 could break other protocols.
52 default 552 - minimum discovered Path MTU
54 ip_forward_use_pmtu - BOOLEAN
55 By default we don't trust protocol path MTUs while forwarding
56 because they could be easily forged and can lead to unwanted
57 fragmentation by the router.
58 You only need to enable this if you have user-space software
59 which tries to discover path mtus by itself and depends on the
60 kernel honoring this information. This is normally not the
70 fwmark_reflect - BOOLEAN
71 Controls the fwmark of kernel-generated IPv4 reply packets that are not
72 associated with a socket for example, TCP RSTs or ICMP echo replies).
73 If unset, these packets have a fwmark of zero. If set, they have the
74 fwmark of the packet they are replying to.
78 fib_multipath_use_neigh - BOOLEAN
79 Use status of existing neighbor entry when determining nexthop for
80 multipath routes. If disabled, neighbor information is not used and
81 packets could be directed to a failed nexthop. Only valid for kernels
82 built with CONFIG_IP_ROUTE_MULTIPATH enabled.
91 fib_multipath_hash_policy - INTEGER
92 Controls which hash policy to use for multipath routes. Only valid
93 for kernels built with CONFIG_IP_ROUTE_MULTIPATH enabled.
101 - 2 - Layer 3 or inner Layer 3 if present
102 - 3 - Custom multipath hash. Fields used for multipath hash calculation
103 are determined by fib_multipath_hash_fields sysctl
105 fib_multipath_hash_fields - UNSIGNED INTEGER
106 When fib_multipath_hash_policy is set to 3 (custom multipath hash), the
107 fields used for multipath hash calculation are determined by this
110 This value is a bitmask which enables various fields for multipath hash
115 ====== ============================
116 0x0001 Source IP address
117 0x0002 Destination IP address
119 0x0008 Unused (Flow Label)
121 0x0020 Destination port
122 0x0040 Inner source IP address
123 0x0080 Inner destination IP address
124 0x0100 Inner IP protocol
125 0x0200 Inner Flow Label
126 0x0400 Inner source port
127 0x0800 Inner destination port
128 ====== ============================
130 Default: 0x0007 (source IP, destination IP and IP protocol)
132 fib_sync_mem - UNSIGNED INTEGER
133 Amount of dirty memory from fib entries that can be backlogged before
134 synchronize_rcu is forced.
136 Default: 512kB Minimum: 64kB Maximum: 64MB
138 ip_forward_update_priority - INTEGER
139 Whether to update SKB priority from "TOS" field in IPv4 header after it
140 is forwarded. The new SKB priority is mapped from TOS field value
141 according to an rt_tos2priority table (see e.g. man tc-prio).
143 Default: 1 (Update priority.)
147 - 0 - Do not update priority.
148 - 1 - Update priority.
150 route/max_size - INTEGER
151 Maximum number of routes allowed in the kernel. Increase
152 this when using large numbers of interfaces and/or routes.
154 From linux kernel 3.6 onwards, this is deprecated for ipv4
155 as route cache is no longer used.
157 neigh/default/gc_thresh1 - INTEGER
158 Minimum number of entries to keep. Garbage collector will not
159 purge entries if there are fewer than this number.
163 neigh/default/gc_thresh2 - INTEGER
164 Threshold when garbage collector becomes more aggressive about
165 purging entries. Entries older than 5 seconds will be cleared
166 when over this number.
170 neigh/default/gc_thresh3 - INTEGER
171 Maximum number of non-PERMANENT neighbor entries allowed. Increase
172 this when using large numbers of interfaces and when communicating
173 with large numbers of directly-connected peers.
177 neigh/default/unres_qlen_bytes - INTEGER
178 The maximum number of bytes which may be used by packets
179 queued for each unresolved address by other network layers.
182 Setting negative value is meaningless and will return error.
184 Default: SK_WMEM_MAX, (same as net.core.wmem_default).
186 Exact value depends on architecture and kernel options,
187 but should be enough to allow queuing 256 packets
190 neigh/default/unres_qlen - INTEGER
191 The maximum number of packets which may be queued for each
192 unresolved address by other network layers.
194 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
196 Prior to linux 3.3, the default value is 3 which may cause
197 unexpected packet loss. The current default value is calculated
198 according to default value of unres_qlen_bytes and true size of
203 mtu_expires - INTEGER
204 Time, in seconds, that cached PMTU information is kept.
206 min_adv_mss - INTEGER
207 The advertised MSS depends on the first hop route MTU, but will
208 never be lower than this setting.
210 fib_notify_on_flag_change - INTEGER
211 Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/
212 RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed.
214 After installing a route to the kernel, user space receives an
215 acknowledgment, which means the route was installed in the kernel,
216 but not necessarily in hardware.
217 It is also possible for a route already installed in hardware to change
218 its action and therefore its flags. For example, a host route that is
219 trapping packets can be "promoted" to perform decapsulation following
220 the installation of an IPinIP/VXLAN tunnel.
221 The notifications will indicate to user-space the state of the route.
223 Default: 0 (Do not emit notifications.)
227 - 0 - Do not emit notifications.
228 - 1 - Emit notifications.
229 - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change.
233 ipfrag_high_thresh - LONG INTEGER
234 Maximum memory used to reassemble IP fragments.
236 ipfrag_low_thresh - LONG INTEGER
237 (Obsolete since linux-4.17)
238 Maximum memory used to reassemble IP fragments before the kernel
239 begins to remove incomplete fragment queues to free up resources.
240 The kernel still accepts new fragments for defragmentation.
242 ipfrag_time - INTEGER
243 Time in seconds to keep an IP fragment in memory.
245 ipfrag_max_dist - INTEGER
246 ipfrag_max_dist is a non-negative integer value which defines the
247 maximum "disorder" which is allowed among fragments which share a
248 common IP source address. Note that reordering of packets is
249 not unusual, but if a large number of fragments arrive from a source
250 IP address while a particular fragment queue remains incomplete, it
251 probably indicates that one or more fragments belonging to that queue
252 have been lost. When ipfrag_max_dist is positive, an additional check
253 is done on fragments before they are added to a reassembly queue - if
254 ipfrag_max_dist (or more) fragments have arrived from a particular IP
255 address between additions to any IP fragment queue using that source
256 address, it's presumed that one or more fragments in the queue are
257 lost. The existing fragment queue will be dropped, and a new one
258 started. An ipfrag_max_dist value of zero disables this check.
260 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
261 result in unnecessarily dropping fragment queues when normal
262 reordering of packets occurs, which could lead to poor application
263 performance. Using a very large value, e.g. 50000, increases the
264 likelihood of incorrectly reassembling IP fragments that originate
265 from different IP datagrams, which could result in data corruption.
271 inet_peer_threshold - INTEGER
272 The approximate size of the storage. Starting from this threshold
273 entries will be thrown aggressively. This threshold also determines
274 entries' time-to-live and time intervals between garbage collection
275 passes. More entries, less time-to-live, less GC interval.
277 inet_peer_minttl - INTEGER
278 Minimum time-to-live of entries. Should be enough to cover fragment
279 time-to-live on the reassembling side. This minimum time-to-live is
280 guaranteed if the pool size is less than inet_peer_threshold.
283 inet_peer_maxttl - INTEGER
284 Maximum time-to-live of entries. Unused entries will expire after
285 this period of time if there is no memory pressure on the pool (i.e.
286 when the number of entries in the pool is very small).
293 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
294 Defaults to 4096. (Was 128 before linux-5.4)
295 See also tcp_max_syn_backlog for additional tuning for TCP sockets.
297 tcp_abort_on_overflow - BOOLEAN
298 If listening service is too slow to accept new connections,
299 reset them. Default state is FALSE. It means that if overflow
300 occurred due to a burst, connection will recover. Enable this
301 option _only_ if you are really sure that listening daemon
302 cannot be tuned to accept connections faster. Enabling this
303 option can harm clients of your server.
305 tcp_adv_win_scale - INTEGER
306 Count buffering overhead as bytes/2^tcp_adv_win_scale
307 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
310 Possible values are [-31, 31], inclusive.
314 tcp_allowed_congestion_control - STRING
315 Show/set the congestion control choices available to non-privileged
316 processes. The list is a subset of those listed in
317 tcp_available_congestion_control.
319 Default is "reno" and the default setting (tcp_congestion_control).
321 tcp_app_win - INTEGER
322 Reserve max(window/2^tcp_app_win, mss) of window for application
323 buffer. Value 0 is special, it means that nothing is reserved.
327 tcp_autocorking - BOOLEAN
328 Enable TCP auto corking :
329 When applications do consecutive small write()/sendmsg() system calls,
330 we try to coalesce these small writes as much as possible, to lower
331 total amount of sent packets. This is done if at least one prior
332 packet for the flow is waiting in Qdisc queues or device transmit
333 queue. Applications can still use TCP_CORK for optimal behavior
334 when they know how/when to uncork their sockets.
338 tcp_available_congestion_control - STRING
339 Shows the available congestion control choices that are registered.
340 More congestion control algorithms may be available as modules,
343 tcp_base_mss - INTEGER
344 The initial value of search_low to be used by the packetization layer
345 Path MTU discovery (MTU probing). If MTU probing is enabled,
346 this is the initial MSS used by the connection.
348 tcp_mtu_probe_floor - INTEGER
349 If MTU probing is enabled this caps the minimum MSS used for search_low
354 tcp_min_snd_mss - INTEGER
355 TCP SYN and SYNACK messages usually advertise an ADVMSS option,
356 as described in RFC 1122 and RFC 6691.
358 If this ADVMSS option is smaller than tcp_min_snd_mss,
359 it is silently capped to tcp_min_snd_mss.
361 Default : 48 (at least 8 bytes of payload per segment)
363 tcp_congestion_control - STRING
364 Set the congestion control algorithm to be used for new
365 connections. The algorithm "reno" is always available, but
366 additional choices may be available based on kernel configuration.
367 Default is set as part of kernel configuration.
368 For passive connections, the listener congestion control choice
371 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
374 Allows TCP to send "duplicate" SACKs.
376 tcp_early_retrans - INTEGER
377 Tail loss probe (TLP) converts RTOs occurring due to tail
378 losses into fast recovery (draft-ietf-tcpm-rack). Note that
379 TLP requires RACK to function properly (see tcp_recovery below)
389 Control use of Explicit Congestion Notification (ECN) by TCP.
390 ECN is used only when both ends of the TCP connection indicate
391 support for it. This feature is useful in avoiding losses due
392 to congestion by allowing supporting routers to signal
393 congestion before having to drop packets.
397 = =====================================================
398 0 Disable ECN. Neither initiate nor accept ECN.
399 1 Enable ECN when requested by incoming connections and
400 also request ECN on outgoing connection attempts.
401 2 Enable ECN when requested by incoming connections
402 but do not request ECN on outgoing connections.
403 = =====================================================
407 tcp_ecn_fallback - BOOLEAN
408 If the kernel detects that ECN connection misbehaves, enable fall
409 back to non-ECN. Currently, this knob implements the fallback
410 from RFC3168, section 6.1.1.1., but we reserve that in future,
411 additional detection mechanisms could be implemented under this
412 knob. The value is not used, if tcp_ecn or per route (or congestion
413 control) ECN settings are disabled.
415 Default: 1 (fallback enabled)
418 This is a legacy option, it has no effect anymore.
420 tcp_fin_timeout - INTEGER
421 The length of time an orphaned (no longer referenced by any
422 application) connection will remain in the FIN_WAIT_2 state
423 before it is aborted at the local end. While a perfectly
424 valid "receive only" state for an un-orphaned connection, an
425 orphaned connection in FIN_WAIT_2 state could otherwise wait
426 forever for the remote to close its end of the connection.
433 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
434 F-RTO is an enhanced recovery algorithm for TCP retransmission
435 timeouts. It is particularly beneficial in networks where the
436 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
437 modification. It does not require any support from the peer.
439 By default it's enabled with a non-zero value. 0 disables F-RTO.
441 tcp_fwmark_accept - BOOLEAN
442 If set, incoming connections to listening sockets that do not have a
443 socket mark will set the mark of the accepting socket to the fwmark of
444 the incoming SYN packet. This will cause all packets on that connection
445 (starting from the first SYNACK) to be sent with that fwmark. The
446 listening socket's mark is unchanged. Listening sockets that already
447 have a fwmark set via setsockopt(SOL_SOCKET, SO_MARK, ...) are
452 tcp_invalid_ratelimit - INTEGER
453 Limit the maximal rate for sending duplicate acknowledgments
454 in response to incoming TCP packets that are for an existing
455 connection but that are invalid due to any of these reasons:
457 (a) out-of-window sequence number,
458 (b) out-of-window acknowledgment number, or
459 (c) PAWS (Protection Against Wrapped Sequence numbers) check failure
461 This can help mitigate simple "ack loop" DoS attacks, wherein
462 a buggy or malicious middlebox or man-in-the-middle can
463 rewrite TCP header fields in manner that causes each endpoint
464 to think that the other is sending invalid TCP segments, thus
465 causing each side to send an unterminating stream of duplicate
466 acknowledgments for invalid segments.
468 Using 0 disables rate-limiting of dupacks in response to
469 invalid segments; otherwise this value specifies the minimal
470 space between sending such dupacks, in milliseconds.
472 Default: 500 (milliseconds).
474 tcp_keepalive_time - INTEGER
475 How often TCP sends out keepalive messages when keepalive is enabled.
478 tcp_keepalive_probes - INTEGER
479 How many keepalive probes TCP sends out, until it decides that the
480 connection is broken. Default value: 9.
482 tcp_keepalive_intvl - INTEGER
483 How frequently the probes are send out. Multiplied by
484 tcp_keepalive_probes it is time to kill not responding connection,
485 after probes started. Default value: 75sec i.e. connection
486 will be aborted after ~11 minutes of retries.
488 tcp_l3mdev_accept - BOOLEAN
489 Enables child sockets to inherit the L3 master device index.
490 Enabling this option allows a "global" listen socket to work
491 across L3 master domains (e.g., VRFs) with connected sockets
492 derived from the listen socket to be bound to the L3 domain in
493 which the packets originated. Only valid when the kernel was
494 compiled with CONFIG_NET_L3_MASTER_DEV.
496 Default: 0 (disabled)
498 tcp_low_latency - BOOLEAN
499 This is a legacy option, it has no effect anymore.
501 tcp_max_orphans - INTEGER
502 Maximal number of TCP sockets not attached to any user file handle,
503 held by system. If this number is exceeded orphaned connections are
504 reset immediately and warning is printed. This limit exists
505 only to prevent simple DoS attacks, you _must_ not rely on this
506 or lower the limit artificially, but rather increase it
507 (probably, after increasing installed memory),
508 if network conditions require more than default value,
509 and tune network services to linger and kill such states
510 more aggressively. Let me to remind again: each orphan eats
511 up to ~64K of unswappable memory.
513 tcp_max_syn_backlog - INTEGER
514 Maximal number of remembered connection requests (SYN_RECV),
515 which have not received an acknowledgment from connecting client.
517 This is a per-listener limit.
519 The minimal value is 128 for low memory machines, and it will
520 increase in proportion to the memory of machine.
522 If server suffers from overload, try increasing this number.
524 Remember to also check /proc/sys/net/core/somaxconn
525 A SYN_RECV request socket consumes about 304 bytes of memory.
527 tcp_max_tw_buckets - INTEGER
528 Maximal number of timewait sockets held by system simultaneously.
529 If this number is exceeded time-wait socket is immediately destroyed
530 and warning is printed. This limit exists only to prevent
531 simple DoS attacks, you _must_ not lower the limit artificially,
532 but rather increase it (probably, after increasing installed memory),
533 if network conditions require more than default value.
535 tcp_mem - vector of 3 INTEGERs: min, pressure, max
536 min: below this number of pages TCP is not bothered about its
539 pressure: when amount of memory allocated by TCP exceeds this number
540 of pages, TCP moderates its memory consumption and enters memory
541 pressure mode, which is exited when memory consumption falls
544 max: number of pages allowed for queueing by all TCP sockets.
546 Defaults are calculated at boot time from amount of available
549 tcp_min_rtt_wlen - INTEGER
550 The window length of the windowed min filter to track the minimum RTT.
551 A shorter window lets a flow more quickly pick up new (higher)
552 minimum RTT when it is moved to a longer path (e.g., due to traffic
553 engineering). A longer window makes the filter more resistant to RTT
554 inflations such as transient congestion. The unit is seconds.
556 Possible values: 0 - 86400 (1 day)
560 tcp_moderate_rcvbuf - BOOLEAN
561 If set, TCP performs receive buffer auto-tuning, attempting to
562 automatically size the buffer (no greater than tcp_rmem[2]) to
563 match the size required by the path for full throughput. Enabled by
566 tcp_mtu_probing - INTEGER
567 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
571 - 1 - Disabled by default, enabled when an ICMP black hole detected
572 - 2 - Always enabled, use initial MSS of tcp_base_mss.
574 tcp_probe_interval - UNSIGNED INTEGER
575 Controls how often to start TCP Packetization-Layer Path MTU
576 Discovery reprobe. The default is reprobing every 10 minutes as
579 tcp_probe_threshold - INTEGER
580 Controls when TCP Packetization-Layer Path MTU Discovery probing
581 will stop in respect to the width of search range in bytes. Default
584 tcp_no_metrics_save - BOOLEAN
585 By default, TCP saves various connection metrics in the route cache
586 when the connection closes, so that connections established in the
587 near future can use these to set initial conditions. Usually, this
588 increases overall performance, but may sometimes cause performance
589 degradation. If set, TCP will not cache metrics on closing
592 tcp_no_ssthresh_metrics_save - BOOLEAN
593 Controls whether TCP saves ssthresh metrics in the route cache.
595 Default is 1, which disables ssthresh metrics.
597 tcp_orphan_retries - INTEGER
598 This value influences the timeout of a locally closed TCP connection,
599 when RTO retransmissions remain unacknowledged.
600 See tcp_retries2 for more details.
602 The default value is 8.
604 If your machine is a loaded WEB server,
605 you should think about lowering this value, such sockets
606 may consume significant resources. Cf. tcp_max_orphans.
608 tcp_recovery - INTEGER
609 This value is a bitmap to enable various experimental loss recovery
612 ========= =============================================================
613 RACK: 0x1 enables the RACK loss detection for fast detection of lost
614 retransmissions and tail drops. It also subsumes and disables
615 RFC6675 recovery for SACK connections.
617 RACK: 0x2 makes RACK's reordering window static (min_rtt/4).
619 RACK: 0x4 disables RACK's DUPACK threshold heuristic
620 ========= =============================================================
624 tcp_reordering - INTEGER
625 Initial reordering level of packets in a TCP stream.
626 TCP stack can then dynamically adjust flow reordering level
627 between this initial value and tcp_max_reordering
631 tcp_max_reordering - INTEGER
632 Maximal reordering level of packets in a TCP stream.
633 300 is a fairly conservative value, but you might increase it
634 if paths are using per packet load balancing (like bonding rr mode)
638 tcp_retrans_collapse - BOOLEAN
639 Bug-to-bug compatibility with some broken printers.
640 On retransmit try to send bigger packets to work around bugs in
643 tcp_retries1 - INTEGER
644 This value influences the time, after which TCP decides, that
645 something is wrong due to unacknowledged RTO retransmissions,
646 and reports this suspicion to the network layer.
647 See tcp_retries2 for more details.
649 RFC 1122 recommends at least 3 retransmissions, which is the
652 tcp_retries2 - INTEGER
653 This value influences the timeout of an alive TCP connection,
654 when RTO retransmissions remain unacknowledged.
655 Given a value of N, a hypothetical TCP connection following
656 exponential backoff with an initial RTO of TCP_RTO_MIN would
657 retransmit N times before killing the connection at the (N+1)th RTO.
659 The default value of 15 yields a hypothetical timeout of 924.6
660 seconds and is a lower bound for the effective timeout.
661 TCP will effectively time out at the first RTO which exceeds the
662 hypothetical timeout.
664 RFC 1122 recommends at least 100 seconds for the timeout,
665 which corresponds to a value of at least 8.
667 tcp_rfc1337 - BOOLEAN
668 If set, the TCP stack behaves conforming to RFC1337. If unset,
669 we are not conforming to RFC, but prevent TCP TIME_WAIT
674 tcp_rmem - vector of 3 INTEGERs: min, default, max
675 min: Minimal size of receive buffer used by TCP sockets.
676 It is guaranteed to each TCP socket, even under moderate memory
681 default: initial size of receive buffer used by TCP sockets.
682 This value overrides net.core.rmem_default used by other protocols.
683 Default: 131072 bytes.
684 This value results in initial window of 65535.
686 max: maximal size of receive buffer allowed for automatically
687 selected receiver buffers for TCP socket. This value does not override
688 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
689 automatic tuning of that socket's receive buffer size, in which
690 case this value is ignored.
691 Default: between 131072 and 6MB, depending on RAM size.
694 Enable select acknowledgments (SACKS).
696 tcp_comp_sack_delay_ns - LONG INTEGER
697 TCP tries to reduce number of SACK sent, using a timer
698 based on 5% of SRTT, capped by this sysctl, in nano seconds.
699 The default is 1ms, based on TSO autosizing period.
701 Default : 1,000,000 ns (1 ms)
703 tcp_comp_sack_slack_ns - LONG INTEGER
704 This sysctl control the slack used when arming the
705 timer used by SACK compression. This gives extra time
706 for small RTT flows, and reduces system overhead by allowing
707 opportunistic reduction of timer interrupts.
709 Default : 100,000 ns (100 us)
711 tcp_comp_sack_nr - INTEGER
712 Max number of SACK that can be compressed.
713 Using 0 disables SACK compression.
717 tcp_slow_start_after_idle - BOOLEAN
718 If set, provide RFC2861 behavior and time out the congestion
719 window after an idle period. An idle period is defined at
720 the current RTO. If unset, the congestion window will not
721 be timed out after an idle period.
726 Use the Host requirements interpretation of the TCP urgent pointer field.
727 Most hosts use the older BSD interpretation, so if you turn this on
728 Linux might not communicate correctly with them.
732 tcp_synack_retries - INTEGER
733 Number of times SYNACKs for a passive TCP connection attempt will
734 be retransmitted. Should not be higher than 255. Default value
735 is 5, which corresponds to 31seconds till the last retransmission
736 with the current initial RTO of 1second. With this the final timeout
737 for a passive TCP connection will happen after 63seconds.
739 tcp_syncookies - INTEGER
740 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
741 Send out syncookies when the syn backlog queue of a socket
742 overflows. This is to prevent against the common 'SYN flood attack'
745 Note, that syncookies is fallback facility.
746 It MUST NOT be used to help highly loaded servers to stand
747 against legal connection rate. If you see SYN flood warnings
748 in your logs, but investigation shows that they occur
749 because of overload with legal connections, you should tune
750 another parameters until this warning disappear.
751 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
753 syncookies seriously violate TCP protocol, do not allow
754 to use TCP extensions, can result in serious degradation
755 of some services (f.e. SMTP relaying), visible not by you,
756 but your clients and relays, contacting you. While you see
757 SYN flood warnings in logs not being really flooded, your server
758 is seriously misconfigured.
760 If you want to test which effects syncookies have to your
761 network connections you can set this knob to 2 to enable
762 unconditionally generation of syncookies.
764 tcp_migrate_req - BOOLEAN
765 The incoming connection is tied to a specific listening socket when
766 the initial SYN packet is received during the three-way handshake.
767 When a listener is closed, in-flight request sockets during the
768 handshake and established sockets in the accept queue are aborted.
770 If the listener has SO_REUSEPORT enabled, other listeners on the
771 same port should have been able to accept such connections. This
772 option makes it possible to migrate such child sockets to another
773 listener after close() or shutdown().
775 The BPF_SK_REUSEPORT_SELECT_OR_MIGRATE type of eBPF program should
776 usually be used to define the policy to pick an alive listener.
777 Otherwise, the kernel will randomly pick an alive listener only if
778 this option is enabled.
780 Note that migration between listeners with different settings may
781 crash applications. Let's say migration happens from listener A to
782 B, and only B has TCP_SAVE_SYN enabled. B cannot read SYN data from
783 the requests migrated from A. To avoid such a situation, cancel
784 migration by returning SK_DROP in the type of eBPF program, or
789 tcp_fastopen - INTEGER
790 Enable TCP Fast Open (RFC7413) to send and accept data in the opening
793 The client support is enabled by flag 0x1 (on by default). The client
794 then must use sendmsg() or sendto() with the MSG_FASTOPEN flag,
795 rather than connect() to send data in SYN.
797 The server support is enabled by flag 0x2 (off by default). Then
798 either enable for all listeners with another flag (0x400) or
799 enable individual listeners via TCP_FASTOPEN socket option with
800 the option value being the length of the syn-data backlog.
802 The values (bitmap) are
804 ===== ======== ======================================================
805 0x1 (client) enables sending data in the opening SYN on the client.
806 0x2 (server) enables the server support, i.e., allowing data in
807 a SYN packet to be accepted and passed to the
808 application before 3-way handshake finishes.
809 0x4 (client) send data in the opening SYN regardless of cookie
810 availability and without a cookie option.
811 0x200 (server) accept data-in-SYN w/o any cookie option present.
812 0x400 (server) enable all listeners to support Fast Open by
813 default without explicit TCP_FASTOPEN socket option.
814 ===== ======== ======================================================
818 Note that additional client or server features are only
819 effective if the basic support (0x1 and 0x2) are enabled respectively.
821 tcp_fastopen_blackhole_timeout_sec - INTEGER
822 Initial time period in second to disable Fastopen on active TCP sockets
823 when a TFO firewall blackhole issue happens.
824 This time period will grow exponentially when more blackhole issues
825 get detected right after Fastopen is re-enabled and will reset to
826 initial value when the blackhole issue goes away.
827 0 to disable the blackhole detection.
829 By default, it is set to 0 (feature is disabled).
831 tcp_fastopen_key - list of comma separated 32-digit hexadecimal INTEGERs
832 The list consists of a primary key and an optional backup key. The
833 primary key is used for both creating and validating cookies, while the
834 optional backup key is only used for validating cookies. The purpose of
835 the backup key is to maximize TFO validation when keys are rotated.
837 A randomly chosen primary key may be configured by the kernel if
838 the tcp_fastopen sysctl is set to 0x400 (see above), or if the
839 TCP_FASTOPEN setsockopt() optname is set and a key has not been
840 previously configured via sysctl. If keys are configured via
841 setsockopt() by using the TCP_FASTOPEN_KEY optname, then those
842 per-socket keys will be used instead of any keys that are specified via
845 A key is specified as 4 8-digit hexadecimal integers which are separated
846 by a '-' as: xxxxxxxx-xxxxxxxx-xxxxxxxx-xxxxxxxx. Leading zeros may be
847 omitted. A primary and a backup key may be specified by separating them
848 by a comma. If only one key is specified, it becomes the primary key and
849 any previously configured backup keys are removed.
851 tcp_syn_retries - INTEGER
852 Number of times initial SYNs for an active TCP connection attempt
853 will be retransmitted. Should not be higher than 127. Default value
854 is 6, which corresponds to 63seconds till the last retransmission
855 with the current initial RTO of 1second. With this the final timeout
856 for an active TCP connection attempt will happen after 127seconds.
858 tcp_timestamps - INTEGER
859 Enable timestamps as defined in RFC1323.
862 - 1: Enable timestamps as defined in RFC1323 and use random offset for
863 each connection rather than only using the current time.
864 - 2: Like 1, but without random offsets.
868 tcp_min_tso_segs - INTEGER
869 Minimal number of segments per TSO frame.
871 Since linux-3.12, TCP does an automatic sizing of TSO frames,
872 depending on flow rate, instead of filling 64Kbytes packets.
873 For specific usages, it's possible to force TCP to build big
874 TSO frames. Note that TCP stack might split too big TSO packets
875 if available window is too small.
879 tcp_pacing_ss_ratio - INTEGER
880 sk->sk_pacing_rate is set by TCP stack using a ratio applied
881 to current rate. (current_rate = cwnd * mss / srtt)
882 If TCP is in slow start, tcp_pacing_ss_ratio is applied
883 to let TCP probe for bigger speeds, assuming cwnd can be
884 doubled every other RTT.
888 tcp_pacing_ca_ratio - INTEGER
889 sk->sk_pacing_rate is set by TCP stack using a ratio applied
890 to current rate. (current_rate = cwnd * mss / srtt)
891 If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio
892 is applied to conservatively probe for bigger throughput.
896 tcp_tso_win_divisor - INTEGER
897 This allows control over what percentage of the congestion window
898 can be consumed by a single TSO frame.
899 The setting of this parameter is a choice between burstiness and
900 building larger TSO frames.
904 tcp_tw_reuse - INTEGER
905 Enable reuse of TIME-WAIT sockets for new connections when it is
906 safe from protocol viewpoint.
910 - 2 - enable for loopback traffic only
912 It should not be changed without advice/request of technical
917 tcp_window_scaling - BOOLEAN
918 Enable window scaling as defined in RFC1323.
920 tcp_wmem - vector of 3 INTEGERs: min, default, max
921 min: Amount of memory reserved for send buffers for TCP sockets.
922 Each TCP socket has rights to use it due to fact of its birth.
926 default: initial size of send buffer used by TCP sockets. This
927 value overrides net.core.wmem_default used by other protocols.
929 It is usually lower than net.core.wmem_default.
933 max: Maximal amount of memory allowed for automatically tuned
934 send buffers for TCP sockets. This value does not override
935 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
936 automatic tuning of that socket's send buffer size, in which case
937 this value is ignored.
939 Default: between 64K and 4MB, depending on RAM size.
941 tcp_notsent_lowat - UNSIGNED INTEGER
942 A TCP socket can control the amount of unsent bytes in its write queue,
943 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
944 reports POLLOUT events if the amount of unsent bytes is below a per
945 socket value, and if the write queue is not full. sendmsg() will
946 also not add new buffers if the limit is hit.
948 This global variable controls the amount of unsent data for
949 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
950 to the global variable has immediate effect.
952 Default: UINT_MAX (0xFFFFFFFF)
954 tcp_workaround_signed_windows - BOOLEAN
955 If set, assume no receipt of a window scaling option means the
956 remote TCP is broken and treats the window as a signed quantity.
957 If unset, assume the remote TCP is not broken even if we do
958 not receive a window scaling option from them.
962 tcp_thin_linear_timeouts - BOOLEAN
963 Enable dynamic triggering of linear timeouts for thin streams.
964 If set, a check is performed upon retransmission by timeout to
965 determine if the stream is thin (less than 4 packets in flight).
966 As long as the stream is found to be thin, up to 6 linear
967 timeouts may be performed before exponential backoff mode is
968 initiated. This improves retransmission latency for
969 non-aggressive thin streams, often found to be time-dependent.
970 For more information on thin streams, see
971 Documentation/networking/tcp-thin.rst
975 tcp_limit_output_bytes - INTEGER
976 Controls TCP Small Queue limit per tcp socket.
977 TCP bulk sender tends to increase packets in flight until it
978 gets losses notifications. With SNDBUF autotuning, this can
979 result in a large amount of packets queued on the local machine
980 (e.g.: qdiscs, CPU backlog, or device) hurting latency of other
981 flows, for typical pfifo_fast qdiscs. tcp_limit_output_bytes
982 limits the number of bytes on qdisc or device to reduce artificial
983 RTT/cwnd and reduce bufferbloat.
985 Default: 1048576 (16 * 65536)
987 tcp_challenge_ack_limit - INTEGER
988 Limits number of Challenge ACK sent per second, as recommended
989 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
995 udp_l3mdev_accept - BOOLEAN
996 Enabling this option allows a "global" bound socket to work
997 across L3 master domains (e.g., VRFs) with packets capable of
998 being received regardless of the L3 domain in which they
999 originated. Only valid when the kernel was compiled with
1000 CONFIG_NET_L3_MASTER_DEV.
1002 Default: 0 (disabled)
1004 udp_mem - vector of 3 INTEGERs: min, pressure, max
1005 Number of pages allowed for queueing by all UDP sockets.
1007 min: Number of pages allowed for queueing by all UDP sockets.
1009 pressure: This value was introduced to follow format of tcp_mem.
1011 max: This value was introduced to follow format of tcp_mem.
1013 Default is calculated at boot time from amount of available memory.
1015 udp_rmem_min - INTEGER
1016 Minimal size of receive buffer used by UDP sockets in moderation.
1017 Each UDP socket is able to use the size for receiving data, even if
1018 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
1022 udp_wmem_min - INTEGER
1023 Minimal size of send buffer used by UDP sockets in moderation.
1024 Each UDP socket is able to use the size for sending data, even if
1025 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
1032 raw_l3mdev_accept - BOOLEAN
1033 Enabling this option allows a "global" bound socket to work
1034 across L3 master domains (e.g., VRFs) with packets capable of
1035 being received regardless of the L3 domain in which they
1036 originated. Only valid when the kernel was compiled with
1037 CONFIG_NET_L3_MASTER_DEV.
1039 Default: 1 (enabled)
1044 cipso_cache_enable - BOOLEAN
1045 If set, enable additions to and lookups from the CIPSO label mapping
1046 cache. If unset, additions are ignored and lookups always result in a
1047 miss. However, regardless of the setting the cache is still
1048 invalidated when required when means you can safely toggle this on and
1049 off and the cache will always be "safe".
1053 cipso_cache_bucket_size - INTEGER
1054 The CIPSO label cache consists of a fixed size hash table with each
1055 hash bucket containing a number of cache entries. This variable limits
1056 the number of entries in each hash bucket; the larger the value the
1057 more CIPSO label mappings that can be cached. When the number of
1058 entries in a given hash bucket reaches this limit adding new entries
1059 causes the oldest entry in the bucket to be removed to make room.
1063 cipso_rbm_optfmt - BOOLEAN
1064 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
1065 the CIPSO draft specification (see Documentation/netlabel for details).
1066 This means that when set the CIPSO tag will be padded with empty
1067 categories in order to make the packet data 32-bit aligned.
1071 cipso_rbm_structvalid - BOOLEAN
1072 If set, do a very strict check of the CIPSO option when
1073 ip_options_compile() is called. If unset, relax the checks done during
1074 ip_options_compile(). Either way is "safe" as errors are caught else
1075 where in the CIPSO processing code but setting this to 0 (False) should
1076 result in less work (i.e. it should be faster) but could cause problems
1077 with other implementations that require strict checking.
1084 ip_local_port_range - 2 INTEGERS
1085 Defines the local port range that is used by TCP and UDP to
1086 choose the local port. The first number is the first, the
1087 second the last local port number.
1088 If possible, it is better these numbers have different parity
1089 (one even and one odd value).
1090 Must be greater than or equal to ip_unprivileged_port_start.
1091 The default values are 32768 and 60999 respectively.
1093 ip_local_reserved_ports - list of comma separated ranges
1094 Specify the ports which are reserved for known third-party
1095 applications. These ports will not be used by automatic port
1096 assignments (e.g. when calling connect() or bind() with port
1097 number 0). Explicit port allocation behavior is unchanged.
1099 The format used for both input and output is a comma separated
1100 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
1101 10). Writing to the file will clear all previously reserved
1102 ports and update the current list with the one given in the
1105 Note that ip_local_port_range and ip_local_reserved_ports
1106 settings are independent and both are considered by the kernel
1107 when determining which ports are available for automatic port
1110 You can reserve ports which are not in the current
1111 ip_local_port_range, e.g.::
1113 $ cat /proc/sys/net/ipv4/ip_local_port_range
1115 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
1118 although this is redundant. However such a setting is useful
1119 if later the port range is changed to a value that will
1120 include the reserved ports. Also keep in mind, that overlapping
1121 of these ranges may affect probability of selecting ephemeral
1122 ports which are right after block of reserved ports.
1126 ip_unprivileged_port_start - INTEGER
1127 This is a per-namespace sysctl. It defines the first
1128 unprivileged port in the network namespace. Privileged ports
1129 require root or CAP_NET_BIND_SERVICE in order to bind to them.
1130 To disable all privileged ports, set this to 0. They must not
1131 overlap with the ip_local_port_range.
1135 ip_nonlocal_bind - BOOLEAN
1136 If set, allows processes to bind() to non-local IP addresses,
1137 which can be quite useful - but may break some applications.
1141 ip_autobind_reuse - BOOLEAN
1142 By default, bind() does not select the ports automatically even if
1143 the new socket and all sockets bound to the port have SO_REUSEADDR.
1144 ip_autobind_reuse allows bind() to reuse the port and this is useful
1145 when you use bind()+connect(), but may break some applications.
1146 The preferred solution is to use IP_BIND_ADDRESS_NO_PORT and this
1147 option should only be set by experts.
1150 ip_dynaddr - BOOLEAN
1151 If set non-zero, enables support for dynamic addresses.
1152 If set to a non-zero value larger than 1, a kernel log
1153 message will be printed when dynamic address rewriting
1158 ip_early_demux - BOOLEAN
1159 Optimize input packet processing down to one demux for
1160 certain kinds of local sockets. Currently we only do this
1161 for established TCP and connected UDP sockets.
1163 It may add an additional cost for pure routing workloads that
1164 reduces overall throughput, in such case you should disable it.
1168 ping_group_range - 2 INTEGERS
1169 Restrict ICMP_PROTO datagram sockets to users in the group range.
1170 The default is "1 0", meaning, that nobody (not even root) may
1171 create ping sockets. Setting it to "100 100" would grant permissions
1172 to the single group. "0 4294967295" would enable it for the world, "100
1173 4294967295" would enable it for the users, but not daemons.
1175 tcp_early_demux - BOOLEAN
1176 Enable early demux for established TCP sockets.
1180 udp_early_demux - BOOLEAN
1181 Enable early demux for connected UDP sockets. Disable this if
1182 your system could experience more unconnected load.
1186 icmp_echo_ignore_all - BOOLEAN
1187 If set non-zero, then the kernel will ignore all ICMP ECHO
1188 requests sent to it.
1192 icmp_echo_enable_probe - BOOLEAN
1193 If set to one, then the kernel will respond to RFC 8335 PROBE
1194 requests sent to it.
1198 icmp_echo_ignore_broadcasts - BOOLEAN
1199 If set non-zero, then the kernel will ignore all ICMP ECHO and
1200 TIMESTAMP requests sent to it via broadcast/multicast.
1204 icmp_ratelimit - INTEGER
1205 Limit the maximal rates for sending ICMP packets whose type matches
1206 icmp_ratemask (see below) to specific targets.
1207 0 to disable any limiting,
1208 otherwise the minimal space between responses in milliseconds.
1209 Note that another sysctl, icmp_msgs_per_sec limits the number
1210 of ICMP packets sent on all targets.
1214 icmp_msgs_per_sec - INTEGER
1215 Limit maximal number of ICMP packets sent per second from this host.
1216 Only messages whose type matches icmp_ratemask (see below) are
1217 controlled by this limit. For security reasons, the precise count
1218 of messages per second is randomized.
1222 icmp_msgs_burst - INTEGER
1223 icmp_msgs_per_sec controls number of ICMP packets sent per second,
1224 while icmp_msgs_burst controls the burst size of these packets.
1225 For security reasons, the precise burst size is randomized.
1229 icmp_ratemask - INTEGER
1230 Mask made of ICMP types for which rates are being limited.
1232 Significant bits: IHGFEDCBA9876543210
1234 Default mask: 0000001100000011000 (6168)
1236 Bit definitions (see include/linux/icmp.h):
1238 = =========================
1240 3 Destination Unreachable [1]_
1241 4 Source Quench [1]_
1244 B Time Exceeded [1]_
1245 C Parameter Problem [1]_
1250 H Address Mask Request
1251 I Address Mask Reply
1252 = =========================
1254 .. [1] These are rate limited by default (see default mask above)
1256 icmp_ignore_bogus_error_responses - BOOLEAN
1257 Some routers violate RFC1122 by sending bogus responses to broadcast
1258 frames. Such violations are normally logged via a kernel warning.
1259 If this is set to TRUE, the kernel will not give such warnings, which
1260 will avoid log file clutter.
1264 icmp_errors_use_inbound_ifaddr - BOOLEAN
1266 If zero, icmp error messages are sent with the primary address of
1267 the exiting interface.
1269 If non-zero, the message will be sent with the primary address of
1270 the interface that received the packet that caused the icmp error.
1271 This is the behaviour many network administrators will expect from
1272 a router. And it can make debugging complicated network layouts
1275 Note that if no primary address exists for the interface selected,
1276 then the primary address of the first non-loopback interface that
1277 has one will be used regardless of this setting.
1281 igmp_max_memberships - INTEGER
1282 Change the maximum number of multicast groups we can subscribe to.
1285 Theoretical maximum value is bounded by having to send a membership
1286 report in a single datagram (i.e. the report can't span multiple
1287 datagrams, or risk confusing the switch and leaving groups you don't
1290 The number of supported groups 'M' is bounded by the number of group
1291 report entries you can fit into a single datagram of 65535 bytes.
1293 M = 65536-sizeof (ip header)/(sizeof(Group record))
1295 Group records are variable length, with a minimum of 12 bytes.
1296 So net.ipv4.igmp_max_memberships should not be set higher than:
1298 (65536-24) / 12 = 5459
1300 The value 5459 assumes no IP header options, so in practice
1301 this number may be lower.
1303 igmp_max_msf - INTEGER
1304 Maximum number of addresses allowed in the source filter list for a
1310 Controls the IGMP query robustness variable (see RFC2236 8.1).
1312 Default: 2 (as specified by RFC2236 8.1)
1314 Minimum: 1 (as specified by RFC6636 4.5)
1316 force_igmp_version - INTEGER
1317 - 0 - (default) No enforcement of a IGMP version, IGMPv1/v2 fallback
1318 allowed. Will back to IGMPv3 mode again if all IGMPv1/v2 Querier
1319 Present timer expires.
1320 - 1 - Enforce to use IGMP version 1. Will also reply IGMPv1 report if
1321 receive IGMPv2/v3 query.
1322 - 2 - Enforce to use IGMP version 2. Will fallback to IGMPv1 if receive
1323 IGMPv1 query message. Will reply report if receive IGMPv3 query.
1324 - 3 - Enforce to use IGMP version 3. The same react with default 0.
1328 this is not the same with force_mld_version because IGMPv3 RFC3376
1329 Security Considerations does not have clear description that we could
1330 ignore other version messages completely as MLDv2 RFC3810. So make
1331 this value as default 0 is recommended.
1333 ``conf/interface/*``
1334 changes special settings per interface (where
1335 interface" is the name of your network interface)
1338 is special, changes the settings for all interfaces
1340 log_martians - BOOLEAN
1341 Log packets with impossible addresses to kernel log.
1342 log_martians for the interface will be enabled if at least one of
1343 conf/{all,interface}/log_martians is set to TRUE,
1344 it will be disabled otherwise
1346 accept_redirects - BOOLEAN
1347 Accept ICMP redirect messages.
1348 accept_redirects for the interface will be enabled if:
1350 - both conf/{all,interface}/accept_redirects are TRUE in the case
1351 forwarding for the interface is enabled
1355 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
1356 case forwarding for the interface is disabled
1358 accept_redirects for the interface will be disabled otherwise
1365 forwarding - BOOLEAN
1366 Enable IP forwarding on this interface. This controls whether packets
1367 received _on_ this interface can be forwarded.
1369 mc_forwarding - BOOLEAN
1370 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
1371 and a multicast routing daemon is required.
1372 conf/all/mc_forwarding must also be set to TRUE to enable multicast
1373 routing for the interface
1376 Integer value used to differentiate the devices by the medium they
1377 are attached to. Two devices can have different id values when
1378 the broadcast packets are received only on one of them.
1379 The default value 0 means that the device is the only interface
1380 to its medium, value of -1 means that medium is not known.
1382 Currently, it is used to change the proxy_arp behavior:
1383 the proxy_arp feature is enabled for packets forwarded between
1384 two devices attached to different media.
1389 proxy_arp for the interface will be enabled if at least one of
1390 conf/{all,interface}/proxy_arp is set to TRUE,
1391 it will be disabled otherwise
1393 proxy_arp_pvlan - BOOLEAN
1394 Private VLAN proxy arp.
1396 Basically allow proxy arp replies back to the same interface
1397 (from which the ARP request/solicitation was received).
1399 This is done to support (ethernet) switch features, like RFC
1400 3069, where the individual ports are NOT allowed to
1401 communicate with each other, but they are allowed to talk to
1402 the upstream router. As described in RFC 3069, it is possible
1403 to allow these hosts to communicate through the upstream
1404 router by proxy_arp'ing. Don't need to be used together with
1407 This technology is known by different names:
1409 In RFC 3069 it is called VLAN Aggregation.
1410 Cisco and Allied Telesyn call it Private VLAN.
1411 Hewlett-Packard call it Source-Port filtering or port-isolation.
1412 Ericsson call it MAC-Forced Forwarding (RFC Draft).
1414 shared_media - BOOLEAN
1415 Send(router) or accept(host) RFC1620 shared media redirects.
1416 Overrides secure_redirects.
1418 shared_media for the interface will be enabled if at least one of
1419 conf/{all,interface}/shared_media is set to TRUE,
1420 it will be disabled otherwise
1424 secure_redirects - BOOLEAN
1425 Accept ICMP redirect messages only to gateways listed in the
1426 interface's current gateway list. Even if disabled, RFC1122 redirect
1429 Overridden by shared_media.
1431 secure_redirects for the interface will be enabled if at least one of
1432 conf/{all,interface}/secure_redirects is set to TRUE,
1433 it will be disabled otherwise
1437 send_redirects - BOOLEAN
1438 Send redirects, if router.
1440 send_redirects for the interface will be enabled if at least one of
1441 conf/{all,interface}/send_redirects is set to TRUE,
1442 it will be disabled otherwise
1446 bootp_relay - BOOLEAN
1447 Accept packets with source address 0.b.c.d destined
1448 not to this host as local ones. It is supposed, that
1449 BOOTP relay daemon will catch and forward such packets.
1450 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
1455 Not Implemented Yet.
1457 accept_source_route - BOOLEAN
1458 Accept packets with SRR option.
1459 conf/all/accept_source_route must also be set to TRUE to accept packets
1460 with SRR option on the interface
1467 accept_local - BOOLEAN
1468 Accept packets with local source addresses. In combination with
1469 suitable routing, this can be used to direct packets between two
1470 local interfaces over the wire and have them accepted properly.
1473 route_localnet - BOOLEAN
1474 Do not consider loopback addresses as martian source or destination
1475 while routing. This enables the use of 127/8 for local routing purposes.
1480 - 0 - No source validation.
1481 - 1 - Strict mode as defined in RFC3704 Strict Reverse Path
1482 Each incoming packet is tested against the FIB and if the interface
1483 is not the best reverse path the packet check will fail.
1484 By default failed packets are discarded.
1485 - 2 - Loose mode as defined in RFC3704 Loose Reverse Path
1486 Each incoming packet's source address is also tested against the FIB
1487 and if the source address is not reachable via any interface
1488 the packet check will fail.
1490 Current recommended practice in RFC3704 is to enable strict mode
1491 to prevent IP spoofing from DDos attacks. If using asymmetric routing
1492 or other complicated routing, then loose mode is recommended.
1494 The max value from conf/{all,interface}/rp_filter is used
1495 when doing source validation on the {interface}.
1497 Default value is 0. Note that some distributions enable it
1500 src_valid_mark - BOOLEAN
1501 - 0 - The fwmark of the packet is not included in reverse path
1502 route lookup. This allows for asymmetric routing configurations
1503 utilizing the fwmark in only one direction, e.g., transparent
1506 - 1 - The fwmark of the packet is included in reverse path route
1507 lookup. This permits rp_filter to function when the fwmark is
1508 used for routing traffic in both directions.
1510 This setting also affects the utilization of fmwark when
1511 performing source address selection for ICMP replies, or
1512 determining addresses stored for the IPOPT_TS_TSANDADDR and
1513 IPOPT_RR IP options.
1515 The max value from conf/{all,interface}/src_valid_mark is used.
1519 arp_filter - BOOLEAN
1520 - 1 - Allows you to have multiple network interfaces on the same
1521 subnet, and have the ARPs for each interface be answered
1522 based on whether or not the kernel would route a packet from
1523 the ARP'd IP out that interface (therefore you must use source
1524 based routing for this to work). In other words it allows control
1525 of which cards (usually 1) will respond to an arp request.
1527 - 0 - (default) The kernel can respond to arp requests with addresses
1528 from other interfaces. This may seem wrong but it usually makes
1529 sense, because it increases the chance of successful communication.
1530 IP addresses are owned by the complete host on Linux, not by
1531 particular interfaces. Only for more complex setups like load-
1532 balancing, does this behaviour cause problems.
1534 arp_filter for the interface will be enabled if at least one of
1535 conf/{all,interface}/arp_filter is set to TRUE,
1536 it will be disabled otherwise
1538 arp_announce - INTEGER
1539 Define different restriction levels for announcing the local
1540 source IP address from IP packets in ARP requests sent on
1543 - 0 - (default) Use any local address, configured on any interface
1544 - 1 - Try to avoid local addresses that are not in the target's
1545 subnet for this interface. This mode is useful when target
1546 hosts reachable via this interface require the source IP
1547 address in ARP requests to be part of their logical network
1548 configured on the receiving interface. When we generate the
1549 request we will check all our subnets that include the
1550 target IP and will preserve the source address if it is from
1551 such subnet. If there is no such subnet we select source
1552 address according to the rules for level 2.
1553 - 2 - Always use the best local address for this target.
1554 In this mode we ignore the source address in the IP packet
1555 and try to select local address that we prefer for talks with
1556 the target host. Such local address is selected by looking
1557 for primary IP addresses on all our subnets on the outgoing
1558 interface that include the target IP address. If no suitable
1559 local address is found we select the first local address
1560 we have on the outgoing interface or on all other interfaces,
1561 with the hope we will receive reply for our request and
1562 even sometimes no matter the source IP address we announce.
1564 The max value from conf/{all,interface}/arp_announce is used.
1566 Increasing the restriction level gives more chance for
1567 receiving answer from the resolved target while decreasing
1568 the level announces more valid sender's information.
1570 arp_ignore - INTEGER
1571 Define different modes for sending replies in response to
1572 received ARP requests that resolve local target IP addresses:
1574 - 0 - (default): reply for any local target IP address, configured
1576 - 1 - reply only if the target IP address is local address
1577 configured on the incoming interface
1578 - 2 - reply only if the target IP address is local address
1579 configured on the incoming interface and both with the
1580 sender's IP address are part from same subnet on this interface
1581 - 3 - do not reply for local addresses configured with scope host,
1582 only resolutions for global and link addresses are replied
1584 - 8 - do not reply for all local addresses
1586 The max value from conf/{all,interface}/arp_ignore is used
1587 when ARP request is received on the {interface}
1589 arp_notify - BOOLEAN
1590 Define mode for notification of address and device changes.
1592 == ==========================================================
1593 0 (default): do nothing
1594 1 Generate gratuitous arp requests when device is brought up
1595 or hardware address changes.
1596 == ==========================================================
1598 arp_accept - BOOLEAN
1599 Define behavior for gratuitous ARP frames who's IP is not
1600 already present in the ARP table:
1602 - 0 - don't create new entries in the ARP table
1603 - 1 - create new entries in the ARP table
1605 Both replies and requests type gratuitous arp will trigger the
1606 ARP table to be updated, if this setting is on.
1608 If the ARP table already contains the IP address of the
1609 gratuitous arp frame, the arp table will be updated regardless
1610 if this setting is on or off.
1612 arp_evict_nocarrier - BOOLEAN
1613 Clears the ARP cache on NOCARRIER events. This option is important for
1614 wireless devices where the ARP cache should not be cleared when roaming
1615 between access points on the same network. In most cases this should
1616 remain as the default (1).
1618 - 1 - (default): Clear the ARP cache on NOCARRIER events
1619 - 0 - Do not clear ARP cache on NOCARRIER events
1621 mcast_solicit - INTEGER
1622 The maximum number of multicast probes in INCOMPLETE state,
1623 when the associated hardware address is unknown. Defaults
1626 ucast_solicit - INTEGER
1627 The maximum number of unicast probes in PROBE state, when
1628 the hardware address is being reconfirmed. Defaults to 3.
1630 app_solicit - INTEGER
1631 The maximum number of probes to send to the user space ARP daemon
1632 via netlink before dropping back to multicast probes (see
1633 mcast_resolicit). Defaults to 0.
1635 mcast_resolicit - INTEGER
1636 The maximum number of multicast probes after unicast and
1637 app probes in PROBE state. Defaults to 0.
1639 disable_policy - BOOLEAN
1640 Disable IPSEC policy (SPD) for this interface
1642 disable_xfrm - BOOLEAN
1643 Disable IPSEC encryption on this interface, whatever the policy
1645 igmpv2_unsolicited_report_interval - INTEGER
1646 The interval in milliseconds in which the next unsolicited
1647 IGMPv1 or IGMPv2 report retransmit will take place.
1649 Default: 10000 (10 seconds)
1651 igmpv3_unsolicited_report_interval - INTEGER
1652 The interval in milliseconds in which the next unsolicited
1653 IGMPv3 report retransmit will take place.
1655 Default: 1000 (1 seconds)
1657 ignore_routes_with_linkdown - BOOLEAN
1658 Ignore routes whose link is down when performing a FIB lookup.
1660 promote_secondaries - BOOLEAN
1661 When a primary IP address is removed from this interface
1662 promote a corresponding secondary IP address instead of
1663 removing all the corresponding secondary IP addresses.
1665 drop_unicast_in_l2_multicast - BOOLEAN
1666 Drop any unicast IP packets that are received in link-layer
1667 multicast (or broadcast) frames.
1669 This behavior (for multicast) is actually a SHOULD in RFC
1670 1122, but is disabled by default for compatibility reasons.
1674 drop_gratuitous_arp - BOOLEAN
1675 Drop all gratuitous ARP frames, for example if there's a known
1676 good ARP proxy on the network and such frames need not be used
1677 (or in the case of 802.11, must not be used to prevent attacks.)
1683 Allows you to write a number, which can be used as required.
1687 xfrm4_gc_thresh - INTEGER
1688 (Obsolete since linux-4.14)
1689 The threshold at which we will start garbage collecting for IPv4
1690 destination cache entries. At twice this value the system will
1691 refuse new allocations.
1693 igmp_link_local_mcast_reports - BOOLEAN
1694 Enable IGMP reports for link local multicast groups in the
1700 kuznet@ms2.inr.ac.ru
1707 delon.nicolas@wanadoo.fr
1712 /proc/sys/net/ipv6/* Variables
1713 ==============================
1715 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1716 apply to IPv6 [XXX?].
1718 bindv6only - BOOLEAN
1719 Default value for IPV6_V6ONLY socket option,
1720 which restricts use of the IPv6 socket to IPv6 communication
1723 - TRUE: disable IPv4-mapped address feature
1724 - FALSE: enable IPv4-mapped address feature
1726 Default: FALSE (as specified in RFC3493)
1728 flowlabel_consistency - BOOLEAN
1729 Protect the consistency (and unicity) of flow label.
1730 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1738 auto_flowlabels - INTEGER
1739 Automatically generate flow labels based on a flow hash of the
1740 packet. This allows intermediate devices, such as routers, to
1741 identify packet flows for mechanisms like Equal Cost Multipath
1742 Routing (see RFC 6438).
1744 = ===========================================================
1745 0 automatic flow labels are completely disabled
1746 1 automatic flow labels are enabled by default, they can be
1747 disabled on a per socket basis using the IPV6_AUTOFLOWLABEL
1749 2 automatic flow labels are allowed, they may be enabled on a
1750 per socket basis using the IPV6_AUTOFLOWLABEL socket option
1751 3 automatic flow labels are enabled and enforced, they cannot
1752 be disabled by the socket option
1753 = ===========================================================
1757 flowlabel_state_ranges - BOOLEAN
1758 Split the flow label number space into two ranges. 0-0x7FFFF is
1759 reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF
1760 is reserved for stateless flow labels as described in RFC6437.
1767 flowlabel_reflect - INTEGER
1768 Control flow label reflection. Needed for Path MTU
1769 Discovery to work with Equal Cost Multipath Routing in anycast
1770 environments. See RFC 7690 and:
1771 https://tools.ietf.org/html/draft-wang-6man-flow-label-reflection-01
1775 - 1: enabled for established flows
1777 Note that this prevents automatic flowlabel changes, as done
1778 in "tcp: change IPv6 flow-label upon receiving spurious retransmission"
1779 and "tcp: Change txhash on every SYN and RTO retransmit"
1781 - 2: enabled for TCP RESET packets (no active listener)
1782 If set, a RST packet sent in response to a SYN packet on a closed
1783 port will reflect the incoming flow label.
1785 - 4: enabled for ICMPv6 echo reply messages.
1789 fib_multipath_hash_policy - INTEGER
1790 Controls which hash policy to use for multipath routes.
1792 Default: 0 (Layer 3)
1796 - 0 - Layer 3 (source and destination addresses plus flow label)
1797 - 1 - Layer 4 (standard 5-tuple)
1798 - 2 - Layer 3 or inner Layer 3 if present
1799 - 3 - Custom multipath hash. Fields used for multipath hash calculation
1800 are determined by fib_multipath_hash_fields sysctl
1802 fib_multipath_hash_fields - UNSIGNED INTEGER
1803 When fib_multipath_hash_policy is set to 3 (custom multipath hash), the
1804 fields used for multipath hash calculation are determined by this
1807 This value is a bitmask which enables various fields for multipath hash
1810 Possible fields are:
1812 ====== ============================
1813 0x0001 Source IP address
1814 0x0002 Destination IP address
1818 0x0020 Destination port
1819 0x0040 Inner source IP address
1820 0x0080 Inner destination IP address
1821 0x0100 Inner IP protocol
1822 0x0200 Inner Flow Label
1823 0x0400 Inner source port
1824 0x0800 Inner destination port
1825 ====== ============================
1827 Default: 0x0007 (source IP, destination IP and IP protocol)
1829 anycast_src_echo_reply - BOOLEAN
1830 Controls the use of anycast addresses as source addresses for ICMPv6
1838 idgen_delay - INTEGER
1839 Controls the delay in seconds after which time to retry
1840 privacy stable address generation if a DAD conflict is
1843 Default: 1 (as specified in RFC7217)
1845 idgen_retries - INTEGER
1846 Controls the number of retries to generate a stable privacy
1847 address if a DAD conflict is detected.
1849 Default: 3 (as specified in RFC7217)
1852 Controls the MLD query robustness variable (see RFC3810 9.1).
1854 Default: 2 (as specified by RFC3810 9.1)
1856 Minimum: 1 (as specified by RFC6636 4.5)
1858 max_dst_opts_number - INTEGER
1859 Maximum number of non-padding TLVs allowed in a Destination
1860 options extension header. If this value is less than zero
1861 then unknown options are disallowed and the number of known
1862 TLVs allowed is the absolute value of this number.
1866 max_hbh_opts_number - INTEGER
1867 Maximum number of non-padding TLVs allowed in a Hop-by-Hop
1868 options extension header. If this value is less than zero
1869 then unknown options are disallowed and the number of known
1870 TLVs allowed is the absolute value of this number.
1874 max_dst_opts_length - INTEGER
1875 Maximum length allowed for a Destination options extension
1878 Default: INT_MAX (unlimited)
1880 max_hbh_length - INTEGER
1881 Maximum length allowed for a Hop-by-Hop options extension
1884 Default: INT_MAX (unlimited)
1886 skip_notify_on_dev_down - BOOLEAN
1887 Controls whether an RTM_DELROUTE message is generated for routes
1888 removed when a device is taken down or deleted. IPv4 does not
1889 generate this message; IPv6 does by default. Setting this sysctl
1890 to true skips the message, making IPv4 and IPv6 on par in relying
1891 on userspace caches to track link events and evict routes.
1893 Default: false (generate message)
1895 nexthop_compat_mode - BOOLEAN
1896 New nexthop API provides a means for managing nexthops independent of
1897 prefixes. Backwards compatibilty with old route format is enabled by
1898 default which means route dumps and notifications contain the new
1899 nexthop attribute but also the full, expanded nexthop definition.
1900 Further, updates or deletes of a nexthop configuration generate route
1901 notifications for each fib entry using the nexthop. Once a system
1902 understands the new API, this sysctl can be disabled to achieve full
1903 performance benefits of the new API by disabling the nexthop expansion
1904 and extraneous notifications.
1905 Default: true (backward compat mode)
1907 fib_notify_on_flag_change - INTEGER
1908 Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/
1909 RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed.
1911 After installing a route to the kernel, user space receives an
1912 acknowledgment, which means the route was installed in the kernel,
1913 but not necessarily in hardware.
1914 It is also possible for a route already installed in hardware to change
1915 its action and therefore its flags. For example, a host route that is
1916 trapping packets can be "promoted" to perform decapsulation following
1917 the installation of an IPinIP/VXLAN tunnel.
1918 The notifications will indicate to user-space the state of the route.
1920 Default: 0 (Do not emit notifications.)
1924 - 0 - Do not emit notifications.
1925 - 1 - Emit notifications.
1926 - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change.
1929 Define the IOAM id of this node. Uses only 24 bits out of 32 in total.
1936 ioam6_id_wide - LONG INTEGER
1937 Define the wide IOAM id of this node. Uses only 56 bits out of 64 in
1938 total. Can be different from ioam6_id.
1941 Max: 0xFFFFFFFFFFFFFF
1943 Default: 0xFFFFFFFFFFFFFF
1947 ip6frag_high_thresh - INTEGER
1948 Maximum memory used to reassemble IPv6 fragments. When
1949 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1950 the fragment handler will toss packets until ip6frag_low_thresh
1953 ip6frag_low_thresh - INTEGER
1954 See ip6frag_high_thresh
1956 ip6frag_time - INTEGER
1957 Time in seconds to keep an IPv6 fragment in memory.
1960 Change the interface-specific default settings.
1962 These settings would be used during creating new interfaces.
1966 Change all the interface-specific settings.
1968 [XXX: Other special features than forwarding?]
1970 conf/all/disable_ipv6 - BOOLEAN
1971 Changing this value is same as changing ``conf/default/disable_ipv6``
1972 setting and also all per-interface ``disable_ipv6`` settings to the same
1975 Reading this value does not have any particular meaning. It does not say
1976 whether IPv6 support is enabled or disabled. Returned value can be 1
1977 also in the case when some interface has ``disable_ipv6`` set to 0 and
1978 has configured IPv6 addresses.
1980 conf/all/forwarding - BOOLEAN
1981 Enable global IPv6 forwarding between all interfaces.
1983 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1984 to control which interfaces may forward packets and which not.
1986 This also sets all interfaces' Host/Router setting
1987 'forwarding' to the specified value. See below for details.
1989 This referred to as global forwarding.
1994 fwmark_reflect - BOOLEAN
1995 Controls the fwmark of kernel-generated IPv6 reply packets that are not
1996 associated with a socket for example, TCP RSTs or ICMPv6 echo replies).
1997 If unset, these packets have a fwmark of zero. If set, they have the
1998 fwmark of the packet they are replying to.
2002 ``conf/interface/*``:
2003 Change special settings per interface.
2005 The functional behaviour for certain settings is different
2006 depending on whether local forwarding is enabled or not.
2009 Accept Router Advertisements; autoconfigure using them.
2011 It also determines whether or not to transmit Router
2012 Solicitations. If and only if the functional setting is to
2013 accept Router Advertisements, Router Solicitations will be
2016 Possible values are:
2018 == ===========================================================
2019 0 Do not accept Router Advertisements.
2020 1 Accept Router Advertisements if forwarding is disabled.
2021 2 Overrule forwarding behaviour. Accept Router Advertisements
2022 even if forwarding is enabled.
2023 == ===========================================================
2027 - enabled if local forwarding is disabled.
2028 - disabled if local forwarding is enabled.
2030 accept_ra_defrtr - BOOLEAN
2031 Learn default router in Router Advertisement.
2035 - enabled if accept_ra is enabled.
2036 - disabled if accept_ra is disabled.
2038 ra_defrtr_metric - UNSIGNED INTEGER
2039 Route metric for default route learned in Router Advertisement. This value
2040 will be assigned as metric for the default route learned via IPv6 Router
2041 Advertisement. Takes affect only if accept_ra_defrtr is enabled.
2046 Default: IP6_RT_PRIO_USER i.e. 1024.
2048 accept_ra_from_local - BOOLEAN
2049 Accept RA with source-address that is found on local machine
2050 if the RA is otherwise proper and able to be accepted.
2052 Default is to NOT accept these as it may be an un-intended
2057 - enabled if accept_ra_from_local is enabled
2058 on a specific interface.
2059 - disabled if accept_ra_from_local is disabled
2060 on a specific interface.
2062 accept_ra_min_hop_limit - INTEGER
2063 Minimum hop limit Information in Router Advertisement.
2065 Hop limit Information in Router Advertisement less than this
2066 variable shall be ignored.
2070 accept_ra_pinfo - BOOLEAN
2071 Learn Prefix Information in Router Advertisement.
2075 - enabled if accept_ra is enabled.
2076 - disabled if accept_ra is disabled.
2078 accept_ra_rt_info_min_plen - INTEGER
2079 Minimum prefix length of Route Information in RA.
2081 Route Information w/ prefix smaller than this variable shall
2086 * 0 if accept_ra_rtr_pref is enabled.
2087 * -1 if accept_ra_rtr_pref is disabled.
2089 accept_ra_rt_info_max_plen - INTEGER
2090 Maximum prefix length of Route Information in RA.
2092 Route Information w/ prefix larger than this variable shall
2097 * 0 if accept_ra_rtr_pref is enabled.
2098 * -1 if accept_ra_rtr_pref is disabled.
2100 accept_ra_rtr_pref - BOOLEAN
2101 Accept Router Preference in RA.
2105 - enabled if accept_ra is enabled.
2106 - disabled if accept_ra is disabled.
2108 accept_ra_mtu - BOOLEAN
2109 Apply the MTU value specified in RA option 5 (RFC4861). If
2110 disabled, the MTU specified in the RA will be ignored.
2114 - enabled if accept_ra is enabled.
2115 - disabled if accept_ra is disabled.
2117 accept_redirects - BOOLEAN
2122 - enabled if local forwarding is disabled.
2123 - disabled if local forwarding is enabled.
2125 accept_source_route - INTEGER
2126 Accept source routing (routing extension header).
2128 - >= 0: Accept only routing header type 2.
2129 - < 0: Do not accept routing header.
2134 Autoconfigure addresses using Prefix Information in Router
2139 - enabled if accept_ra_pinfo is enabled.
2140 - disabled if accept_ra_pinfo is disabled.
2142 dad_transmits - INTEGER
2143 The amount of Duplicate Address Detection probes to send.
2147 forwarding - INTEGER
2148 Configure interface-specific Host/Router behaviour.
2152 It is recommended to have the same setting on all
2153 interfaces; mixed router/host scenarios are rather uncommon.
2155 Possible values are:
2157 - 0 Forwarding disabled
2158 - 1 Forwarding enabled
2162 By default, Host behaviour is assumed. This means:
2164 1. IsRouter flag is not set in Neighbour Advertisements.
2165 2. If accept_ra is TRUE (default), transmit Router
2167 3. If accept_ra is TRUE (default), accept Router
2168 Advertisements (and do autoconfiguration).
2169 4. If accept_redirects is TRUE (default), accept Redirects.
2173 If local forwarding is enabled, Router behaviour is assumed.
2174 This means exactly the reverse from the above:
2176 1. IsRouter flag is set in Neighbour Advertisements.
2177 2. Router Solicitations are not sent unless accept_ra is 2.
2178 3. Router Advertisements are ignored unless accept_ra is 2.
2179 4. Redirects are ignored.
2181 Default: 0 (disabled) if global forwarding is disabled (default),
2182 otherwise 1 (enabled).
2185 Default Hop Limit to set.
2190 Default Maximum Transfer Unit
2192 Default: 1280 (IPv6 required minimum)
2194 ip_nonlocal_bind - BOOLEAN
2195 If set, allows processes to bind() to non-local IPv6 addresses,
2196 which can be quite useful - but may break some applications.
2200 router_probe_interval - INTEGER
2201 Minimum interval (in seconds) between Router Probing described
2206 router_solicitation_delay - INTEGER
2207 Number of seconds to wait after interface is brought up
2208 before sending Router Solicitations.
2212 router_solicitation_interval - INTEGER
2213 Number of seconds to wait between Router Solicitations.
2217 router_solicitations - INTEGER
2218 Number of Router Solicitations to send until assuming no
2219 routers are present.
2223 use_oif_addrs_only - BOOLEAN
2224 When enabled, the candidate source addresses for destinations
2225 routed via this interface are restricted to the set of addresses
2226 configured on this interface (vis. RFC 6724, section 4).
2230 use_tempaddr - INTEGER
2231 Preference for Privacy Extensions (RFC3041).
2233 * <= 0 : disable Privacy Extensions
2234 * == 1 : enable Privacy Extensions, but prefer public
2235 addresses over temporary addresses.
2236 * > 1 : enable Privacy Extensions and prefer temporary
2237 addresses over public addresses.
2241 * 0 (for most devices)
2242 * -1 (for point-to-point devices and loopback devices)
2244 temp_valid_lft - INTEGER
2245 valid lifetime (in seconds) for temporary addresses.
2247 Default: 172800 (2 days)
2249 temp_prefered_lft - INTEGER
2250 Preferred lifetime (in seconds) for temporary addresses.
2252 Default: 86400 (1 day)
2254 keep_addr_on_down - INTEGER
2255 Keep all IPv6 addresses on an interface down event. If set static
2256 global addresses with no expiration time are not flushed.
2259 * 0 : system default
2262 Default: 0 (addresses are removed)
2264 max_desync_factor - INTEGER
2265 Maximum value for DESYNC_FACTOR, which is a random value
2266 that ensures that clients don't synchronize with each
2267 other and generate new addresses at exactly the same time.
2268 value is in seconds.
2272 regen_max_retry - INTEGER
2273 Number of attempts before give up attempting to generate
2274 valid temporary addresses.
2278 max_addresses - INTEGER
2279 Maximum number of autoconfigured addresses per interface. Setting
2280 to zero disables the limitation. It is not recommended to set this
2281 value too large (or to zero) because it would be an easy way to
2282 crash the kernel by allowing too many addresses to be created.
2286 disable_ipv6 - BOOLEAN
2287 Disable IPv6 operation. If accept_dad is set to 2, this value
2288 will be dynamically set to TRUE if DAD fails for the link-local
2291 Default: FALSE (enable IPv6 operation)
2293 When this value is changed from 1 to 0 (IPv6 is being enabled),
2294 it will dynamically create a link-local address on the given
2295 interface and start Duplicate Address Detection, if necessary.
2297 When this value is changed from 0 to 1 (IPv6 is being disabled),
2298 it will dynamically delete all addresses and routes on the given
2299 interface. From now on it will not possible to add addresses/routes
2300 to the selected interface.
2302 accept_dad - INTEGER
2303 Whether to accept DAD (Duplicate Address Detection).
2305 == ==============================================================
2307 1 Enable DAD (default)
2308 2 Enable DAD, and disable IPv6 operation if MAC-based duplicate
2309 link-local address has been found.
2310 == ==============================================================
2312 DAD operation and mode on a given interface will be selected according
2313 to the maximum value of conf/{all,interface}/accept_dad.
2315 force_tllao - BOOLEAN
2316 Enable sending the target link-layer address option even when
2317 responding to a unicast neighbor solicitation.
2321 Quoting from RFC 2461, section 4.4, Target link-layer address:
2323 "The option MUST be included for multicast solicitations in order to
2324 avoid infinite Neighbor Solicitation "recursion" when the peer node
2325 does not have a cache entry to return a Neighbor Advertisements
2326 message. When responding to unicast solicitations, the option can be
2327 omitted since the sender of the solicitation has the correct link-
2328 layer address; otherwise it would not have be able to send the unicast
2329 solicitation in the first place. However, including the link-layer
2330 address in this case adds little overhead and eliminates a potential
2331 race condition where the sender deletes the cached link-layer address
2332 prior to receiving a response to a previous solicitation."
2334 ndisc_notify - BOOLEAN
2335 Define mode for notification of address and device changes.
2337 * 0 - (default): do nothing
2338 * 1 - Generate unsolicited neighbour advertisements when device is brought
2339 up or hardware address changes.
2341 ndisc_tclass - INTEGER
2342 The IPv6 Traffic Class to use by default when sending IPv6 Neighbor
2343 Discovery (Router Solicitation, Router Advertisement, Neighbor
2344 Solicitation, Neighbor Advertisement, Redirect) messages.
2345 These 8 bits can be interpreted as 6 high order bits holding the DSCP
2346 value and 2 low order bits representing ECN (which you probably want
2351 ndisc_evict_nocarrier - BOOLEAN
2352 Clears the neighbor discovery table on NOCARRIER events. This option is
2353 important for wireless devices where the neighbor discovery cache should
2354 not be cleared when roaming between access points on the same network.
2355 In most cases this should remain as the default (1).
2357 - 1 - (default): Clear neighbor discover cache on NOCARRIER events.
2358 - 0 - Do not clear neighbor discovery cache on NOCARRIER events.
2360 mldv1_unsolicited_report_interval - INTEGER
2361 The interval in milliseconds in which the next unsolicited
2362 MLDv1 report retransmit will take place.
2364 Default: 10000 (10 seconds)
2366 mldv2_unsolicited_report_interval - INTEGER
2367 The interval in milliseconds in which the next unsolicited
2368 MLDv2 report retransmit will take place.
2370 Default: 1000 (1 second)
2372 force_mld_version - INTEGER
2373 * 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
2374 * 1 - Enforce to use MLD version 1
2375 * 2 - Enforce to use MLD version 2
2377 suppress_frag_ndisc - INTEGER
2378 Control RFC 6980 (Security Implications of IPv6 Fragmentation
2379 with IPv6 Neighbor Discovery) behavior:
2381 * 1 - (default) discard fragmented neighbor discovery packets
2382 * 0 - allow fragmented neighbor discovery packets
2384 optimistic_dad - BOOLEAN
2385 Whether to perform Optimistic Duplicate Address Detection (RFC 4429).
2387 * 0: disabled (default)
2390 Optimistic Duplicate Address Detection for the interface will be enabled
2391 if at least one of conf/{all,interface}/optimistic_dad is set to 1,
2392 it will be disabled otherwise.
2394 use_optimistic - BOOLEAN
2395 If enabled, do not classify optimistic addresses as deprecated during
2396 source address selection. Preferred addresses will still be chosen
2397 before optimistic addresses, subject to other ranking in the source
2398 address selection algorithm.
2400 * 0: disabled (default)
2403 This will be enabled if at least one of
2404 conf/{all,interface}/use_optimistic is set to 1, disabled otherwise.
2406 stable_secret - IPv6 address
2407 This IPv6 address will be used as a secret to generate IPv6
2408 addresses for link-local addresses and autoconfigured
2409 ones. All addresses generated after setting this secret will
2410 be stable privacy ones by default. This can be changed via the
2411 addrgenmode ip-link. conf/default/stable_secret is used as the
2412 secret for the namespace, the interface specific ones can
2413 overwrite that. Writes to conf/all/stable_secret are refused.
2415 It is recommended to generate this secret during installation
2416 of a system and keep it stable after that.
2418 By default the stable secret is unset.
2420 addr_gen_mode - INTEGER
2421 Defines how link-local and autoconf addresses are generated.
2423 = =================================================================
2424 0 generate address based on EUI64 (default)
2425 1 do no generate a link-local address, use EUI64 for addresses
2426 generated from autoconf
2427 2 generate stable privacy addresses, using the secret from
2428 stable_secret (RFC7217)
2429 3 generate stable privacy addresses, using a random secret if unset
2430 = =================================================================
2432 drop_unicast_in_l2_multicast - BOOLEAN
2433 Drop any unicast IPv6 packets that are received in link-layer
2434 multicast (or broadcast) frames.
2436 By default this is turned off.
2438 drop_unsolicited_na - BOOLEAN
2439 Drop all unsolicited neighbor advertisements, for example if there's
2440 a known good NA proxy on the network and such frames need not be used
2441 (or in the case of 802.11, must not be used to prevent attacks.)
2443 By default this is turned off.
2445 enhanced_dad - BOOLEAN
2446 Include a nonce option in the IPv6 neighbor solicitation messages used for
2447 duplicate address detection per RFC7527. A received DAD NS will only signal
2448 a duplicate address if the nonce is different. This avoids any false
2449 detection of duplicates due to loopback of the NS messages that we send.
2450 The nonce option will be sent on an interface unless both of
2451 conf/{all,interface}/enhanced_dad are set to FALSE.
2459 Limit the maximal rates for sending ICMPv6 messages.
2461 0 to disable any limiting,
2462 otherwise the minimal space between responses in milliseconds.
2466 ratemask - list of comma separated ranges
2467 For ICMPv6 message types matching the ranges in the ratemask, limit
2468 the sending of the message according to ratelimit parameter.
2470 The format used for both input and output is a comma separated
2471 list of ranges (e.g. "0-127,129" for ICMPv6 message type 0 to 127 and
2472 129). Writing to the file will clear all previous ranges of ICMPv6
2473 message types and update the current list with the input.
2475 Refer to: https://www.iana.org/assignments/icmpv6-parameters/icmpv6-parameters.xhtml
2476 for numerical values of ICMPv6 message types, e.g. echo request is 128
2477 and echo reply is 129.
2479 Default: 0-1,3-127 (rate limit ICMPv6 errors except Packet Too Big)
2481 echo_ignore_all - BOOLEAN
2482 If set non-zero, then the kernel will ignore all ICMP ECHO
2483 requests sent to it over the IPv6 protocol.
2487 echo_ignore_multicast - BOOLEAN
2488 If set non-zero, then the kernel will ignore all ICMP ECHO
2489 requests sent to it over the IPv6 protocol via multicast.
2493 echo_ignore_anycast - BOOLEAN
2494 If set non-zero, then the kernel will ignore all ICMP ECHO
2495 requests sent to it over the IPv6 protocol destined to anycast address.
2499 xfrm6_gc_thresh - INTEGER
2500 (Obsolete since linux-4.14)
2501 The threshold at which we will start garbage collecting for IPv6
2502 destination cache entries. At twice this value the system will
2503 refuse new allocations.
2507 Pekka Savola <pekkas@netcore.fi>
2508 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
2511 /proc/sys/net/bridge/* Variables:
2512 =================================
2514 bridge-nf-call-arptables - BOOLEAN
2515 - 1 : pass bridged ARP traffic to arptables' FORWARD chain.
2520 bridge-nf-call-iptables - BOOLEAN
2521 - 1 : pass bridged IPv4 traffic to iptables' chains.
2526 bridge-nf-call-ip6tables - BOOLEAN
2527 - 1 : pass bridged IPv6 traffic to ip6tables' chains.
2532 bridge-nf-filter-vlan-tagged - BOOLEAN
2533 - 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
2538 bridge-nf-filter-pppoe-tagged - BOOLEAN
2539 - 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
2544 bridge-nf-pass-vlan-input-dev - BOOLEAN
2545 - 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
2546 interface on the bridge and set the netfilter input device to the
2547 vlan. This allows use of e.g. "iptables -i br0.1" and makes the
2548 REDIRECT target work with vlan-on-top-of-bridge interfaces. When no
2549 matching vlan interface is found, or this switch is off, the input
2550 device is set to the bridge interface.
2552 - 0: disable bridge netfilter vlan interface lookup.
2556 ``proc/sys/net/sctp/*`` Variables:
2557 ==================================
2559 addip_enable - BOOLEAN
2560 Enable or disable extension of Dynamic Address Reconfiguration
2561 (ADD-IP) functionality specified in RFC5061. This extension provides
2562 the ability to dynamically add and remove new addresses for the SCTP
2565 1: Enable extension.
2567 0: Disable extension.
2572 Enable or disable pf (pf is short for potentially failed) state. A value
2573 of pf_retrans > path_max_retrans also disables pf state. That is, one of
2574 both pf_enable and pf_retrans > path_max_retrans can disable pf state.
2575 Since pf_retrans and path_max_retrans can be changed by userspace
2576 application, sometimes user expects to disable pf state by the value of
2577 pf_retrans > path_max_retrans, but occasionally the value of pf_retrans
2578 or path_max_retrans is changed by the user application, this pf state is
2579 enabled. As such, it is necessary to add this to dynamically enable
2580 and disable pf state. See:
2581 https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for
2591 Unset or enable/disable pf (pf is short for potentially failed) state
2592 exposure. Applications can control the exposure of the PF path state
2593 in the SCTP_PEER_ADDR_CHANGE event and the SCTP_GET_PEER_ADDR_INFO
2594 sockopt. When it's unset, no SCTP_PEER_ADDR_CHANGE event with
2595 SCTP_ADDR_PF state will be sent and a SCTP_PF-state transport info
2596 can be got via SCTP_GET_PEER_ADDR_INFO sockopt; When it's enabled,
2597 a SCTP_PEER_ADDR_CHANGE event will be sent for a transport becoming
2598 SCTP_PF state and a SCTP_PF-state transport info can be got via
2599 SCTP_GET_PEER_ADDR_INFO sockopt; When it's diabled, no
2600 SCTP_PEER_ADDR_CHANGE event will be sent and it returns -EACCES when
2601 trying to get a SCTP_PF-state transport info via SCTP_GET_PEER_ADDR_INFO
2604 0: Unset pf state exposure, Compatible with old applications.
2606 1: Disable pf state exposure.
2608 2: Enable pf state exposure.
2612 addip_noauth_enable - BOOLEAN
2613 Dynamic Address Reconfiguration (ADD-IP) requires the use of
2614 authentication to protect the operations of adding or removing new
2615 addresses. This requirement is mandated so that unauthorized hosts
2616 would not be able to hijack associations. However, older
2617 implementations may not have implemented this requirement while
2618 allowing the ADD-IP extension. For reasons of interoperability,
2619 we provide this variable to control the enforcement of the
2620 authentication requirement.
2622 == ===============================================================
2623 1 Allow ADD-IP extension to be used without authentication. This
2624 should only be set in a closed environment for interoperability
2625 with older implementations.
2627 0 Enforce the authentication requirement
2628 == ===============================================================
2632 auth_enable - BOOLEAN
2633 Enable or disable Authenticated Chunks extension. This extension
2634 provides the ability to send and receive authenticated chunks and is
2635 required for secure operation of Dynamic Address Reconfiguration
2638 - 1: Enable this extension.
2639 - 0: Disable this extension.
2643 prsctp_enable - BOOLEAN
2644 Enable or disable the Partial Reliability extension (RFC3758) which
2645 is used to notify peers that a given DATA should no longer be expected.
2647 - 1: Enable extension
2653 The limit of the number of new packets that can be initially sent. It
2654 controls how bursty the generated traffic can be.
2658 association_max_retrans - INTEGER
2659 Set the maximum number for retransmissions that an association can
2660 attempt deciding that the remote end is unreachable. If this value
2661 is exceeded, the association is terminated.
2665 max_init_retransmits - INTEGER
2666 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
2667 that an association will attempt before declaring the destination
2668 unreachable and terminating.
2672 path_max_retrans - INTEGER
2673 The maximum number of retransmissions that will be attempted on a given
2674 path. Once this threshold is exceeded, the path is considered
2675 unreachable, and new traffic will use a different path when the
2676 association is multihomed.
2680 pf_retrans - INTEGER
2681 The number of retransmissions that will be attempted on a given path
2682 before traffic is redirected to an alternate transport (should one
2683 exist). Note this is distinct from path_max_retrans, as a path that
2684 passes the pf_retrans threshold can still be used. Its only
2685 deprioritized when a transmission path is selected by the stack. This
2686 setting is primarily used to enable fast failover mechanisms without
2687 having to reduce path_max_retrans to a very low value. See:
2688 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
2689 for details. Note also that a value of pf_retrans > path_max_retrans
2690 disables this feature. Since both pf_retrans and path_max_retrans can
2691 be changed by userspace application, a variable pf_enable is used to
2696 ps_retrans - INTEGER
2697 Primary.Switchover.Max.Retrans (PSMR), it's a tunable parameter coming
2698 from section-5 "Primary Path Switchover" in rfc7829. The primary path
2699 will be changed to another active path when the path error counter on
2700 the old primary path exceeds PSMR, so that "the SCTP sender is allowed
2701 to continue data transmission on a new working path even when the old
2702 primary destination address becomes active again". Note this feature
2703 is disabled by initializing 'ps_retrans' per netns as 0xffff by default,
2704 and its value can't be less than 'pf_retrans' when changing by sysctl.
2708 rto_initial - INTEGER
2709 The initial round trip timeout value in milliseconds that will be used
2710 in calculating round trip times. This is the initial time interval
2711 for retransmissions.
2716 The maximum value (in milliseconds) of the round trip timeout. This
2717 is the largest time interval that can elapse between retransmissions.
2722 The minimum value (in milliseconds) of the round trip timeout. This
2723 is the smallest time interval the can elapse between retransmissions.
2727 hb_interval - INTEGER
2728 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
2729 are sent at the specified interval on idle paths to probe the state of
2730 a given path between 2 associations.
2734 sack_timeout - INTEGER
2735 The amount of time (in milliseconds) that the implementation will wait
2740 valid_cookie_life - INTEGER
2741 The default lifetime of the SCTP cookie (in milliseconds). The cookie
2742 is used during association establishment.
2746 cookie_preserve_enable - BOOLEAN
2747 Enable or disable the ability to extend the lifetime of the SCTP cookie
2748 that is used during the establishment phase of SCTP association
2750 - 1: Enable cookie lifetime extension.
2755 cookie_hmac_alg - STRING
2756 Select the hmac algorithm used when generating the cookie value sent by
2757 a listening sctp socket to a connecting client in the INIT-ACK chunk.
2764 Ability to assign md5 or sha1 as the selected alg is predicated on the
2765 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
2766 CONFIG_CRYPTO_SHA1).
2768 Default: Dependent on configuration. MD5 if available, else SHA1 if
2769 available, else none.
2771 rcvbuf_policy - INTEGER
2772 Determines if the receive buffer is attributed to the socket or to
2773 association. SCTP supports the capability to create multiple
2774 associations on a single socket. When using this capability, it is
2775 possible that a single stalled association that's buffering a lot
2776 of data may block other associations from delivering their data by
2777 consuming all of the receive buffer space. To work around this,
2778 the rcvbuf_policy could be set to attribute the receiver buffer space
2779 to each association instead of the socket. This prevents the described
2782 - 1: rcvbuf space is per association
2783 - 0: rcvbuf space is per socket
2787 sndbuf_policy - INTEGER
2788 Similar to rcvbuf_policy above, this applies to send buffer space.
2790 - 1: Send buffer is tracked per association
2791 - 0: Send buffer is tracked per socket.
2795 sctp_mem - vector of 3 INTEGERs: min, pressure, max
2796 Number of pages allowed for queueing by all SCTP sockets.
2798 min: Below this number of pages SCTP is not bothered about its
2799 memory appetite. When amount of memory allocated by SCTP exceeds
2800 this number, SCTP starts to moderate memory usage.
2802 pressure: This value was introduced to follow format of tcp_mem.
2804 max: Number of pages allowed for queueing by all SCTP sockets.
2806 Default is calculated at boot time from amount of available memory.
2808 sctp_rmem - vector of 3 INTEGERs: min, default, max
2809 Only the first value ("min") is used, "default" and "max" are
2812 min: Minimal size of receive buffer used by SCTP socket.
2813 It is guaranteed to each SCTP socket (but not association) even
2814 under moderate memory pressure.
2818 sctp_wmem - vector of 3 INTEGERs: min, default, max
2819 Currently this tunable has no effect.
2821 addr_scope_policy - INTEGER
2822 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
2824 - 0 - Disable IPv4 address scoping
2825 - 1 - Enable IPv4 address scoping
2826 - 2 - Follow draft but allow IPv4 private addresses
2827 - 3 - Follow draft but allow IPv4 link local addresses
2832 The listening port for the local UDP tunneling sock. Normally it's
2833 using the IANA-assigned UDP port number 9899 (sctp-tunneling).
2835 This UDP sock is used for processing the incoming UDP-encapsulated
2836 SCTP packets (from RFC6951), and shared by all applications in the
2837 same net namespace. This UDP sock will be closed when the value is
2840 The value will also be used to set the src port of the UDP header
2841 for the outgoing UDP-encapsulated SCTP packets. For the dest port,
2842 please refer to 'encap_port' below.
2846 encap_port - INTEGER
2847 The default remote UDP encapsulation port.
2849 This value is used to set the dest port of the UDP header for the
2850 outgoing UDP-encapsulated SCTP packets by default. Users can also
2851 change the value for each sock/asoc/transport by using setsockopt.
2852 For further information, please refer to RFC6951.
2854 Note that when connecting to a remote server, the client should set
2855 this to the port that the UDP tunneling sock on the peer server is
2856 listening to and the local UDP tunneling sock on the client also
2857 must be started. On the server, it would get the encap_port from
2858 the incoming packet's source port.
2862 plpmtud_probe_interval - INTEGER
2863 The time interval (in milliseconds) for the PLPMTUD probe timer,
2864 which is configured to expire after this period to receive an
2865 acknowledgment to a probe packet. This is also the time interval
2866 between the probes for the current pmtu when the probe search
2869 PLPMTUD will be disabled when 0 is set, and other values for it
2875 ``/proc/sys/net/core/*``
2876 ========================
2878 Please see: Documentation/admin-guide/sysctl/net.rst for descriptions of these entries.
2881 ``/proc/sys/net/unix/*``
2882 ========================
2884 max_dgram_qlen - INTEGER
2885 The maximum length of dgram socket receive queue