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
103 fib_sync_mem - UNSIGNED INTEGER
104 Amount of dirty memory from fib entries that can be backlogged before
105 synchronize_rcu is forced.
107 Default: 512kB Minimum: 64kB Maximum: 64MB
109 ip_forward_update_priority - INTEGER
110 Whether to update SKB priority from "TOS" field in IPv4 header after it
111 is forwarded. The new SKB priority is mapped from TOS field value
112 according to an rt_tos2priority table (see e.g. man tc-prio).
114 Default: 1 (Update priority.)
118 - 0 - Do not update priority.
119 - 1 - Update priority.
121 route/max_size - INTEGER
122 Maximum number of routes allowed in the kernel. Increase
123 this when using large numbers of interfaces and/or routes.
125 From linux kernel 3.6 onwards, this is deprecated for ipv4
126 as route cache is no longer used.
128 neigh/default/gc_thresh1 - INTEGER
129 Minimum number of entries to keep. Garbage collector will not
130 purge entries if there are fewer than this number.
134 neigh/default/gc_thresh2 - INTEGER
135 Threshold when garbage collector becomes more aggressive about
136 purging entries. Entries older than 5 seconds will be cleared
137 when over this number.
141 neigh/default/gc_thresh3 - INTEGER
142 Maximum number of non-PERMANENT neighbor entries allowed. Increase
143 this when using large numbers of interfaces and when communicating
144 with large numbers of directly-connected peers.
148 neigh/default/unres_qlen_bytes - INTEGER
149 The maximum number of bytes which may be used by packets
150 queued for each unresolved address by other network layers.
153 Setting negative value is meaningless and will return error.
155 Default: SK_WMEM_MAX, (same as net.core.wmem_default).
157 Exact value depends on architecture and kernel options,
158 but should be enough to allow queuing 256 packets
161 neigh/default/unres_qlen - INTEGER
162 The maximum number of packets which may be queued for each
163 unresolved address by other network layers.
165 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
167 Prior to linux 3.3, the default value is 3 which may cause
168 unexpected packet loss. The current default value is calculated
169 according to default value of unres_qlen_bytes and true size of
174 mtu_expires - INTEGER
175 Time, in seconds, that cached PMTU information is kept.
177 min_adv_mss - INTEGER
178 The advertised MSS depends on the first hop route MTU, but will
179 never be lower than this setting.
183 ipfrag_high_thresh - LONG INTEGER
184 Maximum memory used to reassemble IP fragments.
186 ipfrag_low_thresh - LONG INTEGER
187 (Obsolete since linux-4.17)
188 Maximum memory used to reassemble IP fragments before the kernel
189 begins to remove incomplete fragment queues to free up resources.
190 The kernel still accepts new fragments for defragmentation.
192 ipfrag_time - INTEGER
193 Time in seconds to keep an IP fragment in memory.
195 ipfrag_max_dist - INTEGER
196 ipfrag_max_dist is a non-negative integer value which defines the
197 maximum "disorder" which is allowed among fragments which share a
198 common IP source address. Note that reordering of packets is
199 not unusual, but if a large number of fragments arrive from a source
200 IP address while a particular fragment queue remains incomplete, it
201 probably indicates that one or more fragments belonging to that queue
202 have been lost. When ipfrag_max_dist is positive, an additional check
203 is done on fragments before they are added to a reassembly queue - if
204 ipfrag_max_dist (or more) fragments have arrived from a particular IP
205 address between additions to any IP fragment queue using that source
206 address, it's presumed that one or more fragments in the queue are
207 lost. The existing fragment queue will be dropped, and a new one
208 started. An ipfrag_max_dist value of zero disables this check.
210 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
211 result in unnecessarily dropping fragment queues when normal
212 reordering of packets occurs, which could lead to poor application
213 performance. Using a very large value, e.g. 50000, increases the
214 likelihood of incorrectly reassembling IP fragments that originate
215 from different IP datagrams, which could result in data corruption.
221 inet_peer_threshold - INTEGER
222 The approximate size of the storage. Starting from this threshold
223 entries will be thrown aggressively. This threshold also determines
224 entries' time-to-live and time intervals between garbage collection
225 passes. More entries, less time-to-live, less GC interval.
227 inet_peer_minttl - INTEGER
228 Minimum time-to-live of entries. Should be enough to cover fragment
229 time-to-live on the reassembling side. This minimum time-to-live is
230 guaranteed if the pool size is less than inet_peer_threshold.
233 inet_peer_maxttl - INTEGER
234 Maximum time-to-live of entries. Unused entries will expire after
235 this period of time if there is no memory pressure on the pool (i.e.
236 when the number of entries in the pool is very small).
243 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
244 Defaults to 4096. (Was 128 before linux-5.4)
245 See also tcp_max_syn_backlog for additional tuning for TCP sockets.
247 tcp_abort_on_overflow - BOOLEAN
248 If listening service is too slow to accept new connections,
249 reset them. Default state is FALSE. It means that if overflow
250 occurred due to a burst, connection will recover. Enable this
251 option _only_ if you are really sure that listening daemon
252 cannot be tuned to accept connections faster. Enabling this
253 option can harm clients of your server.
255 tcp_adv_win_scale - INTEGER
256 Count buffering overhead as bytes/2^tcp_adv_win_scale
257 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
260 Possible values are [-31, 31], inclusive.
264 tcp_allowed_congestion_control - STRING
265 Show/set the congestion control choices available to non-privileged
266 processes. The list is a subset of those listed in
267 tcp_available_congestion_control.
269 Default is "reno" and the default setting (tcp_congestion_control).
271 tcp_app_win - INTEGER
272 Reserve max(window/2^tcp_app_win, mss) of window for application
273 buffer. Value 0 is special, it means that nothing is reserved.
277 tcp_autocorking - BOOLEAN
278 Enable TCP auto corking :
279 When applications do consecutive small write()/sendmsg() system calls,
280 we try to coalesce these small writes as much as possible, to lower
281 total amount of sent packets. This is done if at least one prior
282 packet for the flow is waiting in Qdisc queues or device transmit
283 queue. Applications can still use TCP_CORK for optimal behavior
284 when they know how/when to uncork their sockets.
288 tcp_available_congestion_control - STRING
289 Shows the available congestion control choices that are registered.
290 More congestion control algorithms may be available as modules,
293 tcp_base_mss - INTEGER
294 The initial value of search_low to be used by the packetization layer
295 Path MTU discovery (MTU probing). If MTU probing is enabled,
296 this is the initial MSS used by the connection.
298 tcp_mtu_probe_floor - INTEGER
299 If MTU probing is enabled this caps the minimum MSS used for search_low
304 tcp_min_snd_mss - INTEGER
305 TCP SYN and SYNACK messages usually advertise an ADVMSS option,
306 as described in RFC 1122 and RFC 6691.
308 If this ADVMSS option is smaller than tcp_min_snd_mss,
309 it is silently capped to tcp_min_snd_mss.
311 Default : 48 (at least 8 bytes of payload per segment)
313 tcp_congestion_control - STRING
314 Set the congestion control algorithm to be used for new
315 connections. The algorithm "reno" is always available, but
316 additional choices may be available based on kernel configuration.
317 Default is set as part of kernel configuration.
318 For passive connections, the listener congestion control choice
321 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
324 Allows TCP to send "duplicate" SACKs.
326 tcp_early_retrans - INTEGER
327 Tail loss probe (TLP) converts RTOs occurring due to tail
328 losses into fast recovery (draft-ietf-tcpm-rack). Note that
329 TLP requires RACK to function properly (see tcp_recovery below)
339 Control use of Explicit Congestion Notification (ECN) by TCP.
340 ECN is used only when both ends of the TCP connection indicate
341 support for it. This feature is useful in avoiding losses due
342 to congestion by allowing supporting routers to signal
343 congestion before having to drop packets.
347 = =====================================================
348 0 Disable ECN. Neither initiate nor accept ECN.
349 1 Enable ECN when requested by incoming connections and
350 also request ECN on outgoing connection attempts.
351 2 Enable ECN when requested by incoming connections
352 but do not request ECN on outgoing connections.
353 = =====================================================
357 tcp_ecn_fallback - BOOLEAN
358 If the kernel detects that ECN connection misbehaves, enable fall
359 back to non-ECN. Currently, this knob implements the fallback
360 from RFC3168, section 6.1.1.1., but we reserve that in future,
361 additional detection mechanisms could be implemented under this
362 knob. The value is not used, if tcp_ecn or per route (or congestion
363 control) ECN settings are disabled.
365 Default: 1 (fallback enabled)
368 This is a legacy option, it has no effect anymore.
370 tcp_fin_timeout - INTEGER
371 The length of time an orphaned (no longer referenced by any
372 application) connection will remain in the FIN_WAIT_2 state
373 before it is aborted at the local end. While a perfectly
374 valid "receive only" state for an un-orphaned connection, an
375 orphaned connection in FIN_WAIT_2 state could otherwise wait
376 forever for the remote to close its end of the connection.
383 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
384 F-RTO is an enhanced recovery algorithm for TCP retransmission
385 timeouts. It is particularly beneficial in networks where the
386 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
387 modification. It does not require any support from the peer.
389 By default it's enabled with a non-zero value. 0 disables F-RTO.
391 tcp_fwmark_accept - BOOLEAN
392 If set, incoming connections to listening sockets that do not have a
393 socket mark will set the mark of the accepting socket to the fwmark of
394 the incoming SYN packet. This will cause all packets on that connection
395 (starting from the first SYNACK) to be sent with that fwmark. The
396 listening socket's mark is unchanged. Listening sockets that already
397 have a fwmark set via setsockopt(SOL_SOCKET, SO_MARK, ...) are
402 tcp_invalid_ratelimit - INTEGER
403 Limit the maximal rate for sending duplicate acknowledgments
404 in response to incoming TCP packets that are for an existing
405 connection but that are invalid due to any of these reasons:
407 (a) out-of-window sequence number,
408 (b) out-of-window acknowledgment number, or
409 (c) PAWS (Protection Against Wrapped Sequence numbers) check failure
411 This can help mitigate simple "ack loop" DoS attacks, wherein
412 a buggy or malicious middlebox or man-in-the-middle can
413 rewrite TCP header fields in manner that causes each endpoint
414 to think that the other is sending invalid TCP segments, thus
415 causing each side to send an unterminating stream of duplicate
416 acknowledgments for invalid segments.
418 Using 0 disables rate-limiting of dupacks in response to
419 invalid segments; otherwise this value specifies the minimal
420 space between sending such dupacks, in milliseconds.
422 Default: 500 (milliseconds).
424 tcp_keepalive_time - INTEGER
425 How often TCP sends out keepalive messages when keepalive is enabled.
428 tcp_keepalive_probes - INTEGER
429 How many keepalive probes TCP sends out, until it decides that the
430 connection is broken. Default value: 9.
432 tcp_keepalive_intvl - INTEGER
433 How frequently the probes are send out. Multiplied by
434 tcp_keepalive_probes it is time to kill not responding connection,
435 after probes started. Default value: 75sec i.e. connection
436 will be aborted after ~11 minutes of retries.
438 tcp_l3mdev_accept - BOOLEAN
439 Enables child sockets to inherit the L3 master device index.
440 Enabling this option allows a "global" listen socket to work
441 across L3 master domains (e.g., VRFs) with connected sockets
442 derived from the listen socket to be bound to the L3 domain in
443 which the packets originated. Only valid when the kernel was
444 compiled with CONFIG_NET_L3_MASTER_DEV.
446 Default: 0 (disabled)
448 tcp_low_latency - BOOLEAN
449 This is a legacy option, it has no effect anymore.
451 tcp_max_orphans - INTEGER
452 Maximal number of TCP sockets not attached to any user file handle,
453 held by system. If this number is exceeded orphaned connections are
454 reset immediately and warning is printed. This limit exists
455 only to prevent simple DoS attacks, you _must_ not rely on this
456 or lower the limit artificially, but rather increase it
457 (probably, after increasing installed memory),
458 if network conditions require more than default value,
459 and tune network services to linger and kill such states
460 more aggressively. Let me to remind again: each orphan eats
461 up to ~64K of unswappable memory.
463 tcp_max_syn_backlog - INTEGER
464 Maximal number of remembered connection requests (SYN_RECV),
465 which have not received an acknowledgment from connecting client.
467 This is a per-listener limit.
469 The minimal value is 128 for low memory machines, and it will
470 increase in proportion to the memory of machine.
472 If server suffers from overload, try increasing this number.
474 Remember to also check /proc/sys/net/core/somaxconn
475 A SYN_RECV request socket consumes about 304 bytes of memory.
477 tcp_max_tw_buckets - INTEGER
478 Maximal number of timewait sockets held by system simultaneously.
479 If this number is exceeded time-wait socket is immediately destroyed
480 and warning is printed. This limit exists only to prevent
481 simple DoS attacks, you _must_ not lower the limit artificially,
482 but rather increase it (probably, after increasing installed memory),
483 if network conditions require more than default value.
485 tcp_mem - vector of 3 INTEGERs: min, pressure, max
486 min: below this number of pages TCP is not bothered about its
489 pressure: when amount of memory allocated by TCP exceeds this number
490 of pages, TCP moderates its memory consumption and enters memory
491 pressure mode, which is exited when memory consumption falls
494 max: number of pages allowed for queueing by all TCP sockets.
496 Defaults are calculated at boot time from amount of available
499 tcp_min_rtt_wlen - INTEGER
500 The window length of the windowed min filter to track the minimum RTT.
501 A shorter window lets a flow more quickly pick up new (higher)
502 minimum RTT when it is moved to a longer path (e.g., due to traffic
503 engineering). A longer window makes the filter more resistant to RTT
504 inflations such as transient congestion. The unit is seconds.
506 Possible values: 0 - 86400 (1 day)
510 tcp_moderate_rcvbuf - BOOLEAN
511 If set, TCP performs receive buffer auto-tuning, attempting to
512 automatically size the buffer (no greater than tcp_rmem[2]) to
513 match the size required by the path for full throughput. Enabled by
516 tcp_mtu_probing - INTEGER
517 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
521 - 1 - Disabled by default, enabled when an ICMP black hole detected
522 - 2 - Always enabled, use initial MSS of tcp_base_mss.
524 tcp_probe_interval - UNSIGNED INTEGER
525 Controls how often to start TCP Packetization-Layer Path MTU
526 Discovery reprobe. The default is reprobing every 10 minutes as
529 tcp_probe_threshold - INTEGER
530 Controls when TCP Packetization-Layer Path MTU Discovery probing
531 will stop in respect to the width of search range in bytes. Default
534 tcp_no_metrics_save - BOOLEAN
535 By default, TCP saves various connection metrics in the route cache
536 when the connection closes, so that connections established in the
537 near future can use these to set initial conditions. Usually, this
538 increases overall performance, but may sometimes cause performance
539 degradation. If set, TCP will not cache metrics on closing
542 tcp_no_ssthresh_metrics_save - BOOLEAN
543 Controls whether TCP saves ssthresh metrics in the route cache.
545 Default is 1, which disables ssthresh metrics.
547 tcp_orphan_retries - INTEGER
548 This value influences the timeout of a locally closed TCP connection,
549 when RTO retransmissions remain unacknowledged.
550 See tcp_retries2 for more details.
552 The default value is 8.
554 If your machine is a loaded WEB server,
555 you should think about lowering this value, such sockets
556 may consume significant resources. Cf. tcp_max_orphans.
558 tcp_recovery - INTEGER
559 This value is a bitmap to enable various experimental loss recovery
562 ========= =============================================================
563 RACK: 0x1 enables the RACK loss detection for fast detection of lost
564 retransmissions and tail drops. It also subsumes and disables
565 RFC6675 recovery for SACK connections.
567 RACK: 0x2 makes RACK's reordering window static (min_rtt/4).
569 RACK: 0x4 disables RACK's DUPACK threshold heuristic
570 ========= =============================================================
574 tcp_reordering - INTEGER
575 Initial reordering level of packets in a TCP stream.
576 TCP stack can then dynamically adjust flow reordering level
577 between this initial value and tcp_max_reordering
581 tcp_max_reordering - INTEGER
582 Maximal reordering level of packets in a TCP stream.
583 300 is a fairly conservative value, but you might increase it
584 if paths are using per packet load balancing (like bonding rr mode)
588 tcp_retrans_collapse - BOOLEAN
589 Bug-to-bug compatibility with some broken printers.
590 On retransmit try to send bigger packets to work around bugs in
593 tcp_retries1 - INTEGER
594 This value influences the time, after which TCP decides, that
595 something is wrong due to unacknowledged RTO retransmissions,
596 and reports this suspicion to the network layer.
597 See tcp_retries2 for more details.
599 RFC 1122 recommends at least 3 retransmissions, which is the
602 tcp_retries2 - INTEGER
603 This value influences the timeout of an alive TCP connection,
604 when RTO retransmissions remain unacknowledged.
605 Given a value of N, a hypothetical TCP connection following
606 exponential backoff with an initial RTO of TCP_RTO_MIN would
607 retransmit N times before killing the connection at the (N+1)th RTO.
609 The default value of 15 yields a hypothetical timeout of 924.6
610 seconds and is a lower bound for the effective timeout.
611 TCP will effectively time out at the first RTO which exceeds the
612 hypothetical timeout.
614 RFC 1122 recommends at least 100 seconds for the timeout,
615 which corresponds to a value of at least 8.
617 tcp_rfc1337 - BOOLEAN
618 If set, the TCP stack behaves conforming to RFC1337. If unset,
619 we are not conforming to RFC, but prevent TCP TIME_WAIT
624 tcp_rmem - vector of 3 INTEGERs: min, default, max
625 min: Minimal size of receive buffer used by TCP sockets.
626 It is guaranteed to each TCP socket, even under moderate memory
631 default: initial size of receive buffer used by TCP sockets.
632 This value overrides net.core.rmem_default used by other protocols.
633 Default: 131072 bytes.
634 This value results in initial window of 65535.
636 max: maximal size of receive buffer allowed for automatically
637 selected receiver buffers for TCP socket. This value does not override
638 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
639 automatic tuning of that socket's receive buffer size, in which
640 case this value is ignored.
641 Default: between 131072 and 6MB, depending on RAM size.
644 Enable select acknowledgments (SACKS).
646 tcp_comp_sack_delay_ns - LONG INTEGER
647 TCP tries to reduce number of SACK sent, using a timer
648 based on 5% of SRTT, capped by this sysctl, in nano seconds.
649 The default is 1ms, based on TSO autosizing period.
651 Default : 1,000,000 ns (1 ms)
653 tcp_comp_sack_slack_ns - LONG INTEGER
654 This sysctl control the slack used when arming the
655 timer used by SACK compression. This gives extra time
656 for small RTT flows, and reduces system overhead by allowing
657 opportunistic reduction of timer interrupts.
659 Default : 100,000 ns (100 us)
661 tcp_comp_sack_nr - INTEGER
662 Max number of SACK that can be compressed.
663 Using 0 disables SACK compression.
667 tcp_slow_start_after_idle - BOOLEAN
668 If set, provide RFC2861 behavior and time out the congestion
669 window after an idle period. An idle period is defined at
670 the current RTO. If unset, the congestion window will not
671 be timed out after an idle period.
676 Use the Host requirements interpretation of the TCP urgent pointer field.
677 Most hosts use the older BSD interpretation, so if you turn this on
678 Linux might not communicate correctly with them.
682 tcp_synack_retries - INTEGER
683 Number of times SYNACKs for a passive TCP connection attempt will
684 be retransmitted. Should not be higher than 255. Default value
685 is 5, which corresponds to 31seconds till the last retransmission
686 with the current initial RTO of 1second. With this the final timeout
687 for a passive TCP connection will happen after 63seconds.
689 tcp_syncookies - INTEGER
690 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
691 Send out syncookies when the syn backlog queue of a socket
692 overflows. This is to prevent against the common 'SYN flood attack'
695 Note, that syncookies is fallback facility.
696 It MUST NOT be used to help highly loaded servers to stand
697 against legal connection rate. If you see SYN flood warnings
698 in your logs, but investigation shows that they occur
699 because of overload with legal connections, you should tune
700 another parameters until this warning disappear.
701 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
703 syncookies seriously violate TCP protocol, do not allow
704 to use TCP extensions, can result in serious degradation
705 of some services (f.e. SMTP relaying), visible not by you,
706 but your clients and relays, contacting you. While you see
707 SYN flood warnings in logs not being really flooded, your server
708 is seriously misconfigured.
710 If you want to test which effects syncookies have to your
711 network connections you can set this knob to 2 to enable
712 unconditionally generation of syncookies.
714 tcp_fastopen - INTEGER
715 Enable TCP Fast Open (RFC7413) to send and accept data in the opening
718 The client support is enabled by flag 0x1 (on by default). The client
719 then must use sendmsg() or sendto() with the MSG_FASTOPEN flag,
720 rather than connect() to send data in SYN.
722 The server support is enabled by flag 0x2 (off by default). Then
723 either enable for all listeners with another flag (0x400) or
724 enable individual listeners via TCP_FASTOPEN socket option with
725 the option value being the length of the syn-data backlog.
727 The values (bitmap) are
729 ===== ======== ======================================================
730 0x1 (client) enables sending data in the opening SYN on the client.
731 0x2 (server) enables the server support, i.e., allowing data in
732 a SYN packet to be accepted and passed to the
733 application before 3-way handshake finishes.
734 0x4 (client) send data in the opening SYN regardless of cookie
735 availability and without a cookie option.
736 0x200 (server) accept data-in-SYN w/o any cookie option present.
737 0x400 (server) enable all listeners to support Fast Open by
738 default without explicit TCP_FASTOPEN socket option.
739 ===== ======== ======================================================
743 Note that additional client or server features are only
744 effective if the basic support (0x1 and 0x2) are enabled respectively.
746 tcp_fastopen_blackhole_timeout_sec - INTEGER
747 Initial time period in second to disable Fastopen on active TCP sockets
748 when a TFO firewall blackhole issue happens.
749 This time period will grow exponentially when more blackhole issues
750 get detected right after Fastopen is re-enabled and will reset to
751 initial value when the blackhole issue goes away.
752 0 to disable the blackhole detection.
754 By default, it is set to 0 (feature is disabled).
756 tcp_fastopen_key - list of comma separated 32-digit hexadecimal INTEGERs
757 The list consists of a primary key and an optional backup key. The
758 primary key is used for both creating and validating cookies, while the
759 optional backup key is only used for validating cookies. The purpose of
760 the backup key is to maximize TFO validation when keys are rotated.
762 A randomly chosen primary key may be configured by the kernel if
763 the tcp_fastopen sysctl is set to 0x400 (see above), or if the
764 TCP_FASTOPEN setsockopt() optname is set and a key has not been
765 previously configured via sysctl. If keys are configured via
766 setsockopt() by using the TCP_FASTOPEN_KEY optname, then those
767 per-socket keys will be used instead of any keys that are specified via
770 A key is specified as 4 8-digit hexadecimal integers which are separated
771 by a '-' as: xxxxxxxx-xxxxxxxx-xxxxxxxx-xxxxxxxx. Leading zeros may be
772 omitted. A primary and a backup key may be specified by separating them
773 by a comma. If only one key is specified, it becomes the primary key and
774 any previously configured backup keys are removed.
776 tcp_syn_retries - INTEGER
777 Number of times initial SYNs for an active TCP connection attempt
778 will be retransmitted. Should not be higher than 127. Default value
779 is 6, which corresponds to 63seconds till the last retransmission
780 with the current initial RTO of 1second. With this the final timeout
781 for an active TCP connection attempt will happen after 127seconds.
783 tcp_timestamps - INTEGER
784 Enable timestamps as defined in RFC1323.
787 - 1: Enable timestamps as defined in RFC1323 and use random offset for
788 each connection rather than only using the current time.
789 - 2: Like 1, but without random offsets.
793 tcp_min_tso_segs - INTEGER
794 Minimal number of segments per TSO frame.
796 Since linux-3.12, TCP does an automatic sizing of TSO frames,
797 depending on flow rate, instead of filling 64Kbytes packets.
798 For specific usages, it's possible to force TCP to build big
799 TSO frames. Note that TCP stack might split too big TSO packets
800 if available window is too small.
804 tcp_pacing_ss_ratio - INTEGER
805 sk->sk_pacing_rate is set by TCP stack using a ratio applied
806 to current rate. (current_rate = cwnd * mss / srtt)
807 If TCP is in slow start, tcp_pacing_ss_ratio is applied
808 to let TCP probe for bigger speeds, assuming cwnd can be
809 doubled every other RTT.
813 tcp_pacing_ca_ratio - INTEGER
814 sk->sk_pacing_rate is set by TCP stack using a ratio applied
815 to current rate. (current_rate = cwnd * mss / srtt)
816 If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio
817 is applied to conservatively probe for bigger throughput.
821 tcp_tso_win_divisor - INTEGER
822 This allows control over what percentage of the congestion window
823 can be consumed by a single TSO frame.
824 The setting of this parameter is a choice between burstiness and
825 building larger TSO frames.
829 tcp_tw_reuse - INTEGER
830 Enable reuse of TIME-WAIT sockets for new connections when it is
831 safe from protocol viewpoint.
835 - 2 - enable for loopback traffic only
837 It should not be changed without advice/request of technical
842 tcp_window_scaling - BOOLEAN
843 Enable window scaling as defined in RFC1323.
845 tcp_wmem - vector of 3 INTEGERs: min, default, max
846 min: Amount of memory reserved for send buffers for TCP sockets.
847 Each TCP socket has rights to use it due to fact of its birth.
851 default: initial size of send buffer used by TCP sockets. This
852 value overrides net.core.wmem_default used by other protocols.
854 It is usually lower than net.core.wmem_default.
858 max: Maximal amount of memory allowed for automatically tuned
859 send buffers for TCP sockets. This value does not override
860 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
861 automatic tuning of that socket's send buffer size, in which case
862 this value is ignored.
864 Default: between 64K and 4MB, depending on RAM size.
866 tcp_notsent_lowat - UNSIGNED INTEGER
867 A TCP socket can control the amount of unsent bytes in its write queue,
868 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
869 reports POLLOUT events if the amount of unsent bytes is below a per
870 socket value, and if the write queue is not full. sendmsg() will
871 also not add new buffers if the limit is hit.
873 This global variable controls the amount of unsent data for
874 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
875 to the global variable has immediate effect.
877 Default: UINT_MAX (0xFFFFFFFF)
879 tcp_workaround_signed_windows - BOOLEAN
880 If set, assume no receipt of a window scaling option means the
881 remote TCP is broken and treats the window as a signed quantity.
882 If unset, assume the remote TCP is not broken even if we do
883 not receive a window scaling option from them.
887 tcp_thin_linear_timeouts - BOOLEAN
888 Enable dynamic triggering of linear timeouts for thin streams.
889 If set, a check is performed upon retransmission by timeout to
890 determine if the stream is thin (less than 4 packets in flight).
891 As long as the stream is found to be thin, up to 6 linear
892 timeouts may be performed before exponential backoff mode is
893 initiated. This improves retransmission latency for
894 non-aggressive thin streams, often found to be time-dependent.
895 For more information on thin streams, see
896 Documentation/networking/tcp-thin.rst
900 tcp_limit_output_bytes - INTEGER
901 Controls TCP Small Queue limit per tcp socket.
902 TCP bulk sender tends to increase packets in flight until it
903 gets losses notifications. With SNDBUF autotuning, this can
904 result in a large amount of packets queued on the local machine
905 (e.g.: qdiscs, CPU backlog, or device) hurting latency of other
906 flows, for typical pfifo_fast qdiscs. tcp_limit_output_bytes
907 limits the number of bytes on qdisc or device to reduce artificial
908 RTT/cwnd and reduce bufferbloat.
910 Default: 1048576 (16 * 65536)
912 tcp_challenge_ack_limit - INTEGER
913 Limits number of Challenge ACK sent per second, as recommended
914 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
917 tcp_rx_skb_cache - BOOLEAN
918 Controls a per TCP socket cache of one skb, that might help
919 performance of some workloads. This might be dangerous
920 on systems with a lot of TCP sockets, since it increases
923 Default: 0 (disabled)
928 udp_l3mdev_accept - BOOLEAN
929 Enabling this option allows a "global" bound socket to work
930 across L3 master domains (e.g., VRFs) with packets capable of
931 being received regardless of the L3 domain in which they
932 originated. Only valid when the kernel was compiled with
933 CONFIG_NET_L3_MASTER_DEV.
935 Default: 0 (disabled)
937 udp_mem - vector of 3 INTEGERs: min, pressure, max
938 Number of pages allowed for queueing by all UDP sockets.
940 min: Below this number of pages UDP is not bothered about its
941 memory appetite. When amount of memory allocated by UDP exceeds
942 this number, UDP starts to moderate memory usage.
944 pressure: This value was introduced to follow format of tcp_mem.
946 max: Number of pages allowed for queueing by all UDP sockets.
948 Default is calculated at boot time from amount of available memory.
950 udp_rmem_min - INTEGER
951 Minimal size of receive buffer used by UDP sockets in moderation.
952 Each UDP socket is able to use the size for receiving data, even if
953 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
957 udp_wmem_min - INTEGER
958 Minimal size of send buffer used by UDP sockets in moderation.
959 Each UDP socket is able to use the size for sending data, even if
960 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
967 raw_l3mdev_accept - BOOLEAN
968 Enabling this option allows a "global" bound socket to work
969 across L3 master domains (e.g., VRFs) with packets capable of
970 being received regardless of the L3 domain in which they
971 originated. Only valid when the kernel was compiled with
972 CONFIG_NET_L3_MASTER_DEV.
979 cipso_cache_enable - BOOLEAN
980 If set, enable additions to and lookups from the CIPSO label mapping
981 cache. If unset, additions are ignored and lookups always result in a
982 miss. However, regardless of the setting the cache is still
983 invalidated when required when means you can safely toggle this on and
984 off and the cache will always be "safe".
988 cipso_cache_bucket_size - INTEGER
989 The CIPSO label cache consists of a fixed size hash table with each
990 hash bucket containing a number of cache entries. This variable limits
991 the number of entries in each hash bucket; the larger the value the
992 more CIPSO label mappings that can be cached. When the number of
993 entries in a given hash bucket reaches this limit adding new entries
994 causes the oldest entry in the bucket to be removed to make room.
998 cipso_rbm_optfmt - BOOLEAN
999 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
1000 the CIPSO draft specification (see Documentation/netlabel for details).
1001 This means that when set the CIPSO tag will be padded with empty
1002 categories in order to make the packet data 32-bit aligned.
1006 cipso_rbm_structvalid - BOOLEAN
1007 If set, do a very strict check of the CIPSO option when
1008 ip_options_compile() is called. If unset, relax the checks done during
1009 ip_options_compile(). Either way is "safe" as errors are caught else
1010 where in the CIPSO processing code but setting this to 0 (False) should
1011 result in less work (i.e. it should be faster) but could cause problems
1012 with other implementations that require strict checking.
1019 ip_local_port_range - 2 INTEGERS
1020 Defines the local port range that is used by TCP and UDP to
1021 choose the local port. The first number is the first, the
1022 second the last local port number.
1023 If possible, it is better these numbers have different parity
1024 (one even and one odd value).
1025 Must be greater than or equal to ip_unprivileged_port_start.
1026 The default values are 32768 and 60999 respectively.
1028 ip_local_reserved_ports - list of comma separated ranges
1029 Specify the ports which are reserved for known third-party
1030 applications. These ports will not be used by automatic port
1031 assignments (e.g. when calling connect() or bind() with port
1032 number 0). Explicit port allocation behavior is unchanged.
1034 The format used for both input and output is a comma separated
1035 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
1036 10). Writing to the file will clear all previously reserved
1037 ports and update the current list with the one given in the
1040 Note that ip_local_port_range and ip_local_reserved_ports
1041 settings are independent and both are considered by the kernel
1042 when determining which ports are available for automatic port
1045 You can reserve ports which are not in the current
1046 ip_local_port_range, e.g.::
1048 $ cat /proc/sys/net/ipv4/ip_local_port_range
1050 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
1053 although this is redundant. However such a setting is useful
1054 if later the port range is changed to a value that will
1055 include the reserved ports.
1059 ip_unprivileged_port_start - INTEGER
1060 This is a per-namespace sysctl. It defines the first
1061 unprivileged port in the network namespace. Privileged ports
1062 require root or CAP_NET_BIND_SERVICE in order to bind to them.
1063 To disable all privileged ports, set this to 0. They must not
1064 overlap with the ip_local_port_range.
1068 ip_nonlocal_bind - BOOLEAN
1069 If set, allows processes to bind() to non-local IP addresses,
1070 which can be quite useful - but may break some applications.
1074 ip_autobind_reuse - BOOLEAN
1075 By default, bind() does not select the ports automatically even if
1076 the new socket and all sockets bound to the port have SO_REUSEADDR.
1077 ip_autobind_reuse allows bind() to reuse the port and this is useful
1078 when you use bind()+connect(), but may break some applications.
1079 The preferred solution is to use IP_BIND_ADDRESS_NO_PORT and this
1080 option should only be set by experts.
1083 ip_dynaddr - BOOLEAN
1084 If set non-zero, enables support for dynamic addresses.
1085 If set to a non-zero value larger than 1, a kernel log
1086 message will be printed when dynamic address rewriting
1091 ip_early_demux - BOOLEAN
1092 Optimize input packet processing down to one demux for
1093 certain kinds of local sockets. Currently we only do this
1094 for established TCP and connected UDP sockets.
1096 It may add an additional cost for pure routing workloads that
1097 reduces overall throughput, in such case you should disable it.
1101 ping_group_range - 2 INTEGERS
1102 Restrict ICMP_PROTO datagram sockets to users in the group range.
1103 The default is "1 0", meaning, that nobody (not even root) may
1104 create ping sockets. Setting it to "100 100" would grant permissions
1105 to the single group. "0 4294967295" would enable it for the world, "100
1106 4294967295" would enable it for the users, but not daemons.
1108 tcp_early_demux - BOOLEAN
1109 Enable early demux for established TCP sockets.
1113 udp_early_demux - BOOLEAN
1114 Enable early demux for connected UDP sockets. Disable this if
1115 your system could experience more unconnected load.
1119 icmp_echo_ignore_all - BOOLEAN
1120 If set non-zero, then the kernel will ignore all ICMP ECHO
1121 requests sent to it.
1125 icmp_echo_ignore_broadcasts - BOOLEAN
1126 If set non-zero, then the kernel will ignore all ICMP ECHO and
1127 TIMESTAMP requests sent to it via broadcast/multicast.
1131 icmp_ratelimit - INTEGER
1132 Limit the maximal rates for sending ICMP packets whose type matches
1133 icmp_ratemask (see below) to specific targets.
1134 0 to disable any limiting,
1135 otherwise the minimal space between responses in milliseconds.
1136 Note that another sysctl, icmp_msgs_per_sec limits the number
1137 of ICMP packets sent on all targets.
1141 icmp_msgs_per_sec - INTEGER
1142 Limit maximal number of ICMP packets sent per second from this host.
1143 Only messages whose type matches icmp_ratemask (see below) are
1144 controlled by this limit. For security reasons, the precise count
1145 of messages per second is randomized.
1149 icmp_msgs_burst - INTEGER
1150 icmp_msgs_per_sec controls number of ICMP packets sent per second,
1151 while icmp_msgs_burst controls the burst size of these packets.
1152 For security reasons, the precise burst size is randomized.
1156 icmp_ratemask - INTEGER
1157 Mask made of ICMP types for which rates are being limited.
1159 Significant bits: IHGFEDCBA9876543210
1161 Default mask: 0000001100000011000 (6168)
1163 Bit definitions (see include/linux/icmp.h):
1165 = =========================
1167 3 Destination Unreachable [1]_
1168 4 Source Quench [1]_
1171 B Time Exceeded [1]_
1172 C Parameter Problem [1]_
1177 H Address Mask Request
1178 I Address Mask Reply
1179 = =========================
1181 .. [1] These are rate limited by default (see default mask above)
1183 icmp_ignore_bogus_error_responses - BOOLEAN
1184 Some routers violate RFC1122 by sending bogus responses to broadcast
1185 frames. Such violations are normally logged via a kernel warning.
1186 If this is set to TRUE, the kernel will not give such warnings, which
1187 will avoid log file clutter.
1191 icmp_errors_use_inbound_ifaddr - BOOLEAN
1193 If zero, icmp error messages are sent with the primary address of
1194 the exiting interface.
1196 If non-zero, the message will be sent with the primary address of
1197 the interface that received the packet that caused the icmp error.
1198 This is the behaviour network many administrators will expect from
1199 a router. And it can make debugging complicated network layouts
1202 Note that if no primary address exists for the interface selected,
1203 then the primary address of the first non-loopback interface that
1204 has one will be used regardless of this setting.
1208 igmp_max_memberships - INTEGER
1209 Change the maximum number of multicast groups we can subscribe to.
1212 Theoretical maximum value is bounded by having to send a membership
1213 report in a single datagram (i.e. the report can't span multiple
1214 datagrams, or risk confusing the switch and leaving groups you don't
1217 The number of supported groups 'M' is bounded by the number of group
1218 report entries you can fit into a single datagram of 65535 bytes.
1220 M = 65536-sizeof (ip header)/(sizeof(Group record))
1222 Group records are variable length, with a minimum of 12 bytes.
1223 So net.ipv4.igmp_max_memberships should not be set higher than:
1225 (65536-24) / 12 = 5459
1227 The value 5459 assumes no IP header options, so in practice
1228 this number may be lower.
1230 igmp_max_msf - INTEGER
1231 Maximum number of addresses allowed in the source filter list for a
1237 Controls the IGMP query robustness variable (see RFC2236 8.1).
1239 Default: 2 (as specified by RFC2236 8.1)
1241 Minimum: 1 (as specified by RFC6636 4.5)
1243 force_igmp_version - INTEGER
1244 - 0 - (default) No enforcement of a IGMP version, IGMPv1/v2 fallback
1245 allowed. Will back to IGMPv3 mode again if all IGMPv1/v2 Querier
1246 Present timer expires.
1247 - 1 - Enforce to use IGMP version 1. Will also reply IGMPv1 report if
1248 receive IGMPv2/v3 query.
1249 - 2 - Enforce to use IGMP version 2. Will fallback to IGMPv1 if receive
1250 IGMPv1 query message. Will reply report if receive IGMPv3 query.
1251 - 3 - Enforce to use IGMP version 3. The same react with default 0.
1255 this is not the same with force_mld_version because IGMPv3 RFC3376
1256 Security Considerations does not have clear description that we could
1257 ignore other version messages completely as MLDv2 RFC3810. So make
1258 this value as default 0 is recommended.
1260 ``conf/interface/*``
1261 changes special settings per interface (where
1262 interface" is the name of your network interface)
1265 is special, changes the settings for all interfaces
1267 log_martians - BOOLEAN
1268 Log packets with impossible addresses to kernel log.
1269 log_martians for the interface will be enabled if at least one of
1270 conf/{all,interface}/log_martians is set to TRUE,
1271 it will be disabled otherwise
1273 accept_redirects - BOOLEAN
1274 Accept ICMP redirect messages.
1275 accept_redirects for the interface will be enabled if:
1277 - both conf/{all,interface}/accept_redirects are TRUE in the case
1278 forwarding for the interface is enabled
1282 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
1283 case forwarding for the interface is disabled
1285 accept_redirects for the interface will be disabled otherwise
1292 forwarding - BOOLEAN
1293 Enable IP forwarding on this interface. This controls whether packets
1294 received _on_ this interface can be forwarded.
1296 mc_forwarding - BOOLEAN
1297 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
1298 and a multicast routing daemon is required.
1299 conf/all/mc_forwarding must also be set to TRUE to enable multicast
1300 routing for the interface
1303 Integer value used to differentiate the devices by the medium they
1304 are attached to. Two devices can have different id values when
1305 the broadcast packets are received only on one of them.
1306 The default value 0 means that the device is the only interface
1307 to its medium, value of -1 means that medium is not known.
1309 Currently, it is used to change the proxy_arp behavior:
1310 the proxy_arp feature is enabled for packets forwarded between
1311 two devices attached to different media.
1316 proxy_arp for the interface will be enabled if at least one of
1317 conf/{all,interface}/proxy_arp is set to TRUE,
1318 it will be disabled otherwise
1320 proxy_arp_pvlan - BOOLEAN
1321 Private VLAN proxy arp.
1323 Basically allow proxy arp replies back to the same interface
1324 (from which the ARP request/solicitation was received).
1326 This is done to support (ethernet) switch features, like RFC
1327 3069, where the individual ports are NOT allowed to
1328 communicate with each other, but they are allowed to talk to
1329 the upstream router. As described in RFC 3069, it is possible
1330 to allow these hosts to communicate through the upstream
1331 router by proxy_arp'ing. Don't need to be used together with
1334 This technology is known by different names:
1336 In RFC 3069 it is called VLAN Aggregation.
1337 Cisco and Allied Telesyn call it Private VLAN.
1338 Hewlett-Packard call it Source-Port filtering or port-isolation.
1339 Ericsson call it MAC-Forced Forwarding (RFC Draft).
1341 shared_media - BOOLEAN
1342 Send(router) or accept(host) RFC1620 shared media redirects.
1343 Overrides secure_redirects.
1345 shared_media for the interface will be enabled if at least one of
1346 conf/{all,interface}/shared_media is set to TRUE,
1347 it will be disabled otherwise
1351 secure_redirects - BOOLEAN
1352 Accept ICMP redirect messages only to gateways listed in the
1353 interface's current gateway list. Even if disabled, RFC1122 redirect
1356 Overridden by shared_media.
1358 secure_redirects for the interface will be enabled if at least one of
1359 conf/{all,interface}/secure_redirects is set to TRUE,
1360 it will be disabled otherwise
1364 send_redirects - BOOLEAN
1365 Send redirects, if router.
1367 send_redirects for the interface will be enabled if at least one of
1368 conf/{all,interface}/send_redirects is set to TRUE,
1369 it will be disabled otherwise
1373 bootp_relay - BOOLEAN
1374 Accept packets with source address 0.b.c.d destined
1375 not to this host as local ones. It is supposed, that
1376 BOOTP relay daemon will catch and forward such packets.
1377 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
1382 Not Implemented Yet.
1384 accept_source_route - BOOLEAN
1385 Accept packets with SRR option.
1386 conf/all/accept_source_route must also be set to TRUE to accept packets
1387 with SRR option on the interface
1394 accept_local - BOOLEAN
1395 Accept packets with local source addresses. In combination with
1396 suitable routing, this can be used to direct packets between two
1397 local interfaces over the wire and have them accepted properly.
1400 route_localnet - BOOLEAN
1401 Do not consider loopback addresses as martian source or destination
1402 while routing. This enables the use of 127/8 for local routing purposes.
1407 - 0 - No source validation.
1408 - 1 - Strict mode as defined in RFC3704 Strict Reverse Path
1409 Each incoming packet is tested against the FIB and if the interface
1410 is not the best reverse path the packet check will fail.
1411 By default failed packets are discarded.
1412 - 2 - Loose mode as defined in RFC3704 Loose Reverse Path
1413 Each incoming packet's source address is also tested against the FIB
1414 and if the source address is not reachable via any interface
1415 the packet check will fail.
1417 Current recommended practice in RFC3704 is to enable strict mode
1418 to prevent IP spoofing from DDos attacks. If using asymmetric routing
1419 or other complicated routing, then loose mode is recommended.
1421 The max value from conf/{all,interface}/rp_filter is used
1422 when doing source validation on the {interface}.
1424 Default value is 0. Note that some distributions enable it
1427 arp_filter - BOOLEAN
1428 - 1 - Allows you to have multiple network interfaces on the same
1429 subnet, and have the ARPs for each interface be answered
1430 based on whether or not the kernel would route a packet from
1431 the ARP'd IP out that interface (therefore you must use source
1432 based routing for this to work). In other words it allows control
1433 of which cards (usually 1) will respond to an arp request.
1435 - 0 - (default) The kernel can respond to arp requests with addresses
1436 from other interfaces. This may seem wrong but it usually makes
1437 sense, because it increases the chance of successful communication.
1438 IP addresses are owned by the complete host on Linux, not by
1439 particular interfaces. Only for more complex setups like load-
1440 balancing, does this behaviour cause problems.
1442 arp_filter for the interface will be enabled if at least one of
1443 conf/{all,interface}/arp_filter is set to TRUE,
1444 it will be disabled otherwise
1446 arp_announce - INTEGER
1447 Define different restriction levels for announcing the local
1448 source IP address from IP packets in ARP requests sent on
1451 - 0 - (default) Use any local address, configured on any interface
1452 - 1 - Try to avoid local addresses that are not in the target's
1453 subnet for this interface. This mode is useful when target
1454 hosts reachable via this interface require the source IP
1455 address in ARP requests to be part of their logical network
1456 configured on the receiving interface. When we generate the
1457 request we will check all our subnets that include the
1458 target IP and will preserve the source address if it is from
1459 such subnet. If there is no such subnet we select source
1460 address according to the rules for level 2.
1461 - 2 - Always use the best local address for this target.
1462 In this mode we ignore the source address in the IP packet
1463 and try to select local address that we prefer for talks with
1464 the target host. Such local address is selected by looking
1465 for primary IP addresses on all our subnets on the outgoing
1466 interface that include the target IP address. If no suitable
1467 local address is found we select the first local address
1468 we have on the outgoing interface or on all other interfaces,
1469 with the hope we will receive reply for our request and
1470 even sometimes no matter the source IP address we announce.
1472 The max value from conf/{all,interface}/arp_announce is used.
1474 Increasing the restriction level gives more chance for
1475 receiving answer from the resolved target while decreasing
1476 the level announces more valid sender's information.
1478 arp_ignore - INTEGER
1479 Define different modes for sending replies in response to
1480 received ARP requests that resolve local target IP addresses:
1482 - 0 - (default): reply for any local target IP address, configured
1484 - 1 - reply only if the target IP address is local address
1485 configured on the incoming interface
1486 - 2 - reply only if the target IP address is local address
1487 configured on the incoming interface and both with the
1488 sender's IP address are part from same subnet on this interface
1489 - 3 - do not reply for local addresses configured with scope host,
1490 only resolutions for global and link addresses are replied
1492 - 8 - do not reply for all local addresses
1494 The max value from conf/{all,interface}/arp_ignore is used
1495 when ARP request is received on the {interface}
1497 arp_notify - BOOLEAN
1498 Define mode for notification of address and device changes.
1500 == ==========================================================
1501 0 (default): do nothing
1502 1 Generate gratuitous arp requests when device is brought up
1503 or hardware address changes.
1504 == ==========================================================
1506 arp_accept - BOOLEAN
1507 Define behavior for gratuitous ARP frames who's IP is not
1508 already present in the ARP table:
1510 - 0 - don't create new entries in the ARP table
1511 - 1 - create new entries in the ARP table
1513 Both replies and requests type gratuitous arp will trigger the
1514 ARP table to be updated, if this setting is on.
1516 If the ARP table already contains the IP address of the
1517 gratuitous arp frame, the arp table will be updated regardless
1518 if this setting is on or off.
1520 mcast_solicit - INTEGER
1521 The maximum number of multicast probes in INCOMPLETE state,
1522 when the associated hardware address is unknown. Defaults
1525 ucast_solicit - INTEGER
1526 The maximum number of unicast probes in PROBE state, when
1527 the hardware address is being reconfirmed. Defaults to 3.
1529 app_solicit - INTEGER
1530 The maximum number of probes to send to the user space ARP daemon
1531 via netlink before dropping back to multicast probes (see
1532 mcast_resolicit). Defaults to 0.
1534 mcast_resolicit - INTEGER
1535 The maximum number of multicast probes after unicast and
1536 app probes in PROBE state. Defaults to 0.
1538 disable_policy - BOOLEAN
1539 Disable IPSEC policy (SPD) for this interface
1541 disable_xfrm - BOOLEAN
1542 Disable IPSEC encryption on this interface, whatever the policy
1544 igmpv2_unsolicited_report_interval - INTEGER
1545 The interval in milliseconds in which the next unsolicited
1546 IGMPv1 or IGMPv2 report retransmit will take place.
1548 Default: 10000 (10 seconds)
1550 igmpv3_unsolicited_report_interval - INTEGER
1551 The interval in milliseconds in which the next unsolicited
1552 IGMPv3 report retransmit will take place.
1554 Default: 1000 (1 seconds)
1556 promote_secondaries - BOOLEAN
1557 When a primary IP address is removed from this interface
1558 promote a corresponding secondary IP address instead of
1559 removing all the corresponding secondary IP addresses.
1561 drop_unicast_in_l2_multicast - BOOLEAN
1562 Drop any unicast IP packets that are received in link-layer
1563 multicast (or broadcast) frames.
1565 This behavior (for multicast) is actually a SHOULD in RFC
1566 1122, but is disabled by default for compatibility reasons.
1570 drop_gratuitous_arp - BOOLEAN
1571 Drop all gratuitous ARP frames, for example if there's a known
1572 good ARP proxy on the network and such frames need not be used
1573 (or in the case of 802.11, must not be used to prevent attacks.)
1579 Allows you to write a number, which can be used as required.
1583 xfrm4_gc_thresh - INTEGER
1584 (Obsolete since linux-4.14)
1585 The threshold at which we will start garbage collecting for IPv4
1586 destination cache entries. At twice this value the system will
1587 refuse new allocations.
1589 igmp_link_local_mcast_reports - BOOLEAN
1590 Enable IGMP reports for link local multicast groups in the
1596 kuznet@ms2.inr.ac.ru
1603 delon.nicolas@wanadoo.fr
1608 /proc/sys/net/ipv6/* Variables
1609 ==============================
1611 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1612 apply to IPv6 [XXX?].
1614 bindv6only - BOOLEAN
1615 Default value for IPV6_V6ONLY socket option,
1616 which restricts use of the IPv6 socket to IPv6 communication
1619 - TRUE: disable IPv4-mapped address feature
1620 - FALSE: enable IPv4-mapped address feature
1622 Default: FALSE (as specified in RFC3493)
1624 flowlabel_consistency - BOOLEAN
1625 Protect the consistency (and unicity) of flow label.
1626 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1634 auto_flowlabels - INTEGER
1635 Automatically generate flow labels based on a flow hash of the
1636 packet. This allows intermediate devices, such as routers, to
1637 identify packet flows for mechanisms like Equal Cost Multipath
1638 Routing (see RFC 6438).
1640 = ===========================================================
1641 0 automatic flow labels are completely disabled
1642 1 automatic flow labels are enabled by default, they can be
1643 disabled on a per socket basis using the IPV6_AUTOFLOWLABEL
1645 2 automatic flow labels are allowed, they may be enabled on a
1646 per socket basis using the IPV6_AUTOFLOWLABEL socket option
1647 3 automatic flow labels are enabled and enforced, they cannot
1648 be disabled by the socket option
1649 = ===========================================================
1653 flowlabel_state_ranges - BOOLEAN
1654 Split the flow label number space into two ranges. 0-0x7FFFF is
1655 reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF
1656 is reserved for stateless flow labels as described in RFC6437.
1663 flowlabel_reflect - INTEGER
1664 Control flow label reflection. Needed for Path MTU
1665 Discovery to work with Equal Cost Multipath Routing in anycast
1666 environments. See RFC 7690 and:
1667 https://tools.ietf.org/html/draft-wang-6man-flow-label-reflection-01
1671 - 1: enabled for established flows
1673 Note that this prevents automatic flowlabel changes, as done
1674 in "tcp: change IPv6 flow-label upon receiving spurious retransmission"
1675 and "tcp: Change txhash on every SYN and RTO retransmit"
1677 - 2: enabled for TCP RESET packets (no active listener)
1678 If set, a RST packet sent in response to a SYN packet on a closed
1679 port will reflect the incoming flow label.
1681 - 4: enabled for ICMPv6 echo reply messages.
1685 fib_multipath_hash_policy - INTEGER
1686 Controls which hash policy to use for multipath routes.
1688 Default: 0 (Layer 3)
1692 - 0 - Layer 3 (source and destination addresses plus flow label)
1693 - 1 - Layer 4 (standard 5-tuple)
1694 - 2 - Layer 3 or inner Layer 3 if present
1696 anycast_src_echo_reply - BOOLEAN
1697 Controls the use of anycast addresses as source addresses for ICMPv6
1705 idgen_delay - INTEGER
1706 Controls the delay in seconds after which time to retry
1707 privacy stable address generation if a DAD conflict is
1710 Default: 1 (as specified in RFC7217)
1712 idgen_retries - INTEGER
1713 Controls the number of retries to generate a stable privacy
1714 address if a DAD conflict is detected.
1716 Default: 3 (as specified in RFC7217)
1719 Controls the MLD query robustness variable (see RFC3810 9.1).
1721 Default: 2 (as specified by RFC3810 9.1)
1723 Minimum: 1 (as specified by RFC6636 4.5)
1725 max_dst_opts_number - INTEGER
1726 Maximum number of non-padding TLVs allowed in a Destination
1727 options extension header. If this value is less than zero
1728 then unknown options are disallowed and the number of known
1729 TLVs allowed is the absolute value of this number.
1733 max_hbh_opts_number - INTEGER
1734 Maximum number of non-padding TLVs allowed in a Hop-by-Hop
1735 options extension header. If this value is less than zero
1736 then unknown options are disallowed and the number of known
1737 TLVs allowed is the absolute value of this number.
1741 max_dst_opts_length - INTEGER
1742 Maximum length allowed for a Destination options extension
1745 Default: INT_MAX (unlimited)
1747 max_hbh_length - INTEGER
1748 Maximum length allowed for a Hop-by-Hop options extension
1751 Default: INT_MAX (unlimited)
1753 skip_notify_on_dev_down - BOOLEAN
1754 Controls whether an RTM_DELROUTE message is generated for routes
1755 removed when a device is taken down or deleted. IPv4 does not
1756 generate this message; IPv6 does by default. Setting this sysctl
1757 to true skips the message, making IPv4 and IPv6 on par in relying
1758 on userspace caches to track link events and evict routes.
1760 Default: false (generate message)
1762 nexthop_compat_mode - BOOLEAN
1763 New nexthop API provides a means for managing nexthops independent of
1764 prefixes. Backwards compatibilty with old route format is enabled by
1765 default which means route dumps and notifications contain the new
1766 nexthop attribute but also the full, expanded nexthop definition.
1767 Further, updates or deletes of a nexthop configuration generate route
1768 notifications for each fib entry using the nexthop. Once a system
1769 understands the new API, this sysctl can be disabled to achieve full
1770 performance benefits of the new API by disabling the nexthop expansion
1771 and extraneous notifications.
1772 Default: true (backward compat mode)
1776 ip6frag_high_thresh - INTEGER
1777 Maximum memory used to reassemble IPv6 fragments. When
1778 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1779 the fragment handler will toss packets until ip6frag_low_thresh
1782 ip6frag_low_thresh - INTEGER
1783 See ip6frag_high_thresh
1785 ip6frag_time - INTEGER
1786 Time in seconds to keep an IPv6 fragment in memory.
1788 IPv6 Segment Routing:
1790 seg6_flowlabel - INTEGER
1791 Controls the behaviour of computing the flowlabel of outer
1792 IPv6 header in case of SR T.encaps
1794 == =======================================================
1795 -1 set flowlabel to zero.
1796 0 copy flowlabel from Inner packet in case of Inner IPv6
1797 (Set flowlabel to 0 in case IPv4/L2)
1798 1 Compute the flowlabel using seg6_make_flowlabel()
1799 == =======================================================
1804 Change the interface-specific default settings.
1808 Change all the interface-specific settings.
1810 [XXX: Other special features than forwarding?]
1812 conf/all/forwarding - BOOLEAN
1813 Enable global IPv6 forwarding between all interfaces.
1815 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1816 to control which interfaces may forward packets and which not.
1818 This also sets all interfaces' Host/Router setting
1819 'forwarding' to the specified value. See below for details.
1821 This referred to as global forwarding.
1826 fwmark_reflect - BOOLEAN
1827 Controls the fwmark of kernel-generated IPv6 reply packets that are not
1828 associated with a socket for example, TCP RSTs or ICMPv6 echo replies).
1829 If unset, these packets have a fwmark of zero. If set, they have the
1830 fwmark of the packet they are replying to.
1834 ``conf/interface/*``:
1835 Change special settings per interface.
1837 The functional behaviour for certain settings is different
1838 depending on whether local forwarding is enabled or not.
1841 Accept Router Advertisements; autoconfigure using them.
1843 It also determines whether or not to transmit Router
1844 Solicitations. If and only if the functional setting is to
1845 accept Router Advertisements, Router Solicitations will be
1848 Possible values are:
1850 == ===========================================================
1851 0 Do not accept Router Advertisements.
1852 1 Accept Router Advertisements if forwarding is disabled.
1853 2 Overrule forwarding behaviour. Accept Router Advertisements
1854 even if forwarding is enabled.
1855 == ===========================================================
1859 - enabled if local forwarding is disabled.
1860 - disabled if local forwarding is enabled.
1862 accept_ra_defrtr - BOOLEAN
1863 Learn default router in Router Advertisement.
1867 - enabled if accept_ra is enabled.
1868 - disabled if accept_ra is disabled.
1870 accept_ra_from_local - BOOLEAN
1871 Accept RA with source-address that is found on local machine
1872 if the RA is otherwise proper and able to be accepted.
1874 Default is to NOT accept these as it may be an un-intended
1879 - enabled if accept_ra_from_local is enabled
1880 on a specific interface.
1881 - disabled if accept_ra_from_local is disabled
1882 on a specific interface.
1884 accept_ra_min_hop_limit - INTEGER
1885 Minimum hop limit Information in Router Advertisement.
1887 Hop limit Information in Router Advertisement less than this
1888 variable shall be ignored.
1892 accept_ra_pinfo - BOOLEAN
1893 Learn Prefix Information in Router Advertisement.
1897 - enabled if accept_ra is enabled.
1898 - disabled if accept_ra is disabled.
1900 accept_ra_rt_info_min_plen - INTEGER
1901 Minimum prefix length of Route Information in RA.
1903 Route Information w/ prefix smaller than this variable shall
1908 * 0 if accept_ra_rtr_pref is enabled.
1909 * -1 if accept_ra_rtr_pref is disabled.
1911 accept_ra_rt_info_max_plen - INTEGER
1912 Maximum prefix length of Route Information in RA.
1914 Route Information w/ prefix larger than this variable shall
1919 * 0 if accept_ra_rtr_pref is enabled.
1920 * -1 if accept_ra_rtr_pref is disabled.
1922 accept_ra_rtr_pref - BOOLEAN
1923 Accept Router Preference in RA.
1927 - enabled if accept_ra is enabled.
1928 - disabled if accept_ra is disabled.
1930 accept_ra_mtu - BOOLEAN
1931 Apply the MTU value specified in RA option 5 (RFC4861). If
1932 disabled, the MTU specified in the RA will be ignored.
1936 - enabled if accept_ra is enabled.
1937 - disabled if accept_ra is disabled.
1939 accept_redirects - BOOLEAN
1944 - enabled if local forwarding is disabled.
1945 - disabled if local forwarding is enabled.
1947 accept_source_route - INTEGER
1948 Accept source routing (routing extension header).
1950 - >= 0: Accept only routing header type 2.
1951 - < 0: Do not accept routing header.
1956 Autoconfigure addresses using Prefix Information in Router
1961 - enabled if accept_ra_pinfo is enabled.
1962 - disabled if accept_ra_pinfo is disabled.
1964 dad_transmits - INTEGER
1965 The amount of Duplicate Address Detection probes to send.
1969 forwarding - INTEGER
1970 Configure interface-specific Host/Router behaviour.
1974 It is recommended to have the same setting on all
1975 interfaces; mixed router/host scenarios are rather uncommon.
1977 Possible values are:
1979 - 0 Forwarding disabled
1980 - 1 Forwarding enabled
1984 By default, Host behaviour is assumed. This means:
1986 1. IsRouter flag is not set in Neighbour Advertisements.
1987 2. If accept_ra is TRUE (default), transmit Router
1989 3. If accept_ra is TRUE (default), accept Router
1990 Advertisements (and do autoconfiguration).
1991 4. If accept_redirects is TRUE (default), accept Redirects.
1995 If local forwarding is enabled, Router behaviour is assumed.
1996 This means exactly the reverse from the above:
1998 1. IsRouter flag is set in Neighbour Advertisements.
1999 2. Router Solicitations are not sent unless accept_ra is 2.
2000 3. Router Advertisements are ignored unless accept_ra is 2.
2001 4. Redirects are ignored.
2003 Default: 0 (disabled) if global forwarding is disabled (default),
2004 otherwise 1 (enabled).
2007 Default Hop Limit to set.
2012 Default Maximum Transfer Unit
2014 Default: 1280 (IPv6 required minimum)
2016 ip_nonlocal_bind - BOOLEAN
2017 If set, allows processes to bind() to non-local IPv6 addresses,
2018 which can be quite useful - but may break some applications.
2022 router_probe_interval - INTEGER
2023 Minimum interval (in seconds) between Router Probing described
2028 router_solicitation_delay - INTEGER
2029 Number of seconds to wait after interface is brought up
2030 before sending Router Solicitations.
2034 router_solicitation_interval - INTEGER
2035 Number of seconds to wait between Router Solicitations.
2039 router_solicitations - INTEGER
2040 Number of Router Solicitations to send until assuming no
2041 routers are present.
2045 use_oif_addrs_only - BOOLEAN
2046 When enabled, the candidate source addresses for destinations
2047 routed via this interface are restricted to the set of addresses
2048 configured on this interface (vis. RFC 6724, section 4).
2052 use_tempaddr - INTEGER
2053 Preference for Privacy Extensions (RFC3041).
2055 * <= 0 : disable Privacy Extensions
2056 * == 1 : enable Privacy Extensions, but prefer public
2057 addresses over temporary addresses.
2058 * > 1 : enable Privacy Extensions and prefer temporary
2059 addresses over public addresses.
2063 * 0 (for most devices)
2064 * -1 (for point-to-point devices and loopback devices)
2066 temp_valid_lft - INTEGER
2067 valid lifetime (in seconds) for temporary addresses.
2069 Default: 172800 (2 days)
2071 temp_prefered_lft - INTEGER
2072 Preferred lifetime (in seconds) for temporary addresses.
2074 Default: 86400 (1 day)
2076 keep_addr_on_down - INTEGER
2077 Keep all IPv6 addresses on an interface down event. If set static
2078 global addresses with no expiration time are not flushed.
2081 * 0 : system default
2084 Default: 0 (addresses are removed)
2086 max_desync_factor - INTEGER
2087 Maximum value for DESYNC_FACTOR, which is a random value
2088 that ensures that clients don't synchronize with each
2089 other and generate new addresses at exactly the same time.
2090 value is in seconds.
2094 regen_max_retry - INTEGER
2095 Number of attempts before give up attempting to generate
2096 valid temporary addresses.
2100 max_addresses - INTEGER
2101 Maximum number of autoconfigured addresses per interface. Setting
2102 to zero disables the limitation. It is not recommended to set this
2103 value too large (or to zero) because it would be an easy way to
2104 crash the kernel by allowing too many addresses to be created.
2108 disable_ipv6 - BOOLEAN
2109 Disable IPv6 operation. If accept_dad is set to 2, this value
2110 will be dynamically set to TRUE if DAD fails for the link-local
2113 Default: FALSE (enable IPv6 operation)
2115 When this value is changed from 1 to 0 (IPv6 is being enabled),
2116 it will dynamically create a link-local address on the given
2117 interface and start Duplicate Address Detection, if necessary.
2119 When this value is changed from 0 to 1 (IPv6 is being disabled),
2120 it will dynamically delete all addresses and routes on the given
2121 interface. From now on it will not possible to add addresses/routes
2122 to the selected interface.
2124 accept_dad - INTEGER
2125 Whether to accept DAD (Duplicate Address Detection).
2127 == ==============================================================
2129 1 Enable DAD (default)
2130 2 Enable DAD, and disable IPv6 operation if MAC-based duplicate
2131 link-local address has been found.
2132 == ==============================================================
2134 DAD operation and mode on a given interface will be selected according
2135 to the maximum value of conf/{all,interface}/accept_dad.
2137 force_tllao - BOOLEAN
2138 Enable sending the target link-layer address option even when
2139 responding to a unicast neighbor solicitation.
2143 Quoting from RFC 2461, section 4.4, Target link-layer address:
2145 "The option MUST be included for multicast solicitations in order to
2146 avoid infinite Neighbor Solicitation "recursion" when the peer node
2147 does not have a cache entry to return a Neighbor Advertisements
2148 message. When responding to unicast solicitations, the option can be
2149 omitted since the sender of the solicitation has the correct link-
2150 layer address; otherwise it would not have be able to send the unicast
2151 solicitation in the first place. However, including the link-layer
2152 address in this case adds little overhead and eliminates a potential
2153 race condition where the sender deletes the cached link-layer address
2154 prior to receiving a response to a previous solicitation."
2156 ndisc_notify - BOOLEAN
2157 Define mode for notification of address and device changes.
2159 * 0 - (default): do nothing
2160 * 1 - Generate unsolicited neighbour advertisements when device is brought
2161 up or hardware address changes.
2163 ndisc_tclass - INTEGER
2164 The IPv6 Traffic Class to use by default when sending IPv6 Neighbor
2165 Discovery (Router Solicitation, Router Advertisement, Neighbor
2166 Solicitation, Neighbor Advertisement, Redirect) messages.
2167 These 8 bits can be interpreted as 6 high order bits holding the DSCP
2168 value and 2 low order bits representing ECN (which you probably want
2173 mldv1_unsolicited_report_interval - INTEGER
2174 The interval in milliseconds in which the next unsolicited
2175 MLDv1 report retransmit will take place.
2177 Default: 10000 (10 seconds)
2179 mldv2_unsolicited_report_interval - INTEGER
2180 The interval in milliseconds in which the next unsolicited
2181 MLDv2 report retransmit will take place.
2183 Default: 1000 (1 second)
2185 force_mld_version - INTEGER
2186 * 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
2187 * 1 - Enforce to use MLD version 1
2188 * 2 - Enforce to use MLD version 2
2190 suppress_frag_ndisc - INTEGER
2191 Control RFC 6980 (Security Implications of IPv6 Fragmentation
2192 with IPv6 Neighbor Discovery) behavior:
2194 * 1 - (default) discard fragmented neighbor discovery packets
2195 * 0 - allow fragmented neighbor discovery packets
2197 optimistic_dad - BOOLEAN
2198 Whether to perform Optimistic Duplicate Address Detection (RFC 4429).
2200 * 0: disabled (default)
2203 Optimistic Duplicate Address Detection for the interface will be enabled
2204 if at least one of conf/{all,interface}/optimistic_dad is set to 1,
2205 it will be disabled otherwise.
2207 use_optimistic - BOOLEAN
2208 If enabled, do not classify optimistic addresses as deprecated during
2209 source address selection. Preferred addresses will still be chosen
2210 before optimistic addresses, subject to other ranking in the source
2211 address selection algorithm.
2213 * 0: disabled (default)
2216 This will be enabled if at least one of
2217 conf/{all,interface}/use_optimistic is set to 1, disabled otherwise.
2219 stable_secret - IPv6 address
2220 This IPv6 address will be used as a secret to generate IPv6
2221 addresses for link-local addresses and autoconfigured
2222 ones. All addresses generated after setting this secret will
2223 be stable privacy ones by default. This can be changed via the
2224 addrgenmode ip-link. conf/default/stable_secret is used as the
2225 secret for the namespace, the interface specific ones can
2226 overwrite that. Writes to conf/all/stable_secret are refused.
2228 It is recommended to generate this secret during installation
2229 of a system and keep it stable after that.
2231 By default the stable secret is unset.
2233 addr_gen_mode - INTEGER
2234 Defines how link-local and autoconf addresses are generated.
2236 = =================================================================
2237 0 generate address based on EUI64 (default)
2238 1 do no generate a link-local address, use EUI64 for addresses
2239 generated from autoconf
2240 2 generate stable privacy addresses, using the secret from
2241 stable_secret (RFC7217)
2242 3 generate stable privacy addresses, using a random secret if unset
2243 = =================================================================
2245 drop_unicast_in_l2_multicast - BOOLEAN
2246 Drop any unicast IPv6 packets that are received in link-layer
2247 multicast (or broadcast) frames.
2249 By default this is turned off.
2251 drop_unsolicited_na - BOOLEAN
2252 Drop all unsolicited neighbor advertisements, for example if there's
2253 a known good NA proxy on the network and such frames need not be used
2254 (or in the case of 802.11, must not be used to prevent attacks.)
2256 By default this is turned off.
2258 enhanced_dad - BOOLEAN
2259 Include a nonce option in the IPv6 neighbor solicitation messages used for
2260 duplicate address detection per RFC7527. A received DAD NS will only signal
2261 a duplicate address if the nonce is different. This avoids any false
2262 detection of duplicates due to loopback of the NS messages that we send.
2263 The nonce option will be sent on an interface unless both of
2264 conf/{all,interface}/enhanced_dad are set to FALSE.
2272 Limit the maximal rates for sending ICMPv6 messages.
2274 0 to disable any limiting,
2275 otherwise the minimal space between responses in milliseconds.
2279 ratemask - list of comma separated ranges
2280 For ICMPv6 message types matching the ranges in the ratemask, limit
2281 the sending of the message according to ratelimit parameter.
2283 The format used for both input and output is a comma separated
2284 list of ranges (e.g. "0-127,129" for ICMPv6 message type 0 to 127 and
2285 129). Writing to the file will clear all previous ranges of ICMPv6
2286 message types and update the current list with the input.
2288 Refer to: https://www.iana.org/assignments/icmpv6-parameters/icmpv6-parameters.xhtml
2289 for numerical values of ICMPv6 message types, e.g. echo request is 128
2290 and echo reply is 129.
2292 Default: 0-1,3-127 (rate limit ICMPv6 errors except Packet Too Big)
2294 echo_ignore_all - BOOLEAN
2295 If set non-zero, then the kernel will ignore all ICMP ECHO
2296 requests sent to it over the IPv6 protocol.
2300 echo_ignore_multicast - BOOLEAN
2301 If set non-zero, then the kernel will ignore all ICMP ECHO
2302 requests sent to it over the IPv6 protocol via multicast.
2306 echo_ignore_anycast - BOOLEAN
2307 If set non-zero, then the kernel will ignore all ICMP ECHO
2308 requests sent to it over the IPv6 protocol destined to anycast address.
2312 xfrm6_gc_thresh - INTEGER
2313 (Obsolete since linux-4.14)
2314 The threshold at which we will start garbage collecting for IPv6
2315 destination cache entries. At twice this value the system will
2316 refuse new allocations.
2320 Pekka Savola <pekkas@netcore.fi>
2321 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
2324 /proc/sys/net/bridge/* Variables:
2325 =================================
2327 bridge-nf-call-arptables - BOOLEAN
2328 - 1 : pass bridged ARP traffic to arptables' FORWARD chain.
2333 bridge-nf-call-iptables - BOOLEAN
2334 - 1 : pass bridged IPv4 traffic to iptables' chains.
2339 bridge-nf-call-ip6tables - BOOLEAN
2340 - 1 : pass bridged IPv6 traffic to ip6tables' chains.
2345 bridge-nf-filter-vlan-tagged - BOOLEAN
2346 - 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
2351 bridge-nf-filter-pppoe-tagged - BOOLEAN
2352 - 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
2357 bridge-nf-pass-vlan-input-dev - BOOLEAN
2358 - 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
2359 interface on the bridge and set the netfilter input device to the
2360 vlan. This allows use of e.g. "iptables -i br0.1" and makes the
2361 REDIRECT target work with vlan-on-top-of-bridge interfaces. When no
2362 matching vlan interface is found, or this switch is off, the input
2363 device is set to the bridge interface.
2365 - 0: disable bridge netfilter vlan interface lookup.
2369 ``proc/sys/net/sctp/*`` Variables:
2370 ==================================
2372 addip_enable - BOOLEAN
2373 Enable or disable extension of Dynamic Address Reconfiguration
2374 (ADD-IP) functionality specified in RFC5061. This extension provides
2375 the ability to dynamically add and remove new addresses for the SCTP
2378 1: Enable extension.
2380 0: Disable extension.
2385 Enable or disable pf (pf is short for potentially failed) state. A value
2386 of pf_retrans > path_max_retrans also disables pf state. That is, one of
2387 both pf_enable and pf_retrans > path_max_retrans can disable pf state.
2388 Since pf_retrans and path_max_retrans can be changed by userspace
2389 application, sometimes user expects to disable pf state by the value of
2390 pf_retrans > path_max_retrans, but occasionally the value of pf_retrans
2391 or path_max_retrans is changed by the user application, this pf state is
2392 enabled. As such, it is necessary to add this to dynamically enable
2393 and disable pf state. See:
2394 https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for
2404 Unset or enable/disable pf (pf is short for potentially failed) state
2405 exposure. Applications can control the exposure of the PF path state
2406 in the SCTP_PEER_ADDR_CHANGE event and the SCTP_GET_PEER_ADDR_INFO
2407 sockopt. When it's unset, no SCTP_PEER_ADDR_CHANGE event with
2408 SCTP_ADDR_PF state will be sent and a SCTP_PF-state transport info
2409 can be got via SCTP_GET_PEER_ADDR_INFO sockopt; When it's enabled,
2410 a SCTP_PEER_ADDR_CHANGE event will be sent for a transport becoming
2411 SCTP_PF state and a SCTP_PF-state transport info can be got via
2412 SCTP_GET_PEER_ADDR_INFO sockopt; When it's diabled, no
2413 SCTP_PEER_ADDR_CHANGE event will be sent and it returns -EACCES when
2414 trying to get a SCTP_PF-state transport info via SCTP_GET_PEER_ADDR_INFO
2417 0: Unset pf state exposure, Compatible with old applications.
2419 1: Disable pf state exposure.
2421 2: Enable pf state exposure.
2425 addip_noauth_enable - BOOLEAN
2426 Dynamic Address Reconfiguration (ADD-IP) requires the use of
2427 authentication to protect the operations of adding or removing new
2428 addresses. This requirement is mandated so that unauthorized hosts
2429 would not be able to hijack associations. However, older
2430 implementations may not have implemented this requirement while
2431 allowing the ADD-IP extension. For reasons of interoperability,
2432 we provide this variable to control the enforcement of the
2433 authentication requirement.
2435 == ===============================================================
2436 1 Allow ADD-IP extension to be used without authentication. This
2437 should only be set in a closed environment for interoperability
2438 with older implementations.
2440 0 Enforce the authentication requirement
2441 == ===============================================================
2445 auth_enable - BOOLEAN
2446 Enable or disable Authenticated Chunks extension. This extension
2447 provides the ability to send and receive authenticated chunks and is
2448 required for secure operation of Dynamic Address Reconfiguration
2451 - 1: Enable this extension.
2452 - 0: Disable this extension.
2456 prsctp_enable - BOOLEAN
2457 Enable or disable the Partial Reliability extension (RFC3758) which
2458 is used to notify peers that a given DATA should no longer be expected.
2460 - 1: Enable extension
2466 The limit of the number of new packets that can be initially sent. It
2467 controls how bursty the generated traffic can be.
2471 association_max_retrans - INTEGER
2472 Set the maximum number for retransmissions that an association can
2473 attempt deciding that the remote end is unreachable. If this value
2474 is exceeded, the association is terminated.
2478 max_init_retransmits - INTEGER
2479 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
2480 that an association will attempt before declaring the destination
2481 unreachable and terminating.
2485 path_max_retrans - INTEGER
2486 The maximum number of retransmissions that will be attempted on a given
2487 path. Once this threshold is exceeded, the path is considered
2488 unreachable, and new traffic will use a different path when the
2489 association is multihomed.
2493 pf_retrans - INTEGER
2494 The number of retransmissions that will be attempted on a given path
2495 before traffic is redirected to an alternate transport (should one
2496 exist). Note this is distinct from path_max_retrans, as a path that
2497 passes the pf_retrans threshold can still be used. Its only
2498 deprioritized when a transmission path is selected by the stack. This
2499 setting is primarily used to enable fast failover mechanisms without
2500 having to reduce path_max_retrans to a very low value. See:
2501 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
2502 for details. Note also that a value of pf_retrans > path_max_retrans
2503 disables this feature. Since both pf_retrans and path_max_retrans can
2504 be changed by userspace application, a variable pf_enable is used to
2509 ps_retrans - INTEGER
2510 Primary.Switchover.Max.Retrans (PSMR), it's a tunable parameter coming
2511 from section-5 "Primary Path Switchover" in rfc7829. The primary path
2512 will be changed to another active path when the path error counter on
2513 the old primary path exceeds PSMR, so that "the SCTP sender is allowed
2514 to continue data transmission on a new working path even when the old
2515 primary destination address becomes active again". Note this feature
2516 is disabled by initializing 'ps_retrans' per netns as 0xffff by default,
2517 and its value can't be less than 'pf_retrans' when changing by sysctl.
2521 rto_initial - INTEGER
2522 The initial round trip timeout value in milliseconds that will be used
2523 in calculating round trip times. This is the initial time interval
2524 for retransmissions.
2529 The maximum value (in milliseconds) of the round trip timeout. This
2530 is the largest time interval that can elapse between retransmissions.
2535 The minimum value (in milliseconds) of the round trip timeout. This
2536 is the smallest time interval the can elapse between retransmissions.
2540 hb_interval - INTEGER
2541 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
2542 are sent at the specified interval on idle paths to probe the state of
2543 a given path between 2 associations.
2547 sack_timeout - INTEGER
2548 The amount of time (in milliseconds) that the implementation will wait
2553 valid_cookie_life - INTEGER
2554 The default lifetime of the SCTP cookie (in milliseconds). The cookie
2555 is used during association establishment.
2559 cookie_preserve_enable - BOOLEAN
2560 Enable or disable the ability to extend the lifetime of the SCTP cookie
2561 that is used during the establishment phase of SCTP association
2563 - 1: Enable cookie lifetime extension.
2568 cookie_hmac_alg - STRING
2569 Select the hmac algorithm used when generating the cookie value sent by
2570 a listening sctp socket to a connecting client in the INIT-ACK chunk.
2577 Ability to assign md5 or sha1 as the selected alg is predicated on the
2578 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
2579 CONFIG_CRYPTO_SHA1).
2581 Default: Dependent on configuration. MD5 if available, else SHA1 if
2582 available, else none.
2584 rcvbuf_policy - INTEGER
2585 Determines if the receive buffer is attributed to the socket or to
2586 association. SCTP supports the capability to create multiple
2587 associations on a single socket. When using this capability, it is
2588 possible that a single stalled association that's buffering a lot
2589 of data may block other associations from delivering their data by
2590 consuming all of the receive buffer space. To work around this,
2591 the rcvbuf_policy could be set to attribute the receiver buffer space
2592 to each association instead of the socket. This prevents the described
2595 - 1: rcvbuf space is per association
2596 - 0: rcvbuf space is per socket
2600 sndbuf_policy - INTEGER
2601 Similar to rcvbuf_policy above, this applies to send buffer space.
2603 - 1: Send buffer is tracked per association
2604 - 0: Send buffer is tracked per socket.
2608 sctp_mem - vector of 3 INTEGERs: min, pressure, max
2609 Number of pages allowed for queueing by all SCTP sockets.
2611 min: Below this number of pages SCTP is not bothered about its
2612 memory appetite. When amount of memory allocated by SCTP exceeds
2613 this number, SCTP starts to moderate memory usage.
2615 pressure: This value was introduced to follow format of tcp_mem.
2617 max: Number of pages allowed for queueing by all SCTP sockets.
2619 Default is calculated at boot time from amount of available memory.
2621 sctp_rmem - vector of 3 INTEGERs: min, default, max
2622 Only the first value ("min") is used, "default" and "max" are
2625 min: Minimal size of receive buffer used by SCTP socket.
2626 It is guaranteed to each SCTP socket (but not association) even
2627 under moderate memory pressure.
2631 sctp_wmem - vector of 3 INTEGERs: min, default, max
2632 Currently this tunable has no effect.
2634 addr_scope_policy - INTEGER
2635 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
2637 - 0 - Disable IPv4 address scoping
2638 - 1 - Enable IPv4 address scoping
2639 - 2 - Follow draft but allow IPv4 private addresses
2640 - 3 - Follow draft but allow IPv4 link local addresses
2645 ``/proc/sys/net/core/*``
2646 ========================
2648 Please see: Documentation/admin-guide/sysctl/net.rst for descriptions of these entries.
2651 ``/proc/sys/net/unix/*``
2652 ========================
2654 max_dgram_qlen - INTEGER
2655 The maximum length of dgram socket receive queue