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 the smallest of the old MTU to
29 this destination and min_pmtu (see below). You will need
30 to raise min_pmtu to the smallest interface MTU on your system
31 manually if you want to avoid locally generated fragments.
33 In mode 2 incoming Path MTU Discovery messages will be
34 discarded. Outgoing frames are handled the same as in mode 1,
35 implicitly setting IP_PMTUDISC_DONT on every created socket.
37 Mode 3 is a hardened pmtu discover mode. The kernel will only
38 accept fragmentation-needed errors if the underlying protocol
39 can verify them besides a plain socket lookup. Current
40 protocols for which pmtu events will be honored are TCP, SCTP
41 and DCCP as they verify e.g. the sequence number or the
42 association. This mode should not be enabled globally but is
43 only intended to secure e.g. name servers in namespaces where
44 TCP path mtu must still work but path MTU information of other
45 protocols should be discarded. If enabled globally this mode
46 could break other protocols.
53 default 552 - minimum Path MTU. Unless this is changed mannually,
54 each cached pmtu will never be lower than this setting.
56 ip_forward_use_pmtu - BOOLEAN
57 By default we don't trust protocol path MTUs while forwarding
58 because they could be easily forged and can lead to unwanted
59 fragmentation by the router.
60 You only need to enable this if you have user-space software
61 which tries to discover path mtus by itself and depends on the
62 kernel honoring this information. This is normally not the
72 fwmark_reflect - BOOLEAN
73 Controls the fwmark of kernel-generated IPv4 reply packets that are not
74 associated with a socket for example, TCP RSTs or ICMP echo replies).
75 If unset, these packets have a fwmark of zero. If set, they have the
76 fwmark of the packet they are replying to.
80 fib_multipath_use_neigh - BOOLEAN
81 Use status of existing neighbor entry when determining nexthop for
82 multipath routes. If disabled, neighbor information is not used and
83 packets could be directed to a failed nexthop. Only valid for kernels
84 built with CONFIG_IP_ROUTE_MULTIPATH enabled.
93 fib_multipath_hash_policy - INTEGER
94 Controls which hash policy to use for multipath routes. Only valid
95 for kernels built with CONFIG_IP_ROUTE_MULTIPATH enabled.
103 - 2 - Layer 3 or inner Layer 3 if present
104 - 3 - Custom multipath hash. Fields used for multipath hash calculation
105 are determined by fib_multipath_hash_fields sysctl
107 fib_multipath_hash_fields - UNSIGNED INTEGER
108 When fib_multipath_hash_policy is set to 3 (custom multipath hash), the
109 fields used for multipath hash calculation are determined by this
112 This value is a bitmask which enables various fields for multipath hash
117 ====== ============================
118 0x0001 Source IP address
119 0x0002 Destination IP address
121 0x0008 Unused (Flow Label)
123 0x0020 Destination port
124 0x0040 Inner source IP address
125 0x0080 Inner destination IP address
126 0x0100 Inner IP protocol
127 0x0200 Inner Flow Label
128 0x0400 Inner source port
129 0x0800 Inner destination port
130 ====== ============================
132 Default: 0x0007 (source IP, destination IP and IP protocol)
134 fib_sync_mem - UNSIGNED INTEGER
135 Amount of dirty memory from fib entries that can be backlogged before
136 synchronize_rcu is forced.
138 Default: 512kB Minimum: 64kB Maximum: 64MB
140 ip_forward_update_priority - INTEGER
141 Whether to update SKB priority from "TOS" field in IPv4 header after it
142 is forwarded. The new SKB priority is mapped from TOS field value
143 according to an rt_tos2priority table (see e.g. man tc-prio).
145 Default: 1 (Update priority.)
149 - 0 - Do not update priority.
150 - 1 - Update priority.
152 route/max_size - INTEGER
153 Maximum number of routes allowed in the kernel. Increase
154 this when using large numbers of interfaces and/or routes.
156 From linux kernel 3.6 onwards, this is deprecated for ipv4
157 as route cache is no longer used.
159 neigh/default/gc_thresh1 - INTEGER
160 Minimum number of entries to keep. Garbage collector will not
161 purge entries if there are fewer than this number.
165 neigh/default/gc_thresh2 - INTEGER
166 Threshold when garbage collector becomes more aggressive about
167 purging entries. Entries older than 5 seconds will be cleared
168 when over this number.
172 neigh/default/gc_thresh3 - INTEGER
173 Maximum number of non-PERMANENT neighbor entries allowed. Increase
174 this when using large numbers of interfaces and when communicating
175 with large numbers of directly-connected peers.
179 neigh/default/unres_qlen_bytes - INTEGER
180 The maximum number of bytes which may be used by packets
181 queued for each unresolved address by other network layers.
184 Setting negative value is meaningless and will return error.
186 Default: SK_WMEM_MAX, (same as net.core.wmem_default).
188 Exact value depends on architecture and kernel options,
189 but should be enough to allow queuing 256 packets
192 neigh/default/unres_qlen - INTEGER
193 The maximum number of packets which may be queued for each
194 unresolved address by other network layers.
196 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
198 Prior to linux 3.3, the default value is 3 which may cause
199 unexpected packet loss. The current default value is calculated
200 according to default value of unres_qlen_bytes and true size of
205 neigh/default/interval_probe_time_ms - INTEGER
206 The probe interval for neighbor entries with NTF_MANAGED flag,
211 mtu_expires - INTEGER
212 Time, in seconds, that cached PMTU information is kept.
214 min_adv_mss - INTEGER
215 The advertised MSS depends on the first hop route MTU, but will
216 never be lower than this setting.
218 fib_notify_on_flag_change - INTEGER
219 Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/
220 RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed.
222 After installing a route to the kernel, user space receives an
223 acknowledgment, which means the route was installed in the kernel,
224 but not necessarily in hardware.
225 It is also possible for a route already installed in hardware to change
226 its action and therefore its flags. For example, a host route that is
227 trapping packets can be "promoted" to perform decapsulation following
228 the installation of an IPinIP/VXLAN tunnel.
229 The notifications will indicate to user-space the state of the route.
231 Default: 0 (Do not emit notifications.)
235 - 0 - Do not emit notifications.
236 - 1 - Emit notifications.
237 - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change.
241 ipfrag_high_thresh - LONG INTEGER
242 Maximum memory used to reassemble IP fragments.
244 ipfrag_low_thresh - LONG INTEGER
245 (Obsolete since linux-4.17)
246 Maximum memory used to reassemble IP fragments before the kernel
247 begins to remove incomplete fragment queues to free up resources.
248 The kernel still accepts new fragments for defragmentation.
250 ipfrag_time - INTEGER
251 Time in seconds to keep an IP fragment in memory.
253 ipfrag_max_dist - INTEGER
254 ipfrag_max_dist is a non-negative integer value which defines the
255 maximum "disorder" which is allowed among fragments which share a
256 common IP source address. Note that reordering of packets is
257 not unusual, but if a large number of fragments arrive from a source
258 IP address while a particular fragment queue remains incomplete, it
259 probably indicates that one or more fragments belonging to that queue
260 have been lost. When ipfrag_max_dist is positive, an additional check
261 is done on fragments before they are added to a reassembly queue - if
262 ipfrag_max_dist (or more) fragments have arrived from a particular IP
263 address between additions to any IP fragment queue using that source
264 address, it's presumed that one or more fragments in the queue are
265 lost. The existing fragment queue will be dropped, and a new one
266 started. An ipfrag_max_dist value of zero disables this check.
268 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
269 result in unnecessarily dropping fragment queues when normal
270 reordering of packets occurs, which could lead to poor application
271 performance. Using a very large value, e.g. 50000, increases the
272 likelihood of incorrectly reassembling IP fragments that originate
273 from different IP datagrams, which could result in data corruption.
276 bc_forwarding - INTEGER
277 bc_forwarding enables the feature described in rfc1812#section-5.3.5.2
278 and rfc2644. It allows the router to forward directed broadcast.
279 To enable this feature, the 'all' entry and the input interface entry
286 inet_peer_threshold - INTEGER
287 The approximate size of the storage. Starting from this threshold
288 entries will be thrown aggressively. This threshold also determines
289 entries' time-to-live and time intervals between garbage collection
290 passes. More entries, less time-to-live, less GC interval.
292 inet_peer_minttl - INTEGER
293 Minimum time-to-live of entries. Should be enough to cover fragment
294 time-to-live on the reassembling side. This minimum time-to-live is
295 guaranteed if the pool size is less than inet_peer_threshold.
298 inet_peer_maxttl - INTEGER
299 Maximum time-to-live of entries. Unused entries will expire after
300 this period of time if there is no memory pressure on the pool (i.e.
301 when the number of entries in the pool is very small).
308 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
309 Defaults to 4096. (Was 128 before linux-5.4)
310 See also tcp_max_syn_backlog for additional tuning for TCP sockets.
312 tcp_abort_on_overflow - BOOLEAN
313 If listening service is too slow to accept new connections,
314 reset them. Default state is FALSE. It means that if overflow
315 occurred due to a burst, connection will recover. Enable this
316 option _only_ if you are really sure that listening daemon
317 cannot be tuned to accept connections faster. Enabling this
318 option can harm clients of your server.
320 tcp_adv_win_scale - INTEGER
321 Count buffering overhead as bytes/2^tcp_adv_win_scale
322 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
325 Possible values are [-31, 31], inclusive.
329 tcp_allowed_congestion_control - STRING
330 Show/set the congestion control choices available to non-privileged
331 processes. The list is a subset of those listed in
332 tcp_available_congestion_control.
334 Default is "reno" and the default setting (tcp_congestion_control).
336 tcp_app_win - INTEGER
337 Reserve max(window/2^tcp_app_win, mss) of window for application
338 buffer. Value 0 is special, it means that nothing is reserved.
342 tcp_autocorking - BOOLEAN
343 Enable TCP auto corking :
344 When applications do consecutive small write()/sendmsg() system calls,
345 we try to coalesce these small writes as much as possible, to lower
346 total amount of sent packets. This is done if at least one prior
347 packet for the flow is waiting in Qdisc queues or device transmit
348 queue. Applications can still use TCP_CORK for optimal behavior
349 when they know how/when to uncork their sockets.
353 tcp_available_congestion_control - STRING
354 Shows the available congestion control choices that are registered.
355 More congestion control algorithms may be available as modules,
358 tcp_base_mss - INTEGER
359 The initial value of search_low to be used by the packetization layer
360 Path MTU discovery (MTU probing). If MTU probing is enabled,
361 this is the initial MSS used by the connection.
363 tcp_mtu_probe_floor - INTEGER
364 If MTU probing is enabled this caps the minimum MSS used for search_low
369 tcp_min_snd_mss - INTEGER
370 TCP SYN and SYNACK messages usually advertise an ADVMSS option,
371 as described in RFC 1122 and RFC 6691.
373 If this ADVMSS option is smaller than tcp_min_snd_mss,
374 it is silently capped to tcp_min_snd_mss.
376 Default : 48 (at least 8 bytes of payload per segment)
378 tcp_congestion_control - STRING
379 Set the congestion control algorithm to be used for new
380 connections. The algorithm "reno" is always available, but
381 additional choices may be available based on kernel configuration.
382 Default is set as part of kernel configuration.
383 For passive connections, the listener congestion control choice
386 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
389 Allows TCP to send "duplicate" SACKs.
391 tcp_early_retrans - INTEGER
392 Tail loss probe (TLP) converts RTOs occurring due to tail
393 losses into fast recovery (draft-ietf-tcpm-rack). Note that
394 TLP requires RACK to function properly (see tcp_recovery below)
404 Control use of Explicit Congestion Notification (ECN) by TCP.
405 ECN is used only when both ends of the TCP connection indicate
406 support for it. This feature is useful in avoiding losses due
407 to congestion by allowing supporting routers to signal
408 congestion before having to drop packets.
412 = =====================================================
413 0 Disable ECN. Neither initiate nor accept ECN.
414 1 Enable ECN when requested by incoming connections and
415 also request ECN on outgoing connection attempts.
416 2 Enable ECN when requested by incoming connections
417 but do not request ECN on outgoing connections.
418 = =====================================================
422 tcp_ecn_fallback - BOOLEAN
423 If the kernel detects that ECN connection misbehaves, enable fall
424 back to non-ECN. Currently, this knob implements the fallback
425 from RFC3168, section 6.1.1.1., but we reserve that in future,
426 additional detection mechanisms could be implemented under this
427 knob. The value is not used, if tcp_ecn or per route (or congestion
428 control) ECN settings are disabled.
430 Default: 1 (fallback enabled)
433 This is a legacy option, it has no effect anymore.
435 tcp_fin_timeout - INTEGER
436 The length of time an orphaned (no longer referenced by any
437 application) connection will remain in the FIN_WAIT_2 state
438 before it is aborted at the local end. While a perfectly
439 valid "receive only" state for an un-orphaned connection, an
440 orphaned connection in FIN_WAIT_2 state could otherwise wait
441 forever for the remote to close its end of the connection.
448 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
449 F-RTO is an enhanced recovery algorithm for TCP retransmission
450 timeouts. It is particularly beneficial in networks where the
451 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
452 modification. It does not require any support from the peer.
454 By default it's enabled with a non-zero value. 0 disables F-RTO.
456 tcp_fwmark_accept - BOOLEAN
457 If set, incoming connections to listening sockets that do not have a
458 socket mark will set the mark of the accepting socket to the fwmark of
459 the incoming SYN packet. This will cause all packets on that connection
460 (starting from the first SYNACK) to be sent with that fwmark. The
461 listening socket's mark is unchanged. Listening sockets that already
462 have a fwmark set via setsockopt(SOL_SOCKET, SO_MARK, ...) are
467 tcp_invalid_ratelimit - INTEGER
468 Limit the maximal rate for sending duplicate acknowledgments
469 in response to incoming TCP packets that are for an existing
470 connection but that are invalid due to any of these reasons:
472 (a) out-of-window sequence number,
473 (b) out-of-window acknowledgment number, or
474 (c) PAWS (Protection Against Wrapped Sequence numbers) check failure
476 This can help mitigate simple "ack loop" DoS attacks, wherein
477 a buggy or malicious middlebox or man-in-the-middle can
478 rewrite TCP header fields in manner that causes each endpoint
479 to think that the other is sending invalid TCP segments, thus
480 causing each side to send an unterminating stream of duplicate
481 acknowledgments for invalid segments.
483 Using 0 disables rate-limiting of dupacks in response to
484 invalid segments; otherwise this value specifies the minimal
485 space between sending such dupacks, in milliseconds.
487 Default: 500 (milliseconds).
489 tcp_keepalive_time - INTEGER
490 How often TCP sends out keepalive messages when keepalive is enabled.
493 tcp_keepalive_probes - INTEGER
494 How many keepalive probes TCP sends out, until it decides that the
495 connection is broken. Default value: 9.
497 tcp_keepalive_intvl - INTEGER
498 How frequently the probes are send out. Multiplied by
499 tcp_keepalive_probes it is time to kill not responding connection,
500 after probes started. Default value: 75sec i.e. connection
501 will be aborted after ~11 minutes of retries.
503 tcp_l3mdev_accept - BOOLEAN
504 Enables child sockets to inherit the L3 master device index.
505 Enabling this option allows a "global" listen socket to work
506 across L3 master domains (e.g., VRFs) with connected sockets
507 derived from the listen socket to be bound to the L3 domain in
508 which the packets originated. Only valid when the kernel was
509 compiled with CONFIG_NET_L3_MASTER_DEV.
511 Default: 0 (disabled)
513 tcp_low_latency - BOOLEAN
514 This is a legacy option, it has no effect anymore.
516 tcp_max_orphans - INTEGER
517 Maximal number of TCP sockets not attached to any user file handle,
518 held by system. If this number is exceeded orphaned connections are
519 reset immediately and warning is printed. This limit exists
520 only to prevent simple DoS attacks, you _must_ not rely on this
521 or lower the limit artificially, but rather increase it
522 (probably, after increasing installed memory),
523 if network conditions require more than default value,
524 and tune network services to linger and kill such states
525 more aggressively. Let me to remind again: each orphan eats
526 up to ~64K of unswappable memory.
528 tcp_max_syn_backlog - INTEGER
529 Maximal number of remembered connection requests (SYN_RECV),
530 which have not received an acknowledgment from connecting client.
532 This is a per-listener limit.
534 The minimal value is 128 for low memory machines, and it will
535 increase in proportion to the memory of machine.
537 If server suffers from overload, try increasing this number.
539 Remember to also check /proc/sys/net/core/somaxconn
540 A SYN_RECV request socket consumes about 304 bytes of memory.
542 tcp_max_tw_buckets - INTEGER
543 Maximal number of timewait sockets held by system simultaneously.
544 If this number is exceeded time-wait socket is immediately destroyed
545 and warning is printed. This limit exists only to prevent
546 simple DoS attacks, you _must_ not lower the limit artificially,
547 but rather increase it (probably, after increasing installed memory),
548 if network conditions require more than default value.
550 tcp_mem - vector of 3 INTEGERs: min, pressure, max
551 min: below this number of pages TCP is not bothered about its
554 pressure: when amount of memory allocated by TCP exceeds this number
555 of pages, TCP moderates its memory consumption and enters memory
556 pressure mode, which is exited when memory consumption falls
559 max: number of pages allowed for queueing by all TCP sockets.
561 Defaults are calculated at boot time from amount of available
564 tcp_min_rtt_wlen - INTEGER
565 The window length of the windowed min filter to track the minimum RTT.
566 A shorter window lets a flow more quickly pick up new (higher)
567 minimum RTT when it is moved to a longer path (e.g., due to traffic
568 engineering). A longer window makes the filter more resistant to RTT
569 inflations such as transient congestion. The unit is seconds.
571 Possible values: 0 - 86400 (1 day)
575 tcp_moderate_rcvbuf - BOOLEAN
576 If set, TCP performs receive buffer auto-tuning, attempting to
577 automatically size the buffer (no greater than tcp_rmem[2]) to
578 match the size required by the path for full throughput. Enabled by
581 tcp_mtu_probing - INTEGER
582 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
586 - 1 - Disabled by default, enabled when an ICMP black hole detected
587 - 2 - Always enabled, use initial MSS of tcp_base_mss.
589 tcp_probe_interval - UNSIGNED INTEGER
590 Controls how often to start TCP Packetization-Layer Path MTU
591 Discovery reprobe. The default is reprobing every 10 minutes as
594 tcp_probe_threshold - INTEGER
595 Controls when TCP Packetization-Layer Path MTU Discovery probing
596 will stop in respect to the width of search range in bytes. Default
599 tcp_no_metrics_save - BOOLEAN
600 By default, TCP saves various connection metrics in the route cache
601 when the connection closes, so that connections established in the
602 near future can use these to set initial conditions. Usually, this
603 increases overall performance, but may sometimes cause performance
604 degradation. If set, TCP will not cache metrics on closing
607 tcp_no_ssthresh_metrics_save - BOOLEAN
608 Controls whether TCP saves ssthresh metrics in the route cache.
610 Default is 1, which disables ssthresh metrics.
612 tcp_orphan_retries - INTEGER
613 This value influences the timeout of a locally closed TCP connection,
614 when RTO retransmissions remain unacknowledged.
615 See tcp_retries2 for more details.
617 The default value is 8.
619 If your machine is a loaded WEB server,
620 you should think about lowering this value, such sockets
621 may consume significant resources. Cf. tcp_max_orphans.
623 tcp_recovery - INTEGER
624 This value is a bitmap to enable various experimental loss recovery
627 ========= =============================================================
628 RACK: 0x1 enables the RACK loss detection for fast detection of lost
629 retransmissions and tail drops. It also subsumes and disables
630 RFC6675 recovery for SACK connections.
632 RACK: 0x2 makes RACK's reordering window static (min_rtt/4).
634 RACK: 0x4 disables RACK's DUPACK threshold heuristic
635 ========= =============================================================
639 tcp_reordering - INTEGER
640 Initial reordering level of packets in a TCP stream.
641 TCP stack can then dynamically adjust flow reordering level
642 between this initial value and tcp_max_reordering
646 tcp_max_reordering - INTEGER
647 Maximal reordering level of packets in a TCP stream.
648 300 is a fairly conservative value, but you might increase it
649 if paths are using per packet load balancing (like bonding rr mode)
653 tcp_retrans_collapse - BOOLEAN
654 Bug-to-bug compatibility with some broken printers.
655 On retransmit try to send bigger packets to work around bugs in
658 tcp_retries1 - INTEGER
659 This value influences the time, after which TCP decides, that
660 something is wrong due to unacknowledged RTO retransmissions,
661 and reports this suspicion to the network layer.
662 See tcp_retries2 for more details.
664 RFC 1122 recommends at least 3 retransmissions, which is the
667 tcp_retries2 - INTEGER
668 This value influences the timeout of an alive TCP connection,
669 when RTO retransmissions remain unacknowledged.
670 Given a value of N, a hypothetical TCP connection following
671 exponential backoff with an initial RTO of TCP_RTO_MIN would
672 retransmit N times before killing the connection at the (N+1)th RTO.
674 The default value of 15 yields a hypothetical timeout of 924.6
675 seconds and is a lower bound for the effective timeout.
676 TCP will effectively time out at the first RTO which exceeds the
677 hypothetical timeout.
679 RFC 1122 recommends at least 100 seconds for the timeout,
680 which corresponds to a value of at least 8.
682 tcp_rfc1337 - BOOLEAN
683 If set, the TCP stack behaves conforming to RFC1337. If unset,
684 we are not conforming to RFC, but prevent TCP TIME_WAIT
689 tcp_rmem - vector of 3 INTEGERs: min, default, max
690 min: Minimal size of receive buffer used by TCP sockets.
691 It is guaranteed to each TCP socket, even under moderate memory
696 default: initial size of receive buffer used by TCP sockets.
697 This value overrides net.core.rmem_default used by other protocols.
698 Default: 131072 bytes.
699 This value results in initial window of 65535.
701 max: maximal size of receive buffer allowed for automatically
702 selected receiver buffers for TCP socket. This value does not override
703 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
704 automatic tuning of that socket's receive buffer size, in which
705 case this value is ignored.
706 Default: between 131072 and 6MB, depending on RAM size.
709 Enable select acknowledgments (SACKS).
711 tcp_comp_sack_delay_ns - LONG INTEGER
712 TCP tries to reduce number of SACK sent, using a timer
713 based on 5% of SRTT, capped by this sysctl, in nano seconds.
714 The default is 1ms, based on TSO autosizing period.
716 Default : 1,000,000 ns (1 ms)
718 tcp_comp_sack_slack_ns - LONG INTEGER
719 This sysctl control the slack used when arming the
720 timer used by SACK compression. This gives extra time
721 for small RTT flows, and reduces system overhead by allowing
722 opportunistic reduction of timer interrupts.
724 Default : 100,000 ns (100 us)
726 tcp_comp_sack_nr - INTEGER
727 Max number of SACK that can be compressed.
728 Using 0 disables SACK compression.
732 tcp_slow_start_after_idle - BOOLEAN
733 If set, provide RFC2861 behavior and time out the congestion
734 window after an idle period. An idle period is defined at
735 the current RTO. If unset, the congestion window will not
736 be timed out after an idle period.
741 Use the Host requirements interpretation of the TCP urgent pointer field.
742 Most hosts use the older BSD interpretation, so if you turn this on
743 Linux might not communicate correctly with them.
747 tcp_synack_retries - INTEGER
748 Number of times SYNACKs for a passive TCP connection attempt will
749 be retransmitted. Should not be higher than 255. Default value
750 is 5, which corresponds to 31seconds till the last retransmission
751 with the current initial RTO of 1second. With this the final timeout
752 for a passive TCP connection will happen after 63seconds.
754 tcp_syncookies - INTEGER
755 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
756 Send out syncookies when the syn backlog queue of a socket
757 overflows. This is to prevent against the common 'SYN flood attack'
760 Note, that syncookies is fallback facility.
761 It MUST NOT be used to help highly loaded servers to stand
762 against legal connection rate. If you see SYN flood warnings
763 in your logs, but investigation shows that they occur
764 because of overload with legal connections, you should tune
765 another parameters until this warning disappear.
766 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
768 syncookies seriously violate TCP protocol, do not allow
769 to use TCP extensions, can result in serious degradation
770 of some services (f.e. SMTP relaying), visible not by you,
771 but your clients and relays, contacting you. While you see
772 SYN flood warnings in logs not being really flooded, your server
773 is seriously misconfigured.
775 If you want to test which effects syncookies have to your
776 network connections you can set this knob to 2 to enable
777 unconditionally generation of syncookies.
779 tcp_migrate_req - BOOLEAN
780 The incoming connection is tied to a specific listening socket when
781 the initial SYN packet is received during the three-way handshake.
782 When a listener is closed, in-flight request sockets during the
783 handshake and established sockets in the accept queue are aborted.
785 If the listener has SO_REUSEPORT enabled, other listeners on the
786 same port should have been able to accept such connections. This
787 option makes it possible to migrate such child sockets to another
788 listener after close() or shutdown().
790 The BPF_SK_REUSEPORT_SELECT_OR_MIGRATE type of eBPF program should
791 usually be used to define the policy to pick an alive listener.
792 Otherwise, the kernel will randomly pick an alive listener only if
793 this option is enabled.
795 Note that migration between listeners with different settings may
796 crash applications. Let's say migration happens from listener A to
797 B, and only B has TCP_SAVE_SYN enabled. B cannot read SYN data from
798 the requests migrated from A. To avoid such a situation, cancel
799 migration by returning SK_DROP in the type of eBPF program, or
804 tcp_fastopen - INTEGER
805 Enable TCP Fast Open (RFC7413) to send and accept data in the opening
808 The client support is enabled by flag 0x1 (on by default). The client
809 then must use sendmsg() or sendto() with the MSG_FASTOPEN flag,
810 rather than connect() to send data in SYN.
812 The server support is enabled by flag 0x2 (off by default). Then
813 either enable for all listeners with another flag (0x400) or
814 enable individual listeners via TCP_FASTOPEN socket option with
815 the option value being the length of the syn-data backlog.
817 The values (bitmap) are
819 ===== ======== ======================================================
820 0x1 (client) enables sending data in the opening SYN on the client.
821 0x2 (server) enables the server support, i.e., allowing data in
822 a SYN packet to be accepted and passed to the
823 application before 3-way handshake finishes.
824 0x4 (client) send data in the opening SYN regardless of cookie
825 availability and without a cookie option.
826 0x200 (server) accept data-in-SYN w/o any cookie option present.
827 0x400 (server) enable all listeners to support Fast Open by
828 default without explicit TCP_FASTOPEN socket option.
829 ===== ======== ======================================================
833 Note that additional client or server features are only
834 effective if the basic support (0x1 and 0x2) are enabled respectively.
836 tcp_fastopen_blackhole_timeout_sec - INTEGER
837 Initial time period in second to disable Fastopen on active TCP sockets
838 when a TFO firewall blackhole issue happens.
839 This time period will grow exponentially when more blackhole issues
840 get detected right after Fastopen is re-enabled and will reset to
841 initial value when the blackhole issue goes away.
842 0 to disable the blackhole detection.
844 By default, it is set to 0 (feature is disabled).
846 tcp_fastopen_key - list of comma separated 32-digit hexadecimal INTEGERs
847 The list consists of a primary key and an optional backup key. The
848 primary key is used for both creating and validating cookies, while the
849 optional backup key is only used for validating cookies. The purpose of
850 the backup key is to maximize TFO validation when keys are rotated.
852 A randomly chosen primary key may be configured by the kernel if
853 the tcp_fastopen sysctl is set to 0x400 (see above), or if the
854 TCP_FASTOPEN setsockopt() optname is set and a key has not been
855 previously configured via sysctl. If keys are configured via
856 setsockopt() by using the TCP_FASTOPEN_KEY optname, then those
857 per-socket keys will be used instead of any keys that are specified via
860 A key is specified as 4 8-digit hexadecimal integers which are separated
861 by a '-' as: xxxxxxxx-xxxxxxxx-xxxxxxxx-xxxxxxxx. Leading zeros may be
862 omitted. A primary and a backup key may be specified by separating them
863 by a comma. If only one key is specified, it becomes the primary key and
864 any previously configured backup keys are removed.
866 tcp_syn_retries - INTEGER
867 Number of times initial SYNs for an active TCP connection attempt
868 will be retransmitted. Should not be higher than 127. Default value
869 is 6, which corresponds to 63seconds till the last retransmission
870 with the current initial RTO of 1second. With this the final timeout
871 for an active TCP connection attempt will happen after 127seconds.
873 tcp_timestamps - INTEGER
874 Enable timestamps as defined in RFC1323.
877 - 1: Enable timestamps as defined in RFC1323 and use random offset for
878 each connection rather than only using the current time.
879 - 2: Like 1, but without random offsets.
883 tcp_min_tso_segs - INTEGER
884 Minimal number of segments per TSO frame.
886 Since linux-3.12, TCP does an automatic sizing of TSO frames,
887 depending on flow rate, instead of filling 64Kbytes packets.
888 For specific usages, it's possible to force TCP to build big
889 TSO frames. Note that TCP stack might split too big TSO packets
890 if available window is too small.
894 tcp_tso_rtt_log - INTEGER
895 Adjustment of TSO packet sizes based on min_rtt
897 Starting from linux-5.18, TCP autosizing can be tweaked
898 for flows having small RTT.
900 Old autosizing was splitting the pacing budget to send 1024 TSO
903 tso_packet_size = sk->sk_pacing_rate / 1024;
905 With the new mechanism, we increase this TSO sizing using:
907 distance = min_rtt_usec / (2^tcp_tso_rtt_log)
908 tso_packet_size += gso_max_size >> distance;
910 This means that flows between very close hosts can use bigger
911 TSO packets, reducing their cpu costs.
913 If you want to use the old autosizing, set this sysctl to 0.
915 Default: 9 (2^9 = 512 usec)
917 tcp_pacing_ss_ratio - INTEGER
918 sk->sk_pacing_rate is set by TCP stack using a ratio applied
919 to current rate. (current_rate = cwnd * mss / srtt)
920 If TCP is in slow start, tcp_pacing_ss_ratio is applied
921 to let TCP probe for bigger speeds, assuming cwnd can be
922 doubled every other RTT.
926 tcp_pacing_ca_ratio - INTEGER
927 sk->sk_pacing_rate is set by TCP stack using a ratio applied
928 to current rate. (current_rate = cwnd * mss / srtt)
929 If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio
930 is applied to conservatively probe for bigger throughput.
934 tcp_tso_win_divisor - INTEGER
935 This allows control over what percentage of the congestion window
936 can be consumed by a single TSO frame.
937 The setting of this parameter is a choice between burstiness and
938 building larger TSO frames.
942 tcp_tw_reuse - INTEGER
943 Enable reuse of TIME-WAIT sockets for new connections when it is
944 safe from protocol viewpoint.
948 - 2 - enable for loopback traffic only
950 It should not be changed without advice/request of technical
955 tcp_window_scaling - BOOLEAN
956 Enable window scaling as defined in RFC1323.
958 tcp_wmem - vector of 3 INTEGERs: min, default, max
959 min: Amount of memory reserved for send buffers for TCP sockets.
960 Each TCP socket has rights to use it due to fact of its birth.
964 default: initial size of send buffer used by TCP sockets. This
965 value overrides net.core.wmem_default used by other protocols.
967 It is usually lower than net.core.wmem_default.
971 max: Maximal amount of memory allowed for automatically tuned
972 send buffers for TCP sockets. This value does not override
973 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
974 automatic tuning of that socket's send buffer size, in which case
975 this value is ignored.
977 Default: between 64K and 4MB, depending on RAM size.
979 tcp_notsent_lowat - UNSIGNED INTEGER
980 A TCP socket can control the amount of unsent bytes in its write queue,
981 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
982 reports POLLOUT events if the amount of unsent bytes is below a per
983 socket value, and if the write queue is not full. sendmsg() will
984 also not add new buffers if the limit is hit.
986 This global variable controls the amount of unsent data for
987 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
988 to the global variable has immediate effect.
990 Default: UINT_MAX (0xFFFFFFFF)
992 tcp_workaround_signed_windows - BOOLEAN
993 If set, assume no receipt of a window scaling option means the
994 remote TCP is broken and treats the window as a signed quantity.
995 If unset, assume the remote TCP is not broken even if we do
996 not receive a window scaling option from them.
1000 tcp_thin_linear_timeouts - BOOLEAN
1001 Enable dynamic triggering of linear timeouts for thin streams.
1002 If set, a check is performed upon retransmission by timeout to
1003 determine if the stream is thin (less than 4 packets in flight).
1004 As long as the stream is found to be thin, up to 6 linear
1005 timeouts may be performed before exponential backoff mode is
1006 initiated. This improves retransmission latency for
1007 non-aggressive thin streams, often found to be time-dependent.
1008 For more information on thin streams, see
1009 Documentation/networking/tcp-thin.rst
1013 tcp_limit_output_bytes - INTEGER
1014 Controls TCP Small Queue limit per tcp socket.
1015 TCP bulk sender tends to increase packets in flight until it
1016 gets losses notifications. With SNDBUF autotuning, this can
1017 result in a large amount of packets queued on the local machine
1018 (e.g.: qdiscs, CPU backlog, or device) hurting latency of other
1019 flows, for typical pfifo_fast qdiscs. tcp_limit_output_bytes
1020 limits the number of bytes on qdisc or device to reduce artificial
1021 RTT/cwnd and reduce bufferbloat.
1023 Default: 1048576 (16 * 65536)
1025 tcp_challenge_ack_limit - INTEGER
1026 Limits number of Challenge ACK sent per second, as recommended
1027 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
1033 udp_l3mdev_accept - BOOLEAN
1034 Enabling this option allows a "global" bound socket to work
1035 across L3 master domains (e.g., VRFs) with packets capable of
1036 being received regardless of the L3 domain in which they
1037 originated. Only valid when the kernel was compiled with
1038 CONFIG_NET_L3_MASTER_DEV.
1040 Default: 0 (disabled)
1042 udp_mem - vector of 3 INTEGERs: min, pressure, max
1043 Number of pages allowed for queueing by all UDP sockets.
1045 min: Number of pages allowed for queueing by all UDP sockets.
1047 pressure: This value was introduced to follow format of tcp_mem.
1049 max: This value was introduced to follow format of tcp_mem.
1051 Default is calculated at boot time from amount of available memory.
1053 udp_rmem_min - INTEGER
1054 Minimal size of receive buffer used by UDP sockets in moderation.
1055 Each UDP socket is able to use the size for receiving data, even if
1056 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
1060 udp_wmem_min - INTEGER
1061 Minimal size of send buffer used by UDP sockets in moderation.
1062 Each UDP socket is able to use the size for sending data, even if
1063 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
1070 raw_l3mdev_accept - BOOLEAN
1071 Enabling this option allows a "global" bound socket to work
1072 across L3 master domains (e.g., VRFs) with packets capable of
1073 being received regardless of the L3 domain in which they
1074 originated. Only valid when the kernel was compiled with
1075 CONFIG_NET_L3_MASTER_DEV.
1077 Default: 1 (enabled)
1082 cipso_cache_enable - BOOLEAN
1083 If set, enable additions to and lookups from the CIPSO label mapping
1084 cache. If unset, additions are ignored and lookups always result in a
1085 miss. However, regardless of the setting the cache is still
1086 invalidated when required when means you can safely toggle this on and
1087 off and the cache will always be "safe".
1091 cipso_cache_bucket_size - INTEGER
1092 The CIPSO label cache consists of a fixed size hash table with each
1093 hash bucket containing a number of cache entries. This variable limits
1094 the number of entries in each hash bucket; the larger the value is, the
1095 more CIPSO label mappings that can be cached. When the number of
1096 entries in a given hash bucket reaches this limit adding new entries
1097 causes the oldest entry in the bucket to be removed to make room.
1101 cipso_rbm_optfmt - BOOLEAN
1102 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
1103 the CIPSO draft specification (see Documentation/netlabel for details).
1104 This means that when set the CIPSO tag will be padded with empty
1105 categories in order to make the packet data 32-bit aligned.
1109 cipso_rbm_structvalid - BOOLEAN
1110 If set, do a very strict check of the CIPSO option when
1111 ip_options_compile() is called. If unset, relax the checks done during
1112 ip_options_compile(). Either way is "safe" as errors are caught else
1113 where in the CIPSO processing code but setting this to 0 (False) should
1114 result in less work (i.e. it should be faster) but could cause problems
1115 with other implementations that require strict checking.
1122 ip_local_port_range - 2 INTEGERS
1123 Defines the local port range that is used by TCP and UDP to
1124 choose the local port. The first number is the first, the
1125 second the last local port number.
1126 If possible, it is better these numbers have different parity
1127 (one even and one odd value).
1128 Must be greater than or equal to ip_unprivileged_port_start.
1129 The default values are 32768 and 60999 respectively.
1131 ip_local_reserved_ports - list of comma separated ranges
1132 Specify the ports which are reserved for known third-party
1133 applications. These ports will not be used by automatic port
1134 assignments (e.g. when calling connect() or bind() with port
1135 number 0). Explicit port allocation behavior is unchanged.
1137 The format used for both input and output is a comma separated
1138 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
1139 10). Writing to the file will clear all previously reserved
1140 ports and update the current list with the one given in the
1143 Note that ip_local_port_range and ip_local_reserved_ports
1144 settings are independent and both are considered by the kernel
1145 when determining which ports are available for automatic port
1148 You can reserve ports which are not in the current
1149 ip_local_port_range, e.g.::
1151 $ cat /proc/sys/net/ipv4/ip_local_port_range
1153 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
1156 although this is redundant. However such a setting is useful
1157 if later the port range is changed to a value that will
1158 include the reserved ports. Also keep in mind, that overlapping
1159 of these ranges may affect probability of selecting ephemeral
1160 ports which are right after block of reserved ports.
1164 ip_unprivileged_port_start - INTEGER
1165 This is a per-namespace sysctl. It defines the first
1166 unprivileged port in the network namespace. Privileged ports
1167 require root or CAP_NET_BIND_SERVICE in order to bind to them.
1168 To disable all privileged ports, set this to 0. They must not
1169 overlap with the ip_local_port_range.
1173 ip_nonlocal_bind - BOOLEAN
1174 If set, allows processes to bind() to non-local IP addresses,
1175 which can be quite useful - but may break some applications.
1179 ip_autobind_reuse - BOOLEAN
1180 By default, bind() does not select the ports automatically even if
1181 the new socket and all sockets bound to the port have SO_REUSEADDR.
1182 ip_autobind_reuse allows bind() to reuse the port and this is useful
1183 when you use bind()+connect(), but may break some applications.
1184 The preferred solution is to use IP_BIND_ADDRESS_NO_PORT and this
1185 option should only be set by experts.
1188 ip_dynaddr - INTEGER
1189 If set non-zero, enables support for dynamic addresses.
1190 If set to a non-zero value larger than 1, a kernel log
1191 message will be printed when dynamic address rewriting
1196 ip_early_demux - BOOLEAN
1197 Optimize input packet processing down to one demux for
1198 certain kinds of local sockets. Currently we only do this
1199 for established TCP and connected UDP sockets.
1201 It may add an additional cost for pure routing workloads that
1202 reduces overall throughput, in such case you should disable it.
1206 ping_group_range - 2 INTEGERS
1207 Restrict ICMP_PROTO datagram sockets to users in the group range.
1208 The default is "1 0", meaning, that nobody (not even root) may
1209 create ping sockets. Setting it to "100 100" would grant permissions
1210 to the single group. "0 4294967295" would enable it for the world, "100
1211 4294967295" would enable it for the users, but not daemons.
1213 tcp_early_demux - BOOLEAN
1214 Enable early demux for established TCP sockets.
1218 udp_early_demux - BOOLEAN
1219 Enable early demux for connected UDP sockets. Disable this if
1220 your system could experience more unconnected load.
1224 icmp_echo_ignore_all - BOOLEAN
1225 If set non-zero, then the kernel will ignore all ICMP ECHO
1226 requests sent to it.
1230 icmp_echo_enable_probe - BOOLEAN
1231 If set to one, then the kernel will respond to RFC 8335 PROBE
1232 requests sent to it.
1236 icmp_echo_ignore_broadcasts - BOOLEAN
1237 If set non-zero, then the kernel will ignore all ICMP ECHO and
1238 TIMESTAMP requests sent to it via broadcast/multicast.
1242 icmp_ratelimit - INTEGER
1243 Limit the maximal rates for sending ICMP packets whose type matches
1244 icmp_ratemask (see below) to specific targets.
1245 0 to disable any limiting,
1246 otherwise the minimal space between responses in milliseconds.
1247 Note that another sysctl, icmp_msgs_per_sec limits the number
1248 of ICMP packets sent on all targets.
1252 icmp_msgs_per_sec - INTEGER
1253 Limit maximal number of ICMP packets sent per second from this host.
1254 Only messages whose type matches icmp_ratemask (see below) are
1255 controlled by this limit. For security reasons, the precise count
1256 of messages per second is randomized.
1260 icmp_msgs_burst - INTEGER
1261 icmp_msgs_per_sec controls number of ICMP packets sent per second,
1262 while icmp_msgs_burst controls the burst size of these packets.
1263 For security reasons, the precise burst size is randomized.
1267 icmp_ratemask - INTEGER
1268 Mask made of ICMP types for which rates are being limited.
1270 Significant bits: IHGFEDCBA9876543210
1272 Default mask: 0000001100000011000 (6168)
1274 Bit definitions (see include/linux/icmp.h):
1276 = =========================
1278 3 Destination Unreachable [1]_
1279 4 Source Quench [1]_
1282 B Time Exceeded [1]_
1283 C Parameter Problem [1]_
1288 H Address Mask Request
1289 I Address Mask Reply
1290 = =========================
1292 .. [1] These are rate limited by default (see default mask above)
1294 icmp_ignore_bogus_error_responses - BOOLEAN
1295 Some routers violate RFC1122 by sending bogus responses to broadcast
1296 frames. Such violations are normally logged via a kernel warning.
1297 If this is set to TRUE, the kernel will not give such warnings, which
1298 will avoid log file clutter.
1302 icmp_errors_use_inbound_ifaddr - BOOLEAN
1304 If zero, icmp error messages are sent with the primary address of
1305 the exiting interface.
1307 If non-zero, the message will be sent with the primary address of
1308 the interface that received the packet that caused the icmp error.
1309 This is the behaviour many network administrators will expect from
1310 a router. And it can make debugging complicated network layouts
1313 Note that if no primary address exists for the interface selected,
1314 then the primary address of the first non-loopback interface that
1315 has one will be used regardless of this setting.
1319 igmp_max_memberships - INTEGER
1320 Change the maximum number of multicast groups we can subscribe to.
1323 Theoretical maximum value is bounded by having to send a membership
1324 report in a single datagram (i.e. the report can't span multiple
1325 datagrams, or risk confusing the switch and leaving groups you don't
1328 The number of supported groups 'M' is bounded by the number of group
1329 report entries you can fit into a single datagram of 65535 bytes.
1331 M = 65536-sizeof (ip header)/(sizeof(Group record))
1333 Group records are variable length, with a minimum of 12 bytes.
1334 So net.ipv4.igmp_max_memberships should not be set higher than:
1336 (65536-24) / 12 = 5459
1338 The value 5459 assumes no IP header options, so in practice
1339 this number may be lower.
1341 igmp_max_msf - INTEGER
1342 Maximum number of addresses allowed in the source filter list for a
1348 Controls the IGMP query robustness variable (see RFC2236 8.1).
1350 Default: 2 (as specified by RFC2236 8.1)
1352 Minimum: 1 (as specified by RFC6636 4.5)
1354 force_igmp_version - INTEGER
1355 - 0 - (default) No enforcement of a IGMP version, IGMPv1/v2 fallback
1356 allowed. Will back to IGMPv3 mode again if all IGMPv1/v2 Querier
1357 Present timer expires.
1358 - 1 - Enforce to use IGMP version 1. Will also reply IGMPv1 report if
1359 receive IGMPv2/v3 query.
1360 - 2 - Enforce to use IGMP version 2. Will fallback to IGMPv1 if receive
1361 IGMPv1 query message. Will reply report if receive IGMPv3 query.
1362 - 3 - Enforce to use IGMP version 3. The same react with default 0.
1366 this is not the same with force_mld_version because IGMPv3 RFC3376
1367 Security Considerations does not have clear description that we could
1368 ignore other version messages completely as MLDv2 RFC3810. So make
1369 this value as default 0 is recommended.
1371 ``conf/interface/*``
1372 changes special settings per interface (where
1373 interface" is the name of your network interface)
1376 is special, changes the settings for all interfaces
1378 log_martians - BOOLEAN
1379 Log packets with impossible addresses to kernel log.
1380 log_martians for the interface will be enabled if at least one of
1381 conf/{all,interface}/log_martians is set to TRUE,
1382 it will be disabled otherwise
1384 accept_redirects - BOOLEAN
1385 Accept ICMP redirect messages.
1386 accept_redirects for the interface will be enabled if:
1388 - both conf/{all,interface}/accept_redirects are TRUE in the case
1389 forwarding for the interface is enabled
1393 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
1394 case forwarding for the interface is disabled
1396 accept_redirects for the interface will be disabled otherwise
1403 forwarding - BOOLEAN
1404 Enable IP forwarding on this interface. This controls whether packets
1405 received _on_ this interface can be forwarded.
1407 mc_forwarding - BOOLEAN
1408 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
1409 and a multicast routing daemon is required.
1410 conf/all/mc_forwarding must also be set to TRUE to enable multicast
1411 routing for the interface
1414 Integer value used to differentiate the devices by the medium they
1415 are attached to. Two devices can have different id values when
1416 the broadcast packets are received only on one of them.
1417 The default value 0 means that the device is the only interface
1418 to its medium, value of -1 means that medium is not known.
1420 Currently, it is used to change the proxy_arp behavior:
1421 the proxy_arp feature is enabled for packets forwarded between
1422 two devices attached to different media.
1427 proxy_arp for the interface will be enabled if at least one of
1428 conf/{all,interface}/proxy_arp is set to TRUE,
1429 it will be disabled otherwise
1431 proxy_arp_pvlan - BOOLEAN
1432 Private VLAN proxy arp.
1434 Basically allow proxy arp replies back to the same interface
1435 (from which the ARP request/solicitation was received).
1437 This is done to support (ethernet) switch features, like RFC
1438 3069, where the individual ports are NOT allowed to
1439 communicate with each other, but they are allowed to talk to
1440 the upstream router. As described in RFC 3069, it is possible
1441 to allow these hosts to communicate through the upstream
1442 router by proxy_arp'ing. Don't need to be used together with
1445 This technology is known by different names:
1447 In RFC 3069 it is called VLAN Aggregation.
1448 Cisco and Allied Telesyn call it Private VLAN.
1449 Hewlett-Packard call it Source-Port filtering or port-isolation.
1450 Ericsson call it MAC-Forced Forwarding (RFC Draft).
1452 shared_media - BOOLEAN
1453 Send(router) or accept(host) RFC1620 shared media redirects.
1454 Overrides secure_redirects.
1456 shared_media for the interface will be enabled if at least one of
1457 conf/{all,interface}/shared_media is set to TRUE,
1458 it will be disabled otherwise
1462 secure_redirects - BOOLEAN
1463 Accept ICMP redirect messages only to gateways listed in the
1464 interface's current gateway list. Even if disabled, RFC1122 redirect
1467 Overridden by shared_media.
1469 secure_redirects for the interface will be enabled if at least one of
1470 conf/{all,interface}/secure_redirects is set to TRUE,
1471 it will be disabled otherwise
1475 send_redirects - BOOLEAN
1476 Send redirects, if router.
1478 send_redirects for the interface will be enabled if at least one of
1479 conf/{all,interface}/send_redirects is set to TRUE,
1480 it will be disabled otherwise
1484 bootp_relay - BOOLEAN
1485 Accept packets with source address 0.b.c.d destined
1486 not to this host as local ones. It is supposed, that
1487 BOOTP relay daemon will catch and forward such packets.
1488 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
1493 Not Implemented Yet.
1495 accept_source_route - BOOLEAN
1496 Accept packets with SRR option.
1497 conf/all/accept_source_route must also be set to TRUE to accept packets
1498 with SRR option on the interface
1505 accept_local - BOOLEAN
1506 Accept packets with local source addresses. In combination with
1507 suitable routing, this can be used to direct packets between two
1508 local interfaces over the wire and have them accepted properly.
1511 route_localnet - BOOLEAN
1512 Do not consider loopback addresses as martian source or destination
1513 while routing. This enables the use of 127/8 for local routing purposes.
1518 - 0 - No source validation.
1519 - 1 - Strict mode as defined in RFC3704 Strict Reverse Path
1520 Each incoming packet is tested against the FIB and if the interface
1521 is not the best reverse path the packet check will fail.
1522 By default failed packets are discarded.
1523 - 2 - Loose mode as defined in RFC3704 Loose Reverse Path
1524 Each incoming packet's source address is also tested against the FIB
1525 and if the source address is not reachable via any interface
1526 the packet check will fail.
1528 Current recommended practice in RFC3704 is to enable strict mode
1529 to prevent IP spoofing from DDos attacks. If using asymmetric routing
1530 or other complicated routing, then loose mode is recommended.
1532 The max value from conf/{all,interface}/rp_filter is used
1533 when doing source validation on the {interface}.
1535 Default value is 0. Note that some distributions enable it
1538 src_valid_mark - BOOLEAN
1539 - 0 - The fwmark of the packet is not included in reverse path
1540 route lookup. This allows for asymmetric routing configurations
1541 utilizing the fwmark in only one direction, e.g., transparent
1544 - 1 - The fwmark of the packet is included in reverse path route
1545 lookup. This permits rp_filter to function when the fwmark is
1546 used for routing traffic in both directions.
1548 This setting also affects the utilization of fmwark when
1549 performing source address selection for ICMP replies, or
1550 determining addresses stored for the IPOPT_TS_TSANDADDR and
1551 IPOPT_RR IP options.
1553 The max value from conf/{all,interface}/src_valid_mark is used.
1557 arp_filter - BOOLEAN
1558 - 1 - Allows you to have multiple network interfaces on the same
1559 subnet, and have the ARPs for each interface be answered
1560 based on whether or not the kernel would route a packet from
1561 the ARP'd IP out that interface (therefore you must use source
1562 based routing for this to work). In other words it allows control
1563 of which cards (usually 1) will respond to an arp request.
1565 - 0 - (default) The kernel can respond to arp requests with addresses
1566 from other interfaces. This may seem wrong but it usually makes
1567 sense, because it increases the chance of successful communication.
1568 IP addresses are owned by the complete host on Linux, not by
1569 particular interfaces. Only for more complex setups like load-
1570 balancing, does this behaviour cause problems.
1572 arp_filter for the interface will be enabled if at least one of
1573 conf/{all,interface}/arp_filter is set to TRUE,
1574 it will be disabled otherwise
1576 arp_announce - INTEGER
1577 Define different restriction levels for announcing the local
1578 source IP address from IP packets in ARP requests sent on
1581 - 0 - (default) Use any local address, configured on any interface
1582 - 1 - Try to avoid local addresses that are not in the target's
1583 subnet for this interface. This mode is useful when target
1584 hosts reachable via this interface require the source IP
1585 address in ARP requests to be part of their logical network
1586 configured on the receiving interface. When we generate the
1587 request we will check all our subnets that include the
1588 target IP and will preserve the source address if it is from
1589 such subnet. If there is no such subnet we select source
1590 address according to the rules for level 2.
1591 - 2 - Always use the best local address for this target.
1592 In this mode we ignore the source address in the IP packet
1593 and try to select local address that we prefer for talks with
1594 the target host. Such local address is selected by looking
1595 for primary IP addresses on all our subnets on the outgoing
1596 interface that include the target IP address. If no suitable
1597 local address is found we select the first local address
1598 we have on the outgoing interface or on all other interfaces,
1599 with the hope we will receive reply for our request and
1600 even sometimes no matter the source IP address we announce.
1602 The max value from conf/{all,interface}/arp_announce is used.
1604 Increasing the restriction level gives more chance for
1605 receiving answer from the resolved target while decreasing
1606 the level announces more valid sender's information.
1608 arp_ignore - INTEGER
1609 Define different modes for sending replies in response to
1610 received ARP requests that resolve local target IP addresses:
1612 - 0 - (default): reply for any local target IP address, configured
1614 - 1 - reply only if the target IP address is local address
1615 configured on the incoming interface
1616 - 2 - reply only if the target IP address is local address
1617 configured on the incoming interface and both with the
1618 sender's IP address are part from same subnet on this interface
1619 - 3 - do not reply for local addresses configured with scope host,
1620 only resolutions for global and link addresses are replied
1622 - 8 - do not reply for all local addresses
1624 The max value from conf/{all,interface}/arp_ignore is used
1625 when ARP request is received on the {interface}
1627 arp_notify - BOOLEAN
1628 Define mode for notification of address and device changes.
1630 == ==========================================================
1631 0 (default): do nothing
1632 1 Generate gratuitous arp requests when device is brought up
1633 or hardware address changes.
1634 == ==========================================================
1636 arp_accept - BOOLEAN
1637 Define behavior for gratuitous ARP frames who's IP is not
1638 already present in the ARP table:
1640 - 0 - don't create new entries in the ARP table
1641 - 1 - create new entries in the ARP table
1643 Both replies and requests type gratuitous arp will trigger the
1644 ARP table to be updated, if this setting is on.
1646 If the ARP table already contains the IP address of the
1647 gratuitous arp frame, the arp table will be updated regardless
1648 if this setting is on or off.
1650 arp_evict_nocarrier - BOOLEAN
1651 Clears the ARP cache on NOCARRIER events. This option is important for
1652 wireless devices where the ARP cache should not be cleared when roaming
1653 between access points on the same network. In most cases this should
1654 remain as the default (1).
1656 - 1 - (default): Clear the ARP cache on NOCARRIER events
1657 - 0 - Do not clear ARP cache on NOCARRIER events
1659 mcast_solicit - INTEGER
1660 The maximum number of multicast probes in INCOMPLETE state,
1661 when the associated hardware address is unknown. Defaults
1664 ucast_solicit - INTEGER
1665 The maximum number of unicast probes in PROBE state, when
1666 the hardware address is being reconfirmed. Defaults to 3.
1668 app_solicit - INTEGER
1669 The maximum number of probes to send to the user space ARP daemon
1670 via netlink before dropping back to multicast probes (see
1671 mcast_resolicit). Defaults to 0.
1673 mcast_resolicit - INTEGER
1674 The maximum number of multicast probes after unicast and
1675 app probes in PROBE state. Defaults to 0.
1677 disable_policy - BOOLEAN
1678 Disable IPSEC policy (SPD) for this interface
1680 disable_xfrm - BOOLEAN
1681 Disable IPSEC encryption on this interface, whatever the policy
1683 igmpv2_unsolicited_report_interval - INTEGER
1684 The interval in milliseconds in which the next unsolicited
1685 IGMPv1 or IGMPv2 report retransmit will take place.
1687 Default: 10000 (10 seconds)
1689 igmpv3_unsolicited_report_interval - INTEGER
1690 The interval in milliseconds in which the next unsolicited
1691 IGMPv3 report retransmit will take place.
1693 Default: 1000 (1 seconds)
1695 ignore_routes_with_linkdown - BOOLEAN
1696 Ignore routes whose link is down when performing a FIB lookup.
1698 promote_secondaries - BOOLEAN
1699 When a primary IP address is removed from this interface
1700 promote a corresponding secondary IP address instead of
1701 removing all the corresponding secondary IP addresses.
1703 drop_unicast_in_l2_multicast - BOOLEAN
1704 Drop any unicast IP packets that are received in link-layer
1705 multicast (or broadcast) frames.
1707 This behavior (for multicast) is actually a SHOULD in RFC
1708 1122, but is disabled by default for compatibility reasons.
1712 drop_gratuitous_arp - BOOLEAN
1713 Drop all gratuitous ARP frames, for example if there's a known
1714 good ARP proxy on the network and such frames need not be used
1715 (or in the case of 802.11, must not be used to prevent attacks.)
1721 Allows you to write a number, which can be used as required.
1725 xfrm4_gc_thresh - INTEGER
1726 (Obsolete since linux-4.14)
1727 The threshold at which we will start garbage collecting for IPv4
1728 destination cache entries. At twice this value the system will
1729 refuse new allocations.
1731 igmp_link_local_mcast_reports - BOOLEAN
1732 Enable IGMP reports for link local multicast groups in the
1738 kuznet@ms2.inr.ac.ru
1745 delon.nicolas@wanadoo.fr
1750 /proc/sys/net/ipv6/* Variables
1751 ==============================
1753 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1754 apply to IPv6 [XXX?].
1756 bindv6only - BOOLEAN
1757 Default value for IPV6_V6ONLY socket option,
1758 which restricts use of the IPv6 socket to IPv6 communication
1761 - TRUE: disable IPv4-mapped address feature
1762 - FALSE: enable IPv4-mapped address feature
1764 Default: FALSE (as specified in RFC3493)
1766 flowlabel_consistency - BOOLEAN
1767 Protect the consistency (and unicity) of flow label.
1768 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1776 auto_flowlabels - INTEGER
1777 Automatically generate flow labels based on a flow hash of the
1778 packet. This allows intermediate devices, such as routers, to
1779 identify packet flows for mechanisms like Equal Cost Multipath
1780 Routing (see RFC 6438).
1782 = ===========================================================
1783 0 automatic flow labels are completely disabled
1784 1 automatic flow labels are enabled by default, they can be
1785 disabled on a per socket basis using the IPV6_AUTOFLOWLABEL
1787 2 automatic flow labels are allowed, they may be enabled on a
1788 per socket basis using the IPV6_AUTOFLOWLABEL socket option
1789 3 automatic flow labels are enabled and enforced, they cannot
1790 be disabled by the socket option
1791 = ===========================================================
1795 flowlabel_state_ranges - BOOLEAN
1796 Split the flow label number space into two ranges. 0-0x7FFFF is
1797 reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF
1798 is reserved for stateless flow labels as described in RFC6437.
1805 flowlabel_reflect - INTEGER
1806 Control flow label reflection. Needed for Path MTU
1807 Discovery to work with Equal Cost Multipath Routing in anycast
1808 environments. See RFC 7690 and:
1809 https://tools.ietf.org/html/draft-wang-6man-flow-label-reflection-01
1813 - 1: enabled for established flows
1815 Note that this prevents automatic flowlabel changes, as done
1816 in "tcp: change IPv6 flow-label upon receiving spurious retransmission"
1817 and "tcp: Change txhash on every SYN and RTO retransmit"
1819 - 2: enabled for TCP RESET packets (no active listener)
1820 If set, a RST packet sent in response to a SYN packet on a closed
1821 port will reflect the incoming flow label.
1823 - 4: enabled for ICMPv6 echo reply messages.
1827 fib_multipath_hash_policy - INTEGER
1828 Controls which hash policy to use for multipath routes.
1830 Default: 0 (Layer 3)
1834 - 0 - Layer 3 (source and destination addresses plus flow label)
1835 - 1 - Layer 4 (standard 5-tuple)
1836 - 2 - Layer 3 or inner Layer 3 if present
1837 - 3 - Custom multipath hash. Fields used for multipath hash calculation
1838 are determined by fib_multipath_hash_fields sysctl
1840 fib_multipath_hash_fields - UNSIGNED INTEGER
1841 When fib_multipath_hash_policy is set to 3 (custom multipath hash), the
1842 fields used for multipath hash calculation are determined by this
1845 This value is a bitmask which enables various fields for multipath hash
1848 Possible fields are:
1850 ====== ============================
1851 0x0001 Source IP address
1852 0x0002 Destination IP address
1856 0x0020 Destination port
1857 0x0040 Inner source IP address
1858 0x0080 Inner destination IP address
1859 0x0100 Inner IP protocol
1860 0x0200 Inner Flow Label
1861 0x0400 Inner source port
1862 0x0800 Inner destination port
1863 ====== ============================
1865 Default: 0x0007 (source IP, destination IP and IP protocol)
1867 anycast_src_echo_reply - BOOLEAN
1868 Controls the use of anycast addresses as source addresses for ICMPv6
1876 idgen_delay - INTEGER
1877 Controls the delay in seconds after which time to retry
1878 privacy stable address generation if a DAD conflict is
1881 Default: 1 (as specified in RFC7217)
1883 idgen_retries - INTEGER
1884 Controls the number of retries to generate a stable privacy
1885 address if a DAD conflict is detected.
1887 Default: 3 (as specified in RFC7217)
1890 Controls the MLD query robustness variable (see RFC3810 9.1).
1892 Default: 2 (as specified by RFC3810 9.1)
1894 Minimum: 1 (as specified by RFC6636 4.5)
1896 max_dst_opts_number - INTEGER
1897 Maximum number of non-padding TLVs allowed in a Destination
1898 options extension header. If this value is less than zero
1899 then unknown options are disallowed and the number of known
1900 TLVs allowed is the absolute value of this number.
1904 max_hbh_opts_number - INTEGER
1905 Maximum number of non-padding TLVs allowed in a Hop-by-Hop
1906 options extension header. If this value is less than zero
1907 then unknown options are disallowed and the number of known
1908 TLVs allowed is the absolute value of this number.
1912 max_dst_opts_length - INTEGER
1913 Maximum length allowed for a Destination options extension
1916 Default: INT_MAX (unlimited)
1918 max_hbh_length - INTEGER
1919 Maximum length allowed for a Hop-by-Hop options extension
1922 Default: INT_MAX (unlimited)
1924 skip_notify_on_dev_down - BOOLEAN
1925 Controls whether an RTM_DELROUTE message is generated for routes
1926 removed when a device is taken down or deleted. IPv4 does not
1927 generate this message; IPv6 does by default. Setting this sysctl
1928 to true skips the message, making IPv4 and IPv6 on par in relying
1929 on userspace caches to track link events and evict routes.
1931 Default: false (generate message)
1933 nexthop_compat_mode - BOOLEAN
1934 New nexthop API provides a means for managing nexthops independent of
1935 prefixes. Backwards compatibilty with old route format is enabled by
1936 default which means route dumps and notifications contain the new
1937 nexthop attribute but also the full, expanded nexthop definition.
1938 Further, updates or deletes of a nexthop configuration generate route
1939 notifications for each fib entry using the nexthop. Once a system
1940 understands the new API, this sysctl can be disabled to achieve full
1941 performance benefits of the new API by disabling the nexthop expansion
1942 and extraneous notifications.
1943 Default: true (backward compat mode)
1945 fib_notify_on_flag_change - INTEGER
1946 Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/
1947 RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed.
1949 After installing a route to the kernel, user space receives an
1950 acknowledgment, which means the route was installed in the kernel,
1951 but not necessarily in hardware.
1952 It is also possible for a route already installed in hardware to change
1953 its action and therefore its flags. For example, a host route that is
1954 trapping packets can be "promoted" to perform decapsulation following
1955 the installation of an IPinIP/VXLAN tunnel.
1956 The notifications will indicate to user-space the state of the route.
1958 Default: 0 (Do not emit notifications.)
1962 - 0 - Do not emit notifications.
1963 - 1 - Emit notifications.
1964 - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change.
1967 Define the IOAM id of this node. Uses only 24 bits out of 32 in total.
1974 ioam6_id_wide - LONG INTEGER
1975 Define the wide IOAM id of this node. Uses only 56 bits out of 64 in
1976 total. Can be different from ioam6_id.
1979 Max: 0xFFFFFFFFFFFFFF
1981 Default: 0xFFFFFFFFFFFFFF
1985 ip6frag_high_thresh - INTEGER
1986 Maximum memory used to reassemble IPv6 fragments. When
1987 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1988 the fragment handler will toss packets until ip6frag_low_thresh
1991 ip6frag_low_thresh - INTEGER
1992 See ip6frag_high_thresh
1994 ip6frag_time - INTEGER
1995 Time in seconds to keep an IPv6 fragment in memory.
1998 Change the interface-specific default settings.
2000 These settings would be used during creating new interfaces.
2004 Change all the interface-specific settings.
2006 [XXX: Other special features than forwarding?]
2008 conf/all/disable_ipv6 - BOOLEAN
2009 Changing this value is same as changing ``conf/default/disable_ipv6``
2010 setting and also all per-interface ``disable_ipv6`` settings to the same
2013 Reading this value does not have any particular meaning. It does not say
2014 whether IPv6 support is enabled or disabled. Returned value can be 1
2015 also in the case when some interface has ``disable_ipv6`` set to 0 and
2016 has configured IPv6 addresses.
2018 conf/all/forwarding - BOOLEAN
2019 Enable global IPv6 forwarding between all interfaces.
2021 IPv4 and IPv6 work differently here; e.g. netfilter must be used
2022 to control which interfaces may forward packets and which not.
2024 This also sets all interfaces' Host/Router setting
2025 'forwarding' to the specified value. See below for details.
2027 This referred to as global forwarding.
2032 fwmark_reflect - BOOLEAN
2033 Controls the fwmark of kernel-generated IPv6 reply packets that are not
2034 associated with a socket for example, TCP RSTs or ICMPv6 echo replies).
2035 If unset, these packets have a fwmark of zero. If set, they have the
2036 fwmark of the packet they are replying to.
2040 ``conf/interface/*``:
2041 Change special settings per interface.
2043 The functional behaviour for certain settings is different
2044 depending on whether local forwarding is enabled or not.
2047 Accept Router Advertisements; autoconfigure using them.
2049 It also determines whether or not to transmit Router
2050 Solicitations. If and only if the functional setting is to
2051 accept Router Advertisements, Router Solicitations will be
2054 Possible values are:
2056 == ===========================================================
2057 0 Do not accept Router Advertisements.
2058 1 Accept Router Advertisements if forwarding is disabled.
2059 2 Overrule forwarding behaviour. Accept Router Advertisements
2060 even if forwarding is enabled.
2061 == ===========================================================
2065 - enabled if local forwarding is disabled.
2066 - disabled if local forwarding is enabled.
2068 accept_ra_defrtr - BOOLEAN
2069 Learn default router in Router Advertisement.
2073 - enabled if accept_ra is enabled.
2074 - disabled if accept_ra is disabled.
2076 ra_defrtr_metric - UNSIGNED INTEGER
2077 Route metric for default route learned in Router Advertisement. This value
2078 will be assigned as metric for the default route learned via IPv6 Router
2079 Advertisement. Takes affect only if accept_ra_defrtr is enabled.
2084 Default: IP6_RT_PRIO_USER i.e. 1024.
2086 accept_ra_from_local - BOOLEAN
2087 Accept RA with source-address that is found on local machine
2088 if the RA is otherwise proper and able to be accepted.
2090 Default is to NOT accept these as it may be an un-intended
2095 - enabled if accept_ra_from_local is enabled
2096 on a specific interface.
2097 - disabled if accept_ra_from_local is disabled
2098 on a specific interface.
2100 accept_ra_min_hop_limit - INTEGER
2101 Minimum hop limit Information in Router Advertisement.
2103 Hop limit Information in Router Advertisement less than this
2104 variable shall be ignored.
2108 accept_ra_pinfo - BOOLEAN
2109 Learn Prefix Information in Router Advertisement.
2113 - enabled if accept_ra is enabled.
2114 - disabled if accept_ra is disabled.
2116 accept_ra_rt_info_min_plen - INTEGER
2117 Minimum prefix length of Route Information in RA.
2119 Route Information w/ prefix smaller than this variable shall
2124 * 0 if accept_ra_rtr_pref is enabled.
2125 * -1 if accept_ra_rtr_pref is disabled.
2127 accept_ra_rt_info_max_plen - INTEGER
2128 Maximum prefix length of Route Information in RA.
2130 Route Information w/ prefix larger than this variable shall
2135 * 0 if accept_ra_rtr_pref is enabled.
2136 * -1 if accept_ra_rtr_pref is disabled.
2138 accept_ra_rtr_pref - BOOLEAN
2139 Accept Router Preference in RA.
2143 - enabled if accept_ra is enabled.
2144 - disabled if accept_ra is disabled.
2146 accept_ra_mtu - BOOLEAN
2147 Apply the MTU value specified in RA option 5 (RFC4861). If
2148 disabled, the MTU specified in the RA will be ignored.
2152 - enabled if accept_ra is enabled.
2153 - disabled if accept_ra is disabled.
2155 accept_redirects - BOOLEAN
2160 - enabled if local forwarding is disabled.
2161 - disabled if local forwarding is enabled.
2163 accept_source_route - INTEGER
2164 Accept source routing (routing extension header).
2166 - >= 0: Accept only routing header type 2.
2167 - < 0: Do not accept routing header.
2172 Autoconfigure addresses using Prefix Information in Router
2177 - enabled if accept_ra_pinfo is enabled.
2178 - disabled if accept_ra_pinfo is disabled.
2180 dad_transmits - INTEGER
2181 The amount of Duplicate Address Detection probes to send.
2185 forwarding - INTEGER
2186 Configure interface-specific Host/Router behaviour.
2190 It is recommended to have the same setting on all
2191 interfaces; mixed router/host scenarios are rather uncommon.
2193 Possible values are:
2195 - 0 Forwarding disabled
2196 - 1 Forwarding enabled
2200 By default, Host behaviour is assumed. This means:
2202 1. IsRouter flag is not set in Neighbour Advertisements.
2203 2. If accept_ra is TRUE (default), transmit Router
2205 3. If accept_ra is TRUE (default), accept Router
2206 Advertisements (and do autoconfiguration).
2207 4. If accept_redirects is TRUE (default), accept Redirects.
2211 If local forwarding is enabled, Router behaviour is assumed.
2212 This means exactly the reverse from the above:
2214 1. IsRouter flag is set in Neighbour Advertisements.
2215 2. Router Solicitations are not sent unless accept_ra is 2.
2216 3. Router Advertisements are ignored unless accept_ra is 2.
2217 4. Redirects are ignored.
2219 Default: 0 (disabled) if global forwarding is disabled (default),
2220 otherwise 1 (enabled).
2223 Default Hop Limit to set.
2228 Default Maximum Transfer Unit
2230 Default: 1280 (IPv6 required minimum)
2232 ip_nonlocal_bind - BOOLEAN
2233 If set, allows processes to bind() to non-local IPv6 addresses,
2234 which can be quite useful - but may break some applications.
2238 router_probe_interval - INTEGER
2239 Minimum interval (in seconds) between Router Probing described
2244 router_solicitation_delay - INTEGER
2245 Number of seconds to wait after interface is brought up
2246 before sending Router Solicitations.
2250 router_solicitation_interval - INTEGER
2251 Number of seconds to wait between Router Solicitations.
2255 router_solicitations - INTEGER
2256 Number of Router Solicitations to send until assuming no
2257 routers are present.
2261 use_oif_addrs_only - BOOLEAN
2262 When enabled, the candidate source addresses for destinations
2263 routed via this interface are restricted to the set of addresses
2264 configured on this interface (vis. RFC 6724, section 4).
2268 use_tempaddr - INTEGER
2269 Preference for Privacy Extensions (RFC3041).
2271 * <= 0 : disable Privacy Extensions
2272 * == 1 : enable Privacy Extensions, but prefer public
2273 addresses over temporary addresses.
2274 * > 1 : enable Privacy Extensions and prefer temporary
2275 addresses over public addresses.
2279 * 0 (for most devices)
2280 * -1 (for point-to-point devices and loopback devices)
2282 temp_valid_lft - INTEGER
2283 valid lifetime (in seconds) for temporary addresses.
2285 Default: 172800 (2 days)
2287 temp_prefered_lft - INTEGER
2288 Preferred lifetime (in seconds) for temporary addresses.
2290 Default: 86400 (1 day)
2292 keep_addr_on_down - INTEGER
2293 Keep all IPv6 addresses on an interface down event. If set static
2294 global addresses with no expiration time are not flushed.
2297 * 0 : system default
2300 Default: 0 (addresses are removed)
2302 max_desync_factor - INTEGER
2303 Maximum value for DESYNC_FACTOR, which is a random value
2304 that ensures that clients don't synchronize with each
2305 other and generate new addresses at exactly the same time.
2306 value is in seconds.
2310 regen_max_retry - INTEGER
2311 Number of attempts before give up attempting to generate
2312 valid temporary addresses.
2316 max_addresses - INTEGER
2317 Maximum number of autoconfigured addresses per interface. Setting
2318 to zero disables the limitation. It is not recommended to set this
2319 value too large (or to zero) because it would be an easy way to
2320 crash the kernel by allowing too many addresses to be created.
2324 disable_ipv6 - BOOLEAN
2325 Disable IPv6 operation. If accept_dad is set to 2, this value
2326 will be dynamically set to TRUE if DAD fails for the link-local
2329 Default: FALSE (enable IPv6 operation)
2331 When this value is changed from 1 to 0 (IPv6 is being enabled),
2332 it will dynamically create a link-local address on the given
2333 interface and start Duplicate Address Detection, if necessary.
2335 When this value is changed from 0 to 1 (IPv6 is being disabled),
2336 it will dynamically delete all addresses and routes on the given
2337 interface. From now on it will not possible to add addresses/routes
2338 to the selected interface.
2340 accept_dad - INTEGER
2341 Whether to accept DAD (Duplicate Address Detection).
2343 == ==============================================================
2345 1 Enable DAD (default)
2346 2 Enable DAD, and disable IPv6 operation if MAC-based duplicate
2347 link-local address has been found.
2348 == ==============================================================
2350 DAD operation and mode on a given interface will be selected according
2351 to the maximum value of conf/{all,interface}/accept_dad.
2353 force_tllao - BOOLEAN
2354 Enable sending the target link-layer address option even when
2355 responding to a unicast neighbor solicitation.
2359 Quoting from RFC 2461, section 4.4, Target link-layer address:
2361 "The option MUST be included for multicast solicitations in order to
2362 avoid infinite Neighbor Solicitation "recursion" when the peer node
2363 does not have a cache entry to return a Neighbor Advertisements
2364 message. When responding to unicast solicitations, the option can be
2365 omitted since the sender of the solicitation has the correct link-
2366 layer address; otherwise it would not have be able to send the unicast
2367 solicitation in the first place. However, including the link-layer
2368 address in this case adds little overhead and eliminates a potential
2369 race condition where the sender deletes the cached link-layer address
2370 prior to receiving a response to a previous solicitation."
2372 ndisc_notify - BOOLEAN
2373 Define mode for notification of address and device changes.
2375 * 0 - (default): do nothing
2376 * 1 - Generate unsolicited neighbour advertisements when device is brought
2377 up or hardware address changes.
2379 ndisc_tclass - INTEGER
2380 The IPv6 Traffic Class to use by default when sending IPv6 Neighbor
2381 Discovery (Router Solicitation, Router Advertisement, Neighbor
2382 Solicitation, Neighbor Advertisement, Redirect) messages.
2383 These 8 bits can be interpreted as 6 high order bits holding the DSCP
2384 value and 2 low order bits representing ECN (which you probably want
2389 ndisc_evict_nocarrier - BOOLEAN
2390 Clears the neighbor discovery table on NOCARRIER events. This option is
2391 important for wireless devices where the neighbor discovery cache should
2392 not be cleared when roaming between access points on the same network.
2393 In most cases this should remain as the default (1).
2395 - 1 - (default): Clear neighbor discover cache on NOCARRIER events.
2396 - 0 - Do not clear neighbor discovery cache on NOCARRIER events.
2398 mldv1_unsolicited_report_interval - INTEGER
2399 The interval in milliseconds in which the next unsolicited
2400 MLDv1 report retransmit will take place.
2402 Default: 10000 (10 seconds)
2404 mldv2_unsolicited_report_interval - INTEGER
2405 The interval in milliseconds in which the next unsolicited
2406 MLDv2 report retransmit will take place.
2408 Default: 1000 (1 second)
2410 force_mld_version - INTEGER
2411 * 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
2412 * 1 - Enforce to use MLD version 1
2413 * 2 - Enforce to use MLD version 2
2415 suppress_frag_ndisc - INTEGER
2416 Control RFC 6980 (Security Implications of IPv6 Fragmentation
2417 with IPv6 Neighbor Discovery) behavior:
2419 * 1 - (default) discard fragmented neighbor discovery packets
2420 * 0 - allow fragmented neighbor discovery packets
2422 optimistic_dad - BOOLEAN
2423 Whether to perform Optimistic Duplicate Address Detection (RFC 4429).
2425 * 0: disabled (default)
2428 Optimistic Duplicate Address Detection for the interface will be enabled
2429 if at least one of conf/{all,interface}/optimistic_dad is set to 1,
2430 it will be disabled otherwise.
2432 use_optimistic - BOOLEAN
2433 If enabled, do not classify optimistic addresses as deprecated during
2434 source address selection. Preferred addresses will still be chosen
2435 before optimistic addresses, subject to other ranking in the source
2436 address selection algorithm.
2438 * 0: disabled (default)
2441 This will be enabled if at least one of
2442 conf/{all,interface}/use_optimistic is set to 1, disabled otherwise.
2444 stable_secret - IPv6 address
2445 This IPv6 address will be used as a secret to generate IPv6
2446 addresses for link-local addresses and autoconfigured
2447 ones. All addresses generated after setting this secret will
2448 be stable privacy ones by default. This can be changed via the
2449 addrgenmode ip-link. conf/default/stable_secret is used as the
2450 secret for the namespace, the interface specific ones can
2451 overwrite that. Writes to conf/all/stable_secret are refused.
2453 It is recommended to generate this secret during installation
2454 of a system and keep it stable after that.
2456 By default the stable secret is unset.
2458 addr_gen_mode - INTEGER
2459 Defines how link-local and autoconf addresses are generated.
2461 = =================================================================
2462 0 generate address based on EUI64 (default)
2463 1 do no generate a link-local address, use EUI64 for addresses
2464 generated from autoconf
2465 2 generate stable privacy addresses, using the secret from
2466 stable_secret (RFC7217)
2467 3 generate stable privacy addresses, using a random secret if unset
2468 = =================================================================
2470 drop_unicast_in_l2_multicast - BOOLEAN
2471 Drop any unicast IPv6 packets that are received in link-layer
2472 multicast (or broadcast) frames.
2474 By default this is turned off.
2476 drop_unsolicited_na - BOOLEAN
2477 Drop all unsolicited neighbor advertisements, for example if there's
2478 a known good NA proxy on the network and such frames need not be used
2479 (or in the case of 802.11, must not be used to prevent attacks.)
2481 By default this is turned off.
2483 accept_untracked_na - BOOLEAN
2484 Add a new neighbour cache entry in STALE state for routers on receiving a
2485 neighbour advertisement (either solicited or unsolicited) with target
2486 link-layer address option specified if no neighbour entry is already
2487 present for the advertised IPv6 address. Without this knob, NAs received
2488 for untracked addresses (absent in neighbour cache) are silently ignored.
2490 This is as per router-side behaviour documented in RFC9131.
2492 This has lower precedence than drop_unsolicited_na.
2494 This will optimize the return path for the initial off-link communication
2495 that is initiated by a directly connected host, by ensuring that
2496 the first-hop router which turns on this setting doesn't have to
2497 buffer the initial return packets to do neighbour-solicitation.
2498 The prerequisite is that the host is configured to send
2499 unsolicited neighbour advertisements on interface bringup.
2500 This setting should be used in conjunction with the ndisc_notify setting
2501 on the host to satisfy this prerequisite.
2503 By default this is turned off.
2505 enhanced_dad - BOOLEAN
2506 Include a nonce option in the IPv6 neighbor solicitation messages used for
2507 duplicate address detection per RFC7527. A received DAD NS will only signal
2508 a duplicate address if the nonce is different. This avoids any false
2509 detection of duplicates due to loopback of the NS messages that we send.
2510 The nonce option will be sent on an interface unless both of
2511 conf/{all,interface}/enhanced_dad are set to FALSE.
2519 Limit the maximal rates for sending ICMPv6 messages.
2521 0 to disable any limiting,
2522 otherwise the minimal space between responses in milliseconds.
2526 ratemask - list of comma separated ranges
2527 For ICMPv6 message types matching the ranges in the ratemask, limit
2528 the sending of the message according to ratelimit parameter.
2530 The format used for both input and output is a comma separated
2531 list of ranges (e.g. "0-127,129" for ICMPv6 message type 0 to 127 and
2532 129). Writing to the file will clear all previous ranges of ICMPv6
2533 message types and update the current list with the input.
2535 Refer to: https://www.iana.org/assignments/icmpv6-parameters/icmpv6-parameters.xhtml
2536 for numerical values of ICMPv6 message types, e.g. echo request is 128
2537 and echo reply is 129.
2539 Default: 0-1,3-127 (rate limit ICMPv6 errors except Packet Too Big)
2541 echo_ignore_all - BOOLEAN
2542 If set non-zero, then the kernel will ignore all ICMP ECHO
2543 requests sent to it over the IPv6 protocol.
2547 echo_ignore_multicast - BOOLEAN
2548 If set non-zero, then the kernel will ignore all ICMP ECHO
2549 requests sent to it over the IPv6 protocol via multicast.
2553 echo_ignore_anycast - BOOLEAN
2554 If set non-zero, then the kernel will ignore all ICMP ECHO
2555 requests sent to it over the IPv6 protocol destined to anycast address.
2559 xfrm6_gc_thresh - INTEGER
2560 (Obsolete since linux-4.14)
2561 The threshold at which we will start garbage collecting for IPv6
2562 destination cache entries. At twice this value the system will
2563 refuse new allocations.
2567 Pekka Savola <pekkas@netcore.fi>
2568 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
2571 /proc/sys/net/bridge/* Variables:
2572 =================================
2574 bridge-nf-call-arptables - BOOLEAN
2575 - 1 : pass bridged ARP traffic to arptables' FORWARD chain.
2580 bridge-nf-call-iptables - BOOLEAN
2581 - 1 : pass bridged IPv4 traffic to iptables' chains.
2586 bridge-nf-call-ip6tables - BOOLEAN
2587 - 1 : pass bridged IPv6 traffic to ip6tables' chains.
2592 bridge-nf-filter-vlan-tagged - BOOLEAN
2593 - 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
2598 bridge-nf-filter-pppoe-tagged - BOOLEAN
2599 - 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
2604 bridge-nf-pass-vlan-input-dev - BOOLEAN
2605 - 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
2606 interface on the bridge and set the netfilter input device to the
2607 vlan. This allows use of e.g. "iptables -i br0.1" and makes the
2608 REDIRECT target work with vlan-on-top-of-bridge interfaces. When no
2609 matching vlan interface is found, or this switch is off, the input
2610 device is set to the bridge interface.
2612 - 0: disable bridge netfilter vlan interface lookup.
2616 ``proc/sys/net/sctp/*`` Variables:
2617 ==================================
2619 addip_enable - BOOLEAN
2620 Enable or disable extension of Dynamic Address Reconfiguration
2621 (ADD-IP) functionality specified in RFC5061. This extension provides
2622 the ability to dynamically add and remove new addresses for the SCTP
2625 1: Enable extension.
2627 0: Disable extension.
2632 Enable or disable pf (pf is short for potentially failed) state. A value
2633 of pf_retrans > path_max_retrans also disables pf state. That is, one of
2634 both pf_enable and pf_retrans > path_max_retrans can disable pf state.
2635 Since pf_retrans and path_max_retrans can be changed by userspace
2636 application, sometimes user expects to disable pf state by the value of
2637 pf_retrans > path_max_retrans, but occasionally the value of pf_retrans
2638 or path_max_retrans is changed by the user application, this pf state is
2639 enabled. As such, it is necessary to add this to dynamically enable
2640 and disable pf state. See:
2641 https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for
2651 Unset or enable/disable pf (pf is short for potentially failed) state
2652 exposure. Applications can control the exposure of the PF path state
2653 in the SCTP_PEER_ADDR_CHANGE event and the SCTP_GET_PEER_ADDR_INFO
2654 sockopt. When it's unset, no SCTP_PEER_ADDR_CHANGE event with
2655 SCTP_ADDR_PF state will be sent and a SCTP_PF-state transport info
2656 can be got via SCTP_GET_PEER_ADDR_INFO sockopt; When it's enabled,
2657 a SCTP_PEER_ADDR_CHANGE event will be sent for a transport becoming
2658 SCTP_PF state and a SCTP_PF-state transport info can be got via
2659 SCTP_GET_PEER_ADDR_INFO sockopt; When it's diabled, no
2660 SCTP_PEER_ADDR_CHANGE event will be sent and it returns -EACCES when
2661 trying to get a SCTP_PF-state transport info via SCTP_GET_PEER_ADDR_INFO
2664 0: Unset pf state exposure, Compatible with old applications.
2666 1: Disable pf state exposure.
2668 2: Enable pf state exposure.
2672 addip_noauth_enable - BOOLEAN
2673 Dynamic Address Reconfiguration (ADD-IP) requires the use of
2674 authentication to protect the operations of adding or removing new
2675 addresses. This requirement is mandated so that unauthorized hosts
2676 would not be able to hijack associations. However, older
2677 implementations may not have implemented this requirement while
2678 allowing the ADD-IP extension. For reasons of interoperability,
2679 we provide this variable to control the enforcement of the
2680 authentication requirement.
2682 == ===============================================================
2683 1 Allow ADD-IP extension to be used without authentication. This
2684 should only be set in a closed environment for interoperability
2685 with older implementations.
2687 0 Enforce the authentication requirement
2688 == ===============================================================
2692 auth_enable - BOOLEAN
2693 Enable or disable Authenticated Chunks extension. This extension
2694 provides the ability to send and receive authenticated chunks and is
2695 required for secure operation of Dynamic Address Reconfiguration
2698 - 1: Enable this extension.
2699 - 0: Disable this extension.
2703 prsctp_enable - BOOLEAN
2704 Enable or disable the Partial Reliability extension (RFC3758) which
2705 is used to notify peers that a given DATA should no longer be expected.
2707 - 1: Enable extension
2713 The limit of the number of new packets that can be initially sent. It
2714 controls how bursty the generated traffic can be.
2718 association_max_retrans - INTEGER
2719 Set the maximum number for retransmissions that an association can
2720 attempt deciding that the remote end is unreachable. If this value
2721 is exceeded, the association is terminated.
2725 max_init_retransmits - INTEGER
2726 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
2727 that an association will attempt before declaring the destination
2728 unreachable and terminating.
2732 path_max_retrans - INTEGER
2733 The maximum number of retransmissions that will be attempted on a given
2734 path. Once this threshold is exceeded, the path is considered
2735 unreachable, and new traffic will use a different path when the
2736 association is multihomed.
2740 pf_retrans - INTEGER
2741 The number of retransmissions that will be attempted on a given path
2742 before traffic is redirected to an alternate transport (should one
2743 exist). Note this is distinct from path_max_retrans, as a path that
2744 passes the pf_retrans threshold can still be used. Its only
2745 deprioritized when a transmission path is selected by the stack. This
2746 setting is primarily used to enable fast failover mechanisms without
2747 having to reduce path_max_retrans to a very low value. See:
2748 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
2749 for details. Note also that a value of pf_retrans > path_max_retrans
2750 disables this feature. Since both pf_retrans and path_max_retrans can
2751 be changed by userspace application, a variable pf_enable is used to
2756 ps_retrans - INTEGER
2757 Primary.Switchover.Max.Retrans (PSMR), it's a tunable parameter coming
2758 from section-5 "Primary Path Switchover" in rfc7829. The primary path
2759 will be changed to another active path when the path error counter on
2760 the old primary path exceeds PSMR, so that "the SCTP sender is allowed
2761 to continue data transmission on a new working path even when the old
2762 primary destination address becomes active again". Note this feature
2763 is disabled by initializing 'ps_retrans' per netns as 0xffff by default,
2764 and its value can't be less than 'pf_retrans' when changing by sysctl.
2768 rto_initial - INTEGER
2769 The initial round trip timeout value in milliseconds that will be used
2770 in calculating round trip times. This is the initial time interval
2771 for retransmissions.
2776 The maximum value (in milliseconds) of the round trip timeout. This
2777 is the largest time interval that can elapse between retransmissions.
2782 The minimum value (in milliseconds) of the round trip timeout. This
2783 is the smallest time interval the can elapse between retransmissions.
2787 hb_interval - INTEGER
2788 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
2789 are sent at the specified interval on idle paths to probe the state of
2790 a given path between 2 associations.
2794 sack_timeout - INTEGER
2795 The amount of time (in milliseconds) that the implementation will wait
2800 valid_cookie_life - INTEGER
2801 The default lifetime of the SCTP cookie (in milliseconds). The cookie
2802 is used during association establishment.
2806 cookie_preserve_enable - BOOLEAN
2807 Enable or disable the ability to extend the lifetime of the SCTP cookie
2808 that is used during the establishment phase of SCTP association
2810 - 1: Enable cookie lifetime extension.
2815 cookie_hmac_alg - STRING
2816 Select the hmac algorithm used when generating the cookie value sent by
2817 a listening sctp socket to a connecting client in the INIT-ACK chunk.
2824 Ability to assign md5 or sha1 as the selected alg is predicated on the
2825 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
2826 CONFIG_CRYPTO_SHA1).
2828 Default: Dependent on configuration. MD5 if available, else SHA1 if
2829 available, else none.
2831 rcvbuf_policy - INTEGER
2832 Determines if the receive buffer is attributed to the socket or to
2833 association. SCTP supports the capability to create multiple
2834 associations on a single socket. When using this capability, it is
2835 possible that a single stalled association that's buffering a lot
2836 of data may block other associations from delivering their data by
2837 consuming all of the receive buffer space. To work around this,
2838 the rcvbuf_policy could be set to attribute the receiver buffer space
2839 to each association instead of the socket. This prevents the described
2842 - 1: rcvbuf space is per association
2843 - 0: rcvbuf space is per socket
2847 sndbuf_policy - INTEGER
2848 Similar to rcvbuf_policy above, this applies to send buffer space.
2850 - 1: Send buffer is tracked per association
2851 - 0: Send buffer is tracked per socket.
2855 sctp_mem - vector of 3 INTEGERs: min, pressure, max
2856 Number of pages allowed for queueing by all SCTP sockets.
2858 min: Below this number of pages SCTP is not bothered about its
2859 memory appetite. When amount of memory allocated by SCTP exceeds
2860 this number, SCTP starts to moderate memory usage.
2862 pressure: This value was introduced to follow format of tcp_mem.
2864 max: Number of pages allowed for queueing by all SCTP sockets.
2866 Default is calculated at boot time from amount of available memory.
2868 sctp_rmem - vector of 3 INTEGERs: min, default, max
2869 Only the first value ("min") is used, "default" and "max" are
2872 min: Minimal size of receive buffer used by SCTP socket.
2873 It is guaranteed to each SCTP socket (but not association) even
2874 under moderate memory pressure.
2878 sctp_wmem - vector of 3 INTEGERs: min, default, max
2879 Currently this tunable has no effect.
2881 addr_scope_policy - INTEGER
2882 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
2884 - 0 - Disable IPv4 address scoping
2885 - 1 - Enable IPv4 address scoping
2886 - 2 - Follow draft but allow IPv4 private addresses
2887 - 3 - Follow draft but allow IPv4 link local addresses
2892 The listening port for the local UDP tunneling sock. Normally it's
2893 using the IANA-assigned UDP port number 9899 (sctp-tunneling).
2895 This UDP sock is used for processing the incoming UDP-encapsulated
2896 SCTP packets (from RFC6951), and shared by all applications in the
2897 same net namespace. This UDP sock will be closed when the value is
2900 The value will also be used to set the src port of the UDP header
2901 for the outgoing UDP-encapsulated SCTP packets. For the dest port,
2902 please refer to 'encap_port' below.
2906 encap_port - INTEGER
2907 The default remote UDP encapsulation port.
2909 This value is used to set the dest port of the UDP header for the
2910 outgoing UDP-encapsulated SCTP packets by default. Users can also
2911 change the value for each sock/asoc/transport by using setsockopt.
2912 For further information, please refer to RFC6951.
2914 Note that when connecting to a remote server, the client should set
2915 this to the port that the UDP tunneling sock on the peer server is
2916 listening to and the local UDP tunneling sock on the client also
2917 must be started. On the server, it would get the encap_port from
2918 the incoming packet's source port.
2922 plpmtud_probe_interval - INTEGER
2923 The time interval (in milliseconds) for the PLPMTUD probe timer,
2924 which is configured to expire after this period to receive an
2925 acknowledgment to a probe packet. This is also the time interval
2926 between the probes for the current pmtu when the probe search
2929 PLPMTUD will be disabled when 0 is set, and other values for it
2934 reconf_enable - BOOLEAN
2935 Enable or disable extension of Stream Reconfiguration functionality
2936 specified in RFC6525. This extension provides the ability to "reset"
2937 a stream, and it includes the Parameters of "Outgoing/Incoming SSN
2938 Reset", "SSN/TSN Reset" and "Add Outgoing/Incoming Streams".
2940 - 1: Enable extension.
2941 - 0: Disable extension.
2945 intl_enable - BOOLEAN
2946 Enable or disable extension of User Message Interleaving functionality
2947 specified in RFC8260. This extension allows the interleaving of user
2948 messages sent on different streams. With this feature enabled, I-DATA
2949 chunk will replace DATA chunk to carry user messages if also supported
2950 by the peer. Note that to use this feature, one needs to set this option
2951 to 1 and also needs to set socket options SCTP_FRAGMENT_INTERLEAVE to 2
2952 and SCTP_INTERLEAVING_SUPPORTED to 1.
2954 - 1: Enable extension.
2955 - 0: Disable extension.
2959 ecn_enable - BOOLEAN
2960 Control use of Explicit Congestion Notification (ECN) by SCTP.
2961 Like in TCP, ECN is used only when both ends of the SCTP connection
2962 indicate support for it. This feature is useful in avoiding losses
2963 due to congestion by allowing supporting routers to signal congestion
2964 before having to drop packets.
2972 ``/proc/sys/net/core/*``
2973 ========================
2975 Please see: Documentation/admin-guide/sysctl/net.rst for descriptions of these entries.
2978 ``/proc/sys/net/unix/*``
2979 ========================
2981 max_dgram_qlen - INTEGER
2982 The maximum length of dgram socket receive queue