1 .. SPDX-License-Identifier: GPL-2.0
7 /proc/sys/net/ipv4/* Variables
8 ==============================
11 - 0 - disabled (default)
14 Forward Packets between interfaces.
16 This variable is special, its change resets all configuration
17 parameters to their default state (RFC1122 for hosts, RFC1812
20 ip_default_ttl - INTEGER
21 Default value of TTL field (Time To Live) for outgoing (but not
22 forwarded) IP packets. Should be between 1 and 255 inclusive.
23 Default: 64 (as recommended by RFC1700)
25 ip_no_pmtu_disc - INTEGER
26 Disable Path MTU Discovery. If enabled in mode 1 and a
27 fragmentation-required ICMP is received, the PMTU to this
28 destination will be set to min_pmtu (see below). You will need
29 to raise min_pmtu to the smallest interface MTU on your system
30 manually if you want to avoid locally generated fragments.
32 In mode 2 incoming Path MTU Discovery messages will be
33 discarded. Outgoing frames are handled the same as in mode 1,
34 implicitly setting IP_PMTUDISC_DONT on every created socket.
36 Mode 3 is a hardened pmtu discover mode. The kernel will only
37 accept fragmentation-needed errors if the underlying protocol
38 can verify them besides a plain socket lookup. Current
39 protocols for which pmtu events will be honored are TCP, SCTP
40 and DCCP as they verify e.g. the sequence number or the
41 association. This mode should not be enabled globally but is
42 only intended to secure e.g. name servers in namespaces where
43 TCP path mtu must still work but path MTU information of other
44 protocols should be discarded. If enabled globally this mode
45 could break other protocols.
52 default 552 - minimum discovered Path MTU
54 ip_forward_use_pmtu - BOOLEAN
55 By default we don't trust protocol path MTUs while forwarding
56 because they could be easily forged and can lead to unwanted
57 fragmentation by the router.
58 You only need to enable this if you have user-space software
59 which tries to discover path mtus by itself and depends on the
60 kernel honoring this information. This is normally not the
70 fwmark_reflect - BOOLEAN
71 Controls the fwmark of kernel-generated IPv4 reply packets that are not
72 associated with a socket for example, TCP RSTs or ICMP echo replies).
73 If unset, these packets have a fwmark of zero. If set, they have the
74 fwmark of the packet they are replying to.
78 fib_multipath_use_neigh - BOOLEAN
79 Use status of existing neighbor entry when determining nexthop for
80 multipath routes. If disabled, neighbor information is not used and
81 packets could be directed to a failed nexthop. Only valid for kernels
82 built with CONFIG_IP_ROUTE_MULTIPATH enabled.
91 fib_multipath_hash_policy - INTEGER
92 Controls which hash policy to use for multipath routes. Only valid
93 for kernels built with CONFIG_IP_ROUTE_MULTIPATH enabled.
101 - 2 - Layer 3 or inner Layer 3 if present
103 fib_sync_mem - UNSIGNED INTEGER
104 Amount of dirty memory from fib entries that can be backlogged before
105 synchronize_rcu is forced.
107 Default: 512kB Minimum: 64kB Maximum: 64MB
109 ip_forward_update_priority - INTEGER
110 Whether to update SKB priority from "TOS" field in IPv4 header after it
111 is forwarded. The new SKB priority is mapped from TOS field value
112 according to an rt_tos2priority table (see e.g. man tc-prio).
114 Default: 1 (Update priority.)
118 - 0 - Do not update priority.
119 - 1 - Update priority.
121 route/max_size - INTEGER
122 Maximum number of routes allowed in the kernel. Increase
123 this when using large numbers of interfaces and/or routes.
125 From linux kernel 3.6 onwards, this is deprecated for ipv4
126 as route cache is no longer used.
128 neigh/default/gc_thresh1 - INTEGER
129 Minimum number of entries to keep. Garbage collector will not
130 purge entries if there are fewer than this number.
134 neigh/default/gc_thresh2 - INTEGER
135 Threshold when garbage collector becomes more aggressive about
136 purging entries. Entries older than 5 seconds will be cleared
137 when over this number.
141 neigh/default/gc_thresh3 - INTEGER
142 Maximum number of non-PERMANENT neighbor entries allowed. Increase
143 this when using large numbers of interfaces and when communicating
144 with large numbers of directly-connected peers.
148 neigh/default/unres_qlen_bytes - INTEGER
149 The maximum number of bytes which may be used by packets
150 queued for each unresolved address by other network layers.
153 Setting negative value is meaningless and will return error.
155 Default: SK_WMEM_MAX, (same as net.core.wmem_default).
157 Exact value depends on architecture and kernel options,
158 but should be enough to allow queuing 256 packets
161 neigh/default/unres_qlen - INTEGER
162 The maximum number of packets which may be queued for each
163 unresolved address by other network layers.
165 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
167 Prior to linux 3.3, the default value is 3 which may cause
168 unexpected packet loss. The current default value is calculated
169 according to default value of unres_qlen_bytes and true size of
174 mtu_expires - INTEGER
175 Time, in seconds, that cached PMTU information is kept.
177 min_adv_mss - INTEGER
178 The advertised MSS depends on the first hop route MTU, but will
179 never be lower than this setting.
181 fib_notify_on_flag_change - INTEGER
182 Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/
183 RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed.
185 After installing a route to the kernel, user space receives an
186 acknowledgment, which means the route was installed in the kernel,
187 but not necessarily in hardware.
188 It is also possible for a route already installed in hardware to change
189 its action and therefore its flags. For example, a host route that is
190 trapping packets can be "promoted" to perform decapsulation following
191 the installation of an IPinIP/VXLAN tunnel.
192 The notifications will indicate to user-space the state of the route.
194 Default: 0 (Do not emit notifications.)
198 - 0 - Do not emit notifications.
199 - 1 - Emit notifications.
200 - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change.
204 ipfrag_high_thresh - LONG INTEGER
205 Maximum memory used to reassemble IP fragments.
207 ipfrag_low_thresh - LONG INTEGER
208 (Obsolete since linux-4.17)
209 Maximum memory used to reassemble IP fragments before the kernel
210 begins to remove incomplete fragment queues to free up resources.
211 The kernel still accepts new fragments for defragmentation.
213 ipfrag_time - INTEGER
214 Time in seconds to keep an IP fragment in memory.
216 ipfrag_max_dist - INTEGER
217 ipfrag_max_dist is a non-negative integer value which defines the
218 maximum "disorder" which is allowed among fragments which share a
219 common IP source address. Note that reordering of packets is
220 not unusual, but if a large number of fragments arrive from a source
221 IP address while a particular fragment queue remains incomplete, it
222 probably indicates that one or more fragments belonging to that queue
223 have been lost. When ipfrag_max_dist is positive, an additional check
224 is done on fragments before they are added to a reassembly queue - if
225 ipfrag_max_dist (or more) fragments have arrived from a particular IP
226 address between additions to any IP fragment queue using that source
227 address, it's presumed that one or more fragments in the queue are
228 lost. The existing fragment queue will be dropped, and a new one
229 started. An ipfrag_max_dist value of zero disables this check.
231 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
232 result in unnecessarily dropping fragment queues when normal
233 reordering of packets occurs, which could lead to poor application
234 performance. Using a very large value, e.g. 50000, increases the
235 likelihood of incorrectly reassembling IP fragments that originate
236 from different IP datagrams, which could result in data corruption.
242 inet_peer_threshold - INTEGER
243 The approximate size of the storage. Starting from this threshold
244 entries will be thrown aggressively. This threshold also determines
245 entries' time-to-live and time intervals between garbage collection
246 passes. More entries, less time-to-live, less GC interval.
248 inet_peer_minttl - INTEGER
249 Minimum time-to-live of entries. Should be enough to cover fragment
250 time-to-live on the reassembling side. This minimum time-to-live is
251 guaranteed if the pool size is less than inet_peer_threshold.
254 inet_peer_maxttl - INTEGER
255 Maximum time-to-live of entries. Unused entries will expire after
256 this period of time if there is no memory pressure on the pool (i.e.
257 when the number of entries in the pool is very small).
264 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
265 Defaults to 4096. (Was 128 before linux-5.4)
266 See also tcp_max_syn_backlog for additional tuning for TCP sockets.
268 tcp_abort_on_overflow - BOOLEAN
269 If listening service is too slow to accept new connections,
270 reset them. Default state is FALSE. It means that if overflow
271 occurred due to a burst, connection will recover. Enable this
272 option _only_ if you are really sure that listening daemon
273 cannot be tuned to accept connections faster. Enabling this
274 option can harm clients of your server.
276 tcp_adv_win_scale - INTEGER
277 Count buffering overhead as bytes/2^tcp_adv_win_scale
278 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
281 Possible values are [-31, 31], inclusive.
285 tcp_allowed_congestion_control - STRING
286 Show/set the congestion control choices available to non-privileged
287 processes. The list is a subset of those listed in
288 tcp_available_congestion_control.
290 Default is "reno" and the default setting (tcp_congestion_control).
292 tcp_app_win - INTEGER
293 Reserve max(window/2^tcp_app_win, mss) of window for application
294 buffer. Value 0 is special, it means that nothing is reserved.
298 tcp_autocorking - BOOLEAN
299 Enable TCP auto corking :
300 When applications do consecutive small write()/sendmsg() system calls,
301 we try to coalesce these small writes as much as possible, to lower
302 total amount of sent packets. This is done if at least one prior
303 packet for the flow is waiting in Qdisc queues or device transmit
304 queue. Applications can still use TCP_CORK for optimal behavior
305 when they know how/when to uncork their sockets.
309 tcp_available_congestion_control - STRING
310 Shows the available congestion control choices that are registered.
311 More congestion control algorithms may be available as modules,
314 tcp_base_mss - INTEGER
315 The initial value of search_low to be used by the packetization layer
316 Path MTU discovery (MTU probing). If MTU probing is enabled,
317 this is the initial MSS used by the connection.
319 tcp_mtu_probe_floor - INTEGER
320 If MTU probing is enabled this caps the minimum MSS used for search_low
325 tcp_min_snd_mss - INTEGER
326 TCP SYN and SYNACK messages usually advertise an ADVMSS option,
327 as described in RFC 1122 and RFC 6691.
329 If this ADVMSS option is smaller than tcp_min_snd_mss,
330 it is silently capped to tcp_min_snd_mss.
332 Default : 48 (at least 8 bytes of payload per segment)
334 tcp_congestion_control - STRING
335 Set the congestion control algorithm to be used for new
336 connections. The algorithm "reno" is always available, but
337 additional choices may be available based on kernel configuration.
338 Default is set as part of kernel configuration.
339 For passive connections, the listener congestion control choice
342 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
345 Allows TCP to send "duplicate" SACKs.
347 tcp_early_retrans - INTEGER
348 Tail loss probe (TLP) converts RTOs occurring due to tail
349 losses into fast recovery (draft-ietf-tcpm-rack). Note that
350 TLP requires RACK to function properly (see tcp_recovery below)
360 Control use of Explicit Congestion Notification (ECN) by TCP.
361 ECN is used only when both ends of the TCP connection indicate
362 support for it. This feature is useful in avoiding losses due
363 to congestion by allowing supporting routers to signal
364 congestion before having to drop packets.
368 = =====================================================
369 0 Disable ECN. Neither initiate nor accept ECN.
370 1 Enable ECN when requested by incoming connections and
371 also request ECN on outgoing connection attempts.
372 2 Enable ECN when requested by incoming connections
373 but do not request ECN on outgoing connections.
374 = =====================================================
378 tcp_ecn_fallback - BOOLEAN
379 If the kernel detects that ECN connection misbehaves, enable fall
380 back to non-ECN. Currently, this knob implements the fallback
381 from RFC3168, section 6.1.1.1., but we reserve that in future,
382 additional detection mechanisms could be implemented under this
383 knob. The value is not used, if tcp_ecn or per route (or congestion
384 control) ECN settings are disabled.
386 Default: 1 (fallback enabled)
389 This is a legacy option, it has no effect anymore.
391 tcp_fin_timeout - INTEGER
392 The length of time an orphaned (no longer referenced by any
393 application) connection will remain in the FIN_WAIT_2 state
394 before it is aborted at the local end. While a perfectly
395 valid "receive only" state for an un-orphaned connection, an
396 orphaned connection in FIN_WAIT_2 state could otherwise wait
397 forever for the remote to close its end of the connection.
404 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
405 F-RTO is an enhanced recovery algorithm for TCP retransmission
406 timeouts. It is particularly beneficial in networks where the
407 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
408 modification. It does not require any support from the peer.
410 By default it's enabled with a non-zero value. 0 disables F-RTO.
412 tcp_fwmark_accept - BOOLEAN
413 If set, incoming connections to listening sockets that do not have a
414 socket mark will set the mark of the accepting socket to the fwmark of
415 the incoming SYN packet. This will cause all packets on that connection
416 (starting from the first SYNACK) to be sent with that fwmark. The
417 listening socket's mark is unchanged. Listening sockets that already
418 have a fwmark set via setsockopt(SOL_SOCKET, SO_MARK, ...) are
423 tcp_invalid_ratelimit - INTEGER
424 Limit the maximal rate for sending duplicate acknowledgments
425 in response to incoming TCP packets that are for an existing
426 connection but that are invalid due to any of these reasons:
428 (a) out-of-window sequence number,
429 (b) out-of-window acknowledgment number, or
430 (c) PAWS (Protection Against Wrapped Sequence numbers) check failure
432 This can help mitigate simple "ack loop" DoS attacks, wherein
433 a buggy or malicious middlebox or man-in-the-middle can
434 rewrite TCP header fields in manner that causes each endpoint
435 to think that the other is sending invalid TCP segments, thus
436 causing each side to send an unterminating stream of duplicate
437 acknowledgments for invalid segments.
439 Using 0 disables rate-limiting of dupacks in response to
440 invalid segments; otherwise this value specifies the minimal
441 space between sending such dupacks, in milliseconds.
443 Default: 500 (milliseconds).
445 tcp_keepalive_time - INTEGER
446 How often TCP sends out keepalive messages when keepalive is enabled.
449 tcp_keepalive_probes - INTEGER
450 How many keepalive probes TCP sends out, until it decides that the
451 connection is broken. Default value: 9.
453 tcp_keepalive_intvl - INTEGER
454 How frequently the probes are send out. Multiplied by
455 tcp_keepalive_probes it is time to kill not responding connection,
456 after probes started. Default value: 75sec i.e. connection
457 will be aborted after ~11 minutes of retries.
459 tcp_l3mdev_accept - BOOLEAN
460 Enables child sockets to inherit the L3 master device index.
461 Enabling this option allows a "global" listen socket to work
462 across L3 master domains (e.g., VRFs) with connected sockets
463 derived from the listen socket to be bound to the L3 domain in
464 which the packets originated. Only valid when the kernel was
465 compiled with CONFIG_NET_L3_MASTER_DEV.
467 Default: 0 (disabled)
469 tcp_low_latency - BOOLEAN
470 This is a legacy option, it has no effect anymore.
472 tcp_max_orphans - INTEGER
473 Maximal number of TCP sockets not attached to any user file handle,
474 held by system. If this number is exceeded orphaned connections are
475 reset immediately and warning is printed. This limit exists
476 only to prevent simple DoS attacks, you _must_ not rely on this
477 or lower the limit artificially, but rather increase it
478 (probably, after increasing installed memory),
479 if network conditions require more than default value,
480 and tune network services to linger and kill such states
481 more aggressively. Let me to remind again: each orphan eats
482 up to ~64K of unswappable memory.
484 tcp_max_syn_backlog - INTEGER
485 Maximal number of remembered connection requests (SYN_RECV),
486 which have not received an acknowledgment from connecting client.
488 This is a per-listener limit.
490 The minimal value is 128 for low memory machines, and it will
491 increase in proportion to the memory of machine.
493 If server suffers from overload, try increasing this number.
495 Remember to also check /proc/sys/net/core/somaxconn
496 A SYN_RECV request socket consumes about 304 bytes of memory.
498 tcp_max_tw_buckets - INTEGER
499 Maximal number of timewait sockets held by system simultaneously.
500 If this number is exceeded time-wait socket is immediately destroyed
501 and warning is printed. This limit exists only to prevent
502 simple DoS attacks, you _must_ not lower the limit artificially,
503 but rather increase it (probably, after increasing installed memory),
504 if network conditions require more than default value.
506 tcp_mem - vector of 3 INTEGERs: min, pressure, max
507 min: below this number of pages TCP is not bothered about its
510 pressure: when amount of memory allocated by TCP exceeds this number
511 of pages, TCP moderates its memory consumption and enters memory
512 pressure mode, which is exited when memory consumption falls
515 max: number of pages allowed for queueing by all TCP sockets.
517 Defaults are calculated at boot time from amount of available
520 tcp_min_rtt_wlen - INTEGER
521 The window length of the windowed min filter to track the minimum RTT.
522 A shorter window lets a flow more quickly pick up new (higher)
523 minimum RTT when it is moved to a longer path (e.g., due to traffic
524 engineering). A longer window makes the filter more resistant to RTT
525 inflations such as transient congestion. The unit is seconds.
527 Possible values: 0 - 86400 (1 day)
531 tcp_moderate_rcvbuf - BOOLEAN
532 If set, TCP performs receive buffer auto-tuning, attempting to
533 automatically size the buffer (no greater than tcp_rmem[2]) to
534 match the size required by the path for full throughput. Enabled by
537 tcp_mtu_probing - INTEGER
538 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
542 - 1 - Disabled by default, enabled when an ICMP black hole detected
543 - 2 - Always enabled, use initial MSS of tcp_base_mss.
545 tcp_probe_interval - UNSIGNED INTEGER
546 Controls how often to start TCP Packetization-Layer Path MTU
547 Discovery reprobe. The default is reprobing every 10 minutes as
550 tcp_probe_threshold - INTEGER
551 Controls when TCP Packetization-Layer Path MTU Discovery probing
552 will stop in respect to the width of search range in bytes. Default
555 tcp_no_metrics_save - BOOLEAN
556 By default, TCP saves various connection metrics in the route cache
557 when the connection closes, so that connections established in the
558 near future can use these to set initial conditions. Usually, this
559 increases overall performance, but may sometimes cause performance
560 degradation. If set, TCP will not cache metrics on closing
563 tcp_no_ssthresh_metrics_save - BOOLEAN
564 Controls whether TCP saves ssthresh metrics in the route cache.
566 Default is 1, which disables ssthresh metrics.
568 tcp_orphan_retries - INTEGER
569 This value influences the timeout of a locally closed TCP connection,
570 when RTO retransmissions remain unacknowledged.
571 See tcp_retries2 for more details.
573 The default value is 8.
575 If your machine is a loaded WEB server,
576 you should think about lowering this value, such sockets
577 may consume significant resources. Cf. tcp_max_orphans.
579 tcp_recovery - INTEGER
580 This value is a bitmap to enable various experimental loss recovery
583 ========= =============================================================
584 RACK: 0x1 enables the RACK loss detection for fast detection of lost
585 retransmissions and tail drops. It also subsumes and disables
586 RFC6675 recovery for SACK connections.
588 RACK: 0x2 makes RACK's reordering window static (min_rtt/4).
590 RACK: 0x4 disables RACK's DUPACK threshold heuristic
591 ========= =============================================================
595 tcp_reordering - INTEGER
596 Initial reordering level of packets in a TCP stream.
597 TCP stack can then dynamically adjust flow reordering level
598 between this initial value and tcp_max_reordering
602 tcp_max_reordering - INTEGER
603 Maximal reordering level of packets in a TCP stream.
604 300 is a fairly conservative value, but you might increase it
605 if paths are using per packet load balancing (like bonding rr mode)
609 tcp_retrans_collapse - BOOLEAN
610 Bug-to-bug compatibility with some broken printers.
611 On retransmit try to send bigger packets to work around bugs in
614 tcp_retries1 - INTEGER
615 This value influences the time, after which TCP decides, that
616 something is wrong due to unacknowledged RTO retransmissions,
617 and reports this suspicion to the network layer.
618 See tcp_retries2 for more details.
620 RFC 1122 recommends at least 3 retransmissions, which is the
623 tcp_retries2 - INTEGER
624 This value influences the timeout of an alive TCP connection,
625 when RTO retransmissions remain unacknowledged.
626 Given a value of N, a hypothetical TCP connection following
627 exponential backoff with an initial RTO of TCP_RTO_MIN would
628 retransmit N times before killing the connection at the (N+1)th RTO.
630 The default value of 15 yields a hypothetical timeout of 924.6
631 seconds and is a lower bound for the effective timeout.
632 TCP will effectively time out at the first RTO which exceeds the
633 hypothetical timeout.
635 RFC 1122 recommends at least 100 seconds for the timeout,
636 which corresponds to a value of at least 8.
638 tcp_rfc1337 - BOOLEAN
639 If set, the TCP stack behaves conforming to RFC1337. If unset,
640 we are not conforming to RFC, but prevent TCP TIME_WAIT
645 tcp_rmem - vector of 3 INTEGERs: min, default, max
646 min: Minimal size of receive buffer used by TCP sockets.
647 It is guaranteed to each TCP socket, even under moderate memory
652 default: initial size of receive buffer used by TCP sockets.
653 This value overrides net.core.rmem_default used by other protocols.
654 Default: 131072 bytes.
655 This value results in initial window of 65535.
657 max: maximal size of receive buffer allowed for automatically
658 selected receiver buffers for TCP socket. This value does not override
659 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
660 automatic tuning of that socket's receive buffer size, in which
661 case this value is ignored.
662 Default: between 131072 and 6MB, depending on RAM size.
665 Enable select acknowledgments (SACKS).
667 tcp_comp_sack_delay_ns - LONG INTEGER
668 TCP tries to reduce number of SACK sent, using a timer
669 based on 5% of SRTT, capped by this sysctl, in nano seconds.
670 The default is 1ms, based on TSO autosizing period.
672 Default : 1,000,000 ns (1 ms)
674 tcp_comp_sack_slack_ns - LONG INTEGER
675 This sysctl control the slack used when arming the
676 timer used by SACK compression. This gives extra time
677 for small RTT flows, and reduces system overhead by allowing
678 opportunistic reduction of timer interrupts.
680 Default : 100,000 ns (100 us)
682 tcp_comp_sack_nr - INTEGER
683 Max number of SACK that can be compressed.
684 Using 0 disables SACK compression.
688 tcp_slow_start_after_idle - BOOLEAN
689 If set, provide RFC2861 behavior and time out the congestion
690 window after an idle period. An idle period is defined at
691 the current RTO. If unset, the congestion window will not
692 be timed out after an idle period.
697 Use the Host requirements interpretation of the TCP urgent pointer field.
698 Most hosts use the older BSD interpretation, so if you turn this on
699 Linux might not communicate correctly with them.
703 tcp_synack_retries - INTEGER
704 Number of times SYNACKs for a passive TCP connection attempt will
705 be retransmitted. Should not be higher than 255. Default value
706 is 5, which corresponds to 31seconds till the last retransmission
707 with the current initial RTO of 1second. With this the final timeout
708 for a passive TCP connection will happen after 63seconds.
710 tcp_syncookies - INTEGER
711 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
712 Send out syncookies when the syn backlog queue of a socket
713 overflows. This is to prevent against the common 'SYN flood attack'
716 Note, that syncookies is fallback facility.
717 It MUST NOT be used to help highly loaded servers to stand
718 against legal connection rate. If you see SYN flood warnings
719 in your logs, but investigation shows that they occur
720 because of overload with legal connections, you should tune
721 another parameters until this warning disappear.
722 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
724 syncookies seriously violate TCP protocol, do not allow
725 to use TCP extensions, can result in serious degradation
726 of some services (f.e. SMTP relaying), visible not by you,
727 but your clients and relays, contacting you. While you see
728 SYN flood warnings in logs not being really flooded, your server
729 is seriously misconfigured.
731 If you want to test which effects syncookies have to your
732 network connections you can set this knob to 2 to enable
733 unconditionally generation of syncookies.
735 tcp_fastopen - INTEGER
736 Enable TCP Fast Open (RFC7413) to send and accept data in the opening
739 The client support is enabled by flag 0x1 (on by default). The client
740 then must use sendmsg() or sendto() with the MSG_FASTOPEN flag,
741 rather than connect() to send data in SYN.
743 The server support is enabled by flag 0x2 (off by default). Then
744 either enable for all listeners with another flag (0x400) or
745 enable individual listeners via TCP_FASTOPEN socket option with
746 the option value being the length of the syn-data backlog.
748 The values (bitmap) are
750 ===== ======== ======================================================
751 0x1 (client) enables sending data in the opening SYN on the client.
752 0x2 (server) enables the server support, i.e., allowing data in
753 a SYN packet to be accepted and passed to the
754 application before 3-way handshake finishes.
755 0x4 (client) send data in the opening SYN regardless of cookie
756 availability and without a cookie option.
757 0x200 (server) accept data-in-SYN w/o any cookie option present.
758 0x400 (server) enable all listeners to support Fast Open by
759 default without explicit TCP_FASTOPEN socket option.
760 ===== ======== ======================================================
764 Note that additional client or server features are only
765 effective if the basic support (0x1 and 0x2) are enabled respectively.
767 tcp_fastopen_blackhole_timeout_sec - INTEGER
768 Initial time period in second to disable Fastopen on active TCP sockets
769 when a TFO firewall blackhole issue happens.
770 This time period will grow exponentially when more blackhole issues
771 get detected right after Fastopen is re-enabled and will reset to
772 initial value when the blackhole issue goes away.
773 0 to disable the blackhole detection.
775 By default, it is set to 1hr.
777 tcp_fastopen_key - list of comma separated 32-digit hexadecimal INTEGERs
778 The list consists of a primary key and an optional backup key. The
779 primary key is used for both creating and validating cookies, while the
780 optional backup key is only used for validating cookies. The purpose of
781 the backup key is to maximize TFO validation when keys are rotated.
783 A randomly chosen primary key may be configured by the kernel if
784 the tcp_fastopen sysctl is set to 0x400 (see above), or if the
785 TCP_FASTOPEN setsockopt() optname is set and a key has not been
786 previously configured via sysctl. If keys are configured via
787 setsockopt() by using the TCP_FASTOPEN_KEY optname, then those
788 per-socket keys will be used instead of any keys that are specified via
791 A key is specified as 4 8-digit hexadecimal integers which are separated
792 by a '-' as: xxxxxxxx-xxxxxxxx-xxxxxxxx-xxxxxxxx. Leading zeros may be
793 omitted. A primary and a backup key may be specified by separating them
794 by a comma. If only one key is specified, it becomes the primary key and
795 any previously configured backup keys are removed.
797 tcp_syn_retries - INTEGER
798 Number of times initial SYNs for an active TCP connection attempt
799 will be retransmitted. Should not be higher than 127. Default value
800 is 6, which corresponds to 63seconds till the last retransmission
801 with the current initial RTO of 1second. With this the final timeout
802 for an active TCP connection attempt will happen after 127seconds.
804 tcp_timestamps - INTEGER
805 Enable timestamps as defined in RFC1323.
808 - 1: Enable timestamps as defined in RFC1323 and use random offset for
809 each connection rather than only using the current time.
810 - 2: Like 1, but without random offsets.
814 tcp_min_tso_segs - INTEGER
815 Minimal number of segments per TSO frame.
817 Since linux-3.12, TCP does an automatic sizing of TSO frames,
818 depending on flow rate, instead of filling 64Kbytes packets.
819 For specific usages, it's possible to force TCP to build big
820 TSO frames. Note that TCP stack might split too big TSO packets
821 if available window is too small.
825 tcp_pacing_ss_ratio - INTEGER
826 sk->sk_pacing_rate is set by TCP stack using a ratio applied
827 to current rate. (current_rate = cwnd * mss / srtt)
828 If TCP is in slow start, tcp_pacing_ss_ratio is applied
829 to let TCP probe for bigger speeds, assuming cwnd can be
830 doubled every other RTT.
834 tcp_pacing_ca_ratio - INTEGER
835 sk->sk_pacing_rate is set by TCP stack using a ratio applied
836 to current rate. (current_rate = cwnd * mss / srtt)
837 If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio
838 is applied to conservatively probe for bigger throughput.
842 tcp_tso_win_divisor - INTEGER
843 This allows control over what percentage of the congestion window
844 can be consumed by a single TSO frame.
845 The setting of this parameter is a choice between burstiness and
846 building larger TSO frames.
850 tcp_tw_reuse - INTEGER
851 Enable reuse of TIME-WAIT sockets for new connections when it is
852 safe from protocol viewpoint.
856 - 2 - enable for loopback traffic only
858 It should not be changed without advice/request of technical
863 tcp_window_scaling - BOOLEAN
864 Enable window scaling as defined in RFC1323.
866 tcp_wmem - vector of 3 INTEGERs: min, default, max
867 min: Amount of memory reserved for send buffers for TCP sockets.
868 Each TCP socket has rights to use it due to fact of its birth.
872 default: initial size of send buffer used by TCP sockets. This
873 value overrides net.core.wmem_default used by other protocols.
875 It is usually lower than net.core.wmem_default.
879 max: Maximal amount of memory allowed for automatically tuned
880 send buffers for TCP sockets. This value does not override
881 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
882 automatic tuning of that socket's send buffer size, in which case
883 this value is ignored.
885 Default: between 64K and 4MB, depending on RAM size.
887 tcp_notsent_lowat - UNSIGNED INTEGER
888 A TCP socket can control the amount of unsent bytes in its write queue,
889 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
890 reports POLLOUT events if the amount of unsent bytes is below a per
891 socket value, and if the write queue is not full. sendmsg() will
892 also not add new buffers if the limit is hit.
894 This global variable controls the amount of unsent data for
895 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
896 to the global variable has immediate effect.
898 Default: UINT_MAX (0xFFFFFFFF)
900 tcp_workaround_signed_windows - BOOLEAN
901 If set, assume no receipt of a window scaling option means the
902 remote TCP is broken and treats the window as a signed quantity.
903 If unset, assume the remote TCP is not broken even if we do
904 not receive a window scaling option from them.
908 tcp_thin_linear_timeouts - BOOLEAN
909 Enable dynamic triggering of linear timeouts for thin streams.
910 If set, a check is performed upon retransmission by timeout to
911 determine if the stream is thin (less than 4 packets in flight).
912 As long as the stream is found to be thin, up to 6 linear
913 timeouts may be performed before exponential backoff mode is
914 initiated. This improves retransmission latency for
915 non-aggressive thin streams, often found to be time-dependent.
916 For more information on thin streams, see
917 Documentation/networking/tcp-thin.rst
921 tcp_limit_output_bytes - INTEGER
922 Controls TCP Small Queue limit per tcp socket.
923 TCP bulk sender tends to increase packets in flight until it
924 gets losses notifications. With SNDBUF autotuning, this can
925 result in a large amount of packets queued on the local machine
926 (e.g.: qdiscs, CPU backlog, or device) hurting latency of other
927 flows, for typical pfifo_fast qdiscs. tcp_limit_output_bytes
928 limits the number of bytes on qdisc or device to reduce artificial
929 RTT/cwnd and reduce bufferbloat.
931 Default: 1048576 (16 * 65536)
933 tcp_challenge_ack_limit - INTEGER
934 Limits number of Challenge ACK sent per second, as recommended
935 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
938 tcp_rx_skb_cache - BOOLEAN
939 Controls a per TCP socket cache of one skb, that might help
940 performance of some workloads. This might be dangerous
941 on systems with a lot of TCP sockets, since it increases
944 Default: 0 (disabled)
949 udp_l3mdev_accept - BOOLEAN
950 Enabling this option allows a "global" bound socket to work
951 across L3 master domains (e.g., VRFs) with packets capable of
952 being received regardless of the L3 domain in which they
953 originated. Only valid when the kernel was compiled with
954 CONFIG_NET_L3_MASTER_DEV.
956 Default: 0 (disabled)
958 udp_mem - vector of 3 INTEGERs: min, pressure, max
959 Number of pages allowed for queueing by all UDP sockets.
961 min: Below this number of pages UDP is not bothered about its
962 memory appetite. When amount of memory allocated by UDP exceeds
963 this number, UDP starts to moderate memory usage.
965 pressure: This value was introduced to follow format of tcp_mem.
967 max: Number of pages allowed for queueing by all UDP sockets.
969 Default is calculated at boot time from amount of available memory.
971 udp_rmem_min - INTEGER
972 Minimal size of receive buffer used by UDP sockets in moderation.
973 Each UDP socket is able to use the size for receiving data, even if
974 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
978 udp_wmem_min - INTEGER
979 Minimal size of send buffer used by UDP sockets in moderation.
980 Each UDP socket is able to use the size for sending data, even if
981 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
988 raw_l3mdev_accept - BOOLEAN
989 Enabling this option allows a "global" bound socket to work
990 across L3 master domains (e.g., VRFs) with packets capable of
991 being received regardless of the L3 domain in which they
992 originated. Only valid when the kernel was compiled with
993 CONFIG_NET_L3_MASTER_DEV.
1000 cipso_cache_enable - BOOLEAN
1001 If set, enable additions to and lookups from the CIPSO label mapping
1002 cache. If unset, additions are ignored and lookups always result in a
1003 miss. However, regardless of the setting the cache is still
1004 invalidated when required when means you can safely toggle this on and
1005 off and the cache will always be "safe".
1009 cipso_cache_bucket_size - INTEGER
1010 The CIPSO label cache consists of a fixed size hash table with each
1011 hash bucket containing a number of cache entries. This variable limits
1012 the number of entries in each hash bucket; the larger the value the
1013 more CIPSO label mappings that can be cached. When the number of
1014 entries in a given hash bucket reaches this limit adding new entries
1015 causes the oldest entry in the bucket to be removed to make room.
1019 cipso_rbm_optfmt - BOOLEAN
1020 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
1021 the CIPSO draft specification (see Documentation/netlabel for details).
1022 This means that when set the CIPSO tag will be padded with empty
1023 categories in order to make the packet data 32-bit aligned.
1027 cipso_rbm_structvalid - BOOLEAN
1028 If set, do a very strict check of the CIPSO option when
1029 ip_options_compile() is called. If unset, relax the checks done during
1030 ip_options_compile(). Either way is "safe" as errors are caught else
1031 where in the CIPSO processing code but setting this to 0 (False) should
1032 result in less work (i.e. it should be faster) but could cause problems
1033 with other implementations that require strict checking.
1040 ip_local_port_range - 2 INTEGERS
1041 Defines the local port range that is used by TCP and UDP to
1042 choose the local port. The first number is the first, the
1043 second the last local port number.
1044 If possible, it is better these numbers have different parity
1045 (one even and one odd value).
1046 Must be greater than or equal to ip_unprivileged_port_start.
1047 The default values are 32768 and 60999 respectively.
1049 ip_local_reserved_ports - list of comma separated ranges
1050 Specify the ports which are reserved for known third-party
1051 applications. These ports will not be used by automatic port
1052 assignments (e.g. when calling connect() or bind() with port
1053 number 0). Explicit port allocation behavior is unchanged.
1055 The format used for both input and output is a comma separated
1056 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
1057 10). Writing to the file will clear all previously reserved
1058 ports and update the current list with the one given in the
1061 Note that ip_local_port_range and ip_local_reserved_ports
1062 settings are independent and both are considered by the kernel
1063 when determining which ports are available for automatic port
1066 You can reserve ports which are not in the current
1067 ip_local_port_range, e.g.::
1069 $ cat /proc/sys/net/ipv4/ip_local_port_range
1071 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
1074 although this is redundant. However such a setting is useful
1075 if later the port range is changed to a value that will
1076 include the reserved ports. Also keep in mind, that overlapping
1077 of these ranges may affect probability of selecting ephemeral
1078 ports which are right after block of reserved ports.
1082 ip_unprivileged_port_start - INTEGER
1083 This is a per-namespace sysctl. It defines the first
1084 unprivileged port in the network namespace. Privileged ports
1085 require root or CAP_NET_BIND_SERVICE in order to bind to them.
1086 To disable all privileged ports, set this to 0. They must not
1087 overlap with the ip_local_port_range.
1091 ip_nonlocal_bind - BOOLEAN
1092 If set, allows processes to bind() to non-local IP addresses,
1093 which can be quite useful - but may break some applications.
1097 ip_autobind_reuse - BOOLEAN
1098 By default, bind() does not select the ports automatically even if
1099 the new socket and all sockets bound to the port have SO_REUSEADDR.
1100 ip_autobind_reuse allows bind() to reuse the port and this is useful
1101 when you use bind()+connect(), but may break some applications.
1102 The preferred solution is to use IP_BIND_ADDRESS_NO_PORT and this
1103 option should only be set by experts.
1106 ip_dynaddr - BOOLEAN
1107 If set non-zero, enables support for dynamic addresses.
1108 If set to a non-zero value larger than 1, a kernel log
1109 message will be printed when dynamic address rewriting
1114 ip_early_demux - BOOLEAN
1115 Optimize input packet processing down to one demux for
1116 certain kinds of local sockets. Currently we only do this
1117 for established TCP and connected UDP sockets.
1119 It may add an additional cost for pure routing workloads that
1120 reduces overall throughput, in such case you should disable it.
1124 ping_group_range - 2 INTEGERS
1125 Restrict ICMP_PROTO datagram sockets to users in the group range.
1126 The default is "1 0", meaning, that nobody (not even root) may
1127 create ping sockets. Setting it to "100 100" would grant permissions
1128 to the single group. "0 4294967295" would enable it for the world, "100
1129 4294967295" would enable it for the users, but not daemons.
1131 tcp_early_demux - BOOLEAN
1132 Enable early demux for established TCP sockets.
1136 udp_early_demux - BOOLEAN
1137 Enable early demux for connected UDP sockets. Disable this if
1138 your system could experience more unconnected load.
1142 icmp_echo_ignore_all - BOOLEAN
1143 If set non-zero, then the kernel will ignore all ICMP ECHO
1144 requests sent to it.
1148 icmp_echo_enable_probe - BOOLEAN
1149 If set to one, then the kernel will respond to RFC 8335 PROBE
1150 requests sent to it.
1154 icmp_echo_ignore_broadcasts - BOOLEAN
1155 If set non-zero, then the kernel will ignore all ICMP ECHO and
1156 TIMESTAMP requests sent to it via broadcast/multicast.
1160 icmp_ratelimit - INTEGER
1161 Limit the maximal rates for sending ICMP packets whose type matches
1162 icmp_ratemask (see below) to specific targets.
1163 0 to disable any limiting,
1164 otherwise the minimal space between responses in milliseconds.
1165 Note that another sysctl, icmp_msgs_per_sec limits the number
1166 of ICMP packets sent on all targets.
1170 icmp_msgs_per_sec - INTEGER
1171 Limit maximal number of ICMP packets sent per second from this host.
1172 Only messages whose type matches icmp_ratemask (see below) are
1173 controlled by this limit. For security reasons, the precise count
1174 of messages per second is randomized.
1178 icmp_msgs_burst - INTEGER
1179 icmp_msgs_per_sec controls number of ICMP packets sent per second,
1180 while icmp_msgs_burst controls the burst size of these packets.
1181 For security reasons, the precise burst size is randomized.
1185 icmp_ratemask - INTEGER
1186 Mask made of ICMP types for which rates are being limited.
1188 Significant bits: IHGFEDCBA9876543210
1190 Default mask: 0000001100000011000 (6168)
1192 Bit definitions (see include/linux/icmp.h):
1194 = =========================
1196 3 Destination Unreachable [1]_
1197 4 Source Quench [1]_
1200 B Time Exceeded [1]_
1201 C Parameter Problem [1]_
1206 H Address Mask Request
1207 I Address Mask Reply
1208 = =========================
1210 .. [1] These are rate limited by default (see default mask above)
1212 icmp_ignore_bogus_error_responses - BOOLEAN
1213 Some routers violate RFC1122 by sending bogus responses to broadcast
1214 frames. Such violations are normally logged via a kernel warning.
1215 If this is set to TRUE, the kernel will not give such warnings, which
1216 will avoid log file clutter.
1220 icmp_errors_use_inbound_ifaddr - BOOLEAN
1222 If zero, icmp error messages are sent with the primary address of
1223 the exiting interface.
1225 If non-zero, the message will be sent with the primary address of
1226 the interface that received the packet that caused the icmp error.
1227 This is the behaviour many network administrators will expect from
1228 a router. And it can make debugging complicated network layouts
1231 Note that if no primary address exists for the interface selected,
1232 then the primary address of the first non-loopback interface that
1233 has one will be used regardless of this setting.
1237 igmp_max_memberships - INTEGER
1238 Change the maximum number of multicast groups we can subscribe to.
1241 Theoretical maximum value is bounded by having to send a membership
1242 report in a single datagram (i.e. the report can't span multiple
1243 datagrams, or risk confusing the switch and leaving groups you don't
1246 The number of supported groups 'M' is bounded by the number of group
1247 report entries you can fit into a single datagram of 65535 bytes.
1249 M = 65536-sizeof (ip header)/(sizeof(Group record))
1251 Group records are variable length, with a minimum of 12 bytes.
1252 So net.ipv4.igmp_max_memberships should not be set higher than:
1254 (65536-24) / 12 = 5459
1256 The value 5459 assumes no IP header options, so in practice
1257 this number may be lower.
1259 igmp_max_msf - INTEGER
1260 Maximum number of addresses allowed in the source filter list for a
1266 Controls the IGMP query robustness variable (see RFC2236 8.1).
1268 Default: 2 (as specified by RFC2236 8.1)
1270 Minimum: 1 (as specified by RFC6636 4.5)
1272 force_igmp_version - INTEGER
1273 - 0 - (default) No enforcement of a IGMP version, IGMPv1/v2 fallback
1274 allowed. Will back to IGMPv3 mode again if all IGMPv1/v2 Querier
1275 Present timer expires.
1276 - 1 - Enforce to use IGMP version 1. Will also reply IGMPv1 report if
1277 receive IGMPv2/v3 query.
1278 - 2 - Enforce to use IGMP version 2. Will fallback to IGMPv1 if receive
1279 IGMPv1 query message. Will reply report if receive IGMPv3 query.
1280 - 3 - Enforce to use IGMP version 3. The same react with default 0.
1284 this is not the same with force_mld_version because IGMPv3 RFC3376
1285 Security Considerations does not have clear description that we could
1286 ignore other version messages completely as MLDv2 RFC3810. So make
1287 this value as default 0 is recommended.
1289 ``conf/interface/*``
1290 changes special settings per interface (where
1291 interface" is the name of your network interface)
1294 is special, changes the settings for all interfaces
1296 log_martians - BOOLEAN
1297 Log packets with impossible addresses to kernel log.
1298 log_martians for the interface will be enabled if at least one of
1299 conf/{all,interface}/log_martians is set to TRUE,
1300 it will be disabled otherwise
1302 accept_redirects - BOOLEAN
1303 Accept ICMP redirect messages.
1304 accept_redirects for the interface will be enabled if:
1306 - both conf/{all,interface}/accept_redirects are TRUE in the case
1307 forwarding for the interface is enabled
1311 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
1312 case forwarding for the interface is disabled
1314 accept_redirects for the interface will be disabled otherwise
1321 forwarding - BOOLEAN
1322 Enable IP forwarding on this interface. This controls whether packets
1323 received _on_ this interface can be forwarded.
1325 mc_forwarding - BOOLEAN
1326 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
1327 and a multicast routing daemon is required.
1328 conf/all/mc_forwarding must also be set to TRUE to enable multicast
1329 routing for the interface
1332 Integer value used to differentiate the devices by the medium they
1333 are attached to. Two devices can have different id values when
1334 the broadcast packets are received only on one of them.
1335 The default value 0 means that the device is the only interface
1336 to its medium, value of -1 means that medium is not known.
1338 Currently, it is used to change the proxy_arp behavior:
1339 the proxy_arp feature is enabled for packets forwarded between
1340 two devices attached to different media.
1345 proxy_arp for the interface will be enabled if at least one of
1346 conf/{all,interface}/proxy_arp is set to TRUE,
1347 it will be disabled otherwise
1349 proxy_arp_pvlan - BOOLEAN
1350 Private VLAN proxy arp.
1352 Basically allow proxy arp replies back to the same interface
1353 (from which the ARP request/solicitation was received).
1355 This is done to support (ethernet) switch features, like RFC
1356 3069, where the individual ports are NOT allowed to
1357 communicate with each other, but they are allowed to talk to
1358 the upstream router. As described in RFC 3069, it is possible
1359 to allow these hosts to communicate through the upstream
1360 router by proxy_arp'ing. Don't need to be used together with
1363 This technology is known by different names:
1365 In RFC 3069 it is called VLAN Aggregation.
1366 Cisco and Allied Telesyn call it Private VLAN.
1367 Hewlett-Packard call it Source-Port filtering or port-isolation.
1368 Ericsson call it MAC-Forced Forwarding (RFC Draft).
1370 shared_media - BOOLEAN
1371 Send(router) or accept(host) RFC1620 shared media redirects.
1372 Overrides secure_redirects.
1374 shared_media for the interface will be enabled if at least one of
1375 conf/{all,interface}/shared_media is set to TRUE,
1376 it will be disabled otherwise
1380 secure_redirects - BOOLEAN
1381 Accept ICMP redirect messages only to gateways listed in the
1382 interface's current gateway list. Even if disabled, RFC1122 redirect
1385 Overridden by shared_media.
1387 secure_redirects for the interface will be enabled if at least one of
1388 conf/{all,interface}/secure_redirects is set to TRUE,
1389 it will be disabled otherwise
1393 send_redirects - BOOLEAN
1394 Send redirects, if router.
1396 send_redirects for the interface will be enabled if at least one of
1397 conf/{all,interface}/send_redirects is set to TRUE,
1398 it will be disabled otherwise
1402 bootp_relay - BOOLEAN
1403 Accept packets with source address 0.b.c.d destined
1404 not to this host as local ones. It is supposed, that
1405 BOOTP relay daemon will catch and forward such packets.
1406 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
1411 Not Implemented Yet.
1413 accept_source_route - BOOLEAN
1414 Accept packets with SRR option.
1415 conf/all/accept_source_route must also be set to TRUE to accept packets
1416 with SRR option on the interface
1423 accept_local - BOOLEAN
1424 Accept packets with local source addresses. In combination with
1425 suitable routing, this can be used to direct packets between two
1426 local interfaces over the wire and have them accepted properly.
1429 route_localnet - BOOLEAN
1430 Do not consider loopback addresses as martian source or destination
1431 while routing. This enables the use of 127/8 for local routing purposes.
1436 - 0 - No source validation.
1437 - 1 - Strict mode as defined in RFC3704 Strict Reverse Path
1438 Each incoming packet is tested against the FIB and if the interface
1439 is not the best reverse path the packet check will fail.
1440 By default failed packets are discarded.
1441 - 2 - Loose mode as defined in RFC3704 Loose Reverse Path
1442 Each incoming packet's source address is also tested against the FIB
1443 and if the source address is not reachable via any interface
1444 the packet check will fail.
1446 Current recommended practice in RFC3704 is to enable strict mode
1447 to prevent IP spoofing from DDos attacks. If using asymmetric routing
1448 or other complicated routing, then loose mode is recommended.
1450 The max value from conf/{all,interface}/rp_filter is used
1451 when doing source validation on the {interface}.
1453 Default value is 0. Note that some distributions enable it
1456 src_valid_mark - BOOLEAN
1457 - 0 - The fwmark of the packet is not included in reverse path
1458 route lookup. This allows for asymmetric routing configurations
1459 utilizing the fwmark in only one direction, e.g., transparent
1462 - 1 - The fwmark of the packet is included in reverse path route
1463 lookup. This permits rp_filter to function when the fwmark is
1464 used for routing traffic in both directions.
1466 This setting also affects the utilization of fmwark when
1467 performing source address selection for ICMP replies, or
1468 determining addresses stored for the IPOPT_TS_TSANDADDR and
1469 IPOPT_RR IP options.
1471 The max value from conf/{all,interface}/src_valid_mark is used.
1475 arp_filter - BOOLEAN
1476 - 1 - Allows you to have multiple network interfaces on the same
1477 subnet, and have the ARPs for each interface be answered
1478 based on whether or not the kernel would route a packet from
1479 the ARP'd IP out that interface (therefore you must use source
1480 based routing for this to work). In other words it allows control
1481 of which cards (usually 1) will respond to an arp request.
1483 - 0 - (default) The kernel can respond to arp requests with addresses
1484 from other interfaces. This may seem wrong but it usually makes
1485 sense, because it increases the chance of successful communication.
1486 IP addresses are owned by the complete host on Linux, not by
1487 particular interfaces. Only for more complex setups like load-
1488 balancing, does this behaviour cause problems.
1490 arp_filter for the interface will be enabled if at least one of
1491 conf/{all,interface}/arp_filter is set to TRUE,
1492 it will be disabled otherwise
1494 arp_announce - INTEGER
1495 Define different restriction levels for announcing the local
1496 source IP address from IP packets in ARP requests sent on
1499 - 0 - (default) Use any local address, configured on any interface
1500 - 1 - Try to avoid local addresses that are not in the target's
1501 subnet for this interface. This mode is useful when target
1502 hosts reachable via this interface require the source IP
1503 address in ARP requests to be part of their logical network
1504 configured on the receiving interface. When we generate the
1505 request we will check all our subnets that include the
1506 target IP and will preserve the source address if it is from
1507 such subnet. If there is no such subnet we select source
1508 address according to the rules for level 2.
1509 - 2 - Always use the best local address for this target.
1510 In this mode we ignore the source address in the IP packet
1511 and try to select local address that we prefer for talks with
1512 the target host. Such local address is selected by looking
1513 for primary IP addresses on all our subnets on the outgoing
1514 interface that include the target IP address. If no suitable
1515 local address is found we select the first local address
1516 we have on the outgoing interface or on all other interfaces,
1517 with the hope we will receive reply for our request and
1518 even sometimes no matter the source IP address we announce.
1520 The max value from conf/{all,interface}/arp_announce is used.
1522 Increasing the restriction level gives more chance for
1523 receiving answer from the resolved target while decreasing
1524 the level announces more valid sender's information.
1526 arp_ignore - INTEGER
1527 Define different modes for sending replies in response to
1528 received ARP requests that resolve local target IP addresses:
1530 - 0 - (default): reply for any local target IP address, configured
1532 - 1 - reply only if the target IP address is local address
1533 configured on the incoming interface
1534 - 2 - reply only if the target IP address is local address
1535 configured on the incoming interface and both with the
1536 sender's IP address are part from same subnet on this interface
1537 - 3 - do not reply for local addresses configured with scope host,
1538 only resolutions for global and link addresses are replied
1540 - 8 - do not reply for all local addresses
1542 The max value from conf/{all,interface}/arp_ignore is used
1543 when ARP request is received on the {interface}
1545 arp_notify - BOOLEAN
1546 Define mode for notification of address and device changes.
1548 == ==========================================================
1549 0 (default): do nothing
1550 1 Generate gratuitous arp requests when device is brought up
1551 or hardware address changes.
1552 == ==========================================================
1554 arp_accept - BOOLEAN
1555 Define behavior for gratuitous ARP frames who's IP is not
1556 already present in the ARP table:
1558 - 0 - don't create new entries in the ARP table
1559 - 1 - create new entries in the ARP table
1561 Both replies and requests type gratuitous arp will trigger the
1562 ARP table to be updated, if this setting is on.
1564 If the ARP table already contains the IP address of the
1565 gratuitous arp frame, the arp table will be updated regardless
1566 if this setting is on or off.
1568 mcast_solicit - INTEGER
1569 The maximum number of multicast probes in INCOMPLETE state,
1570 when the associated hardware address is unknown. Defaults
1573 ucast_solicit - INTEGER
1574 The maximum number of unicast probes in PROBE state, when
1575 the hardware address is being reconfirmed. Defaults to 3.
1577 app_solicit - INTEGER
1578 The maximum number of probes to send to the user space ARP daemon
1579 via netlink before dropping back to multicast probes (see
1580 mcast_resolicit). Defaults to 0.
1582 mcast_resolicit - INTEGER
1583 The maximum number of multicast probes after unicast and
1584 app probes in PROBE state. Defaults to 0.
1586 disable_policy - BOOLEAN
1587 Disable IPSEC policy (SPD) for this interface
1589 disable_xfrm - BOOLEAN
1590 Disable IPSEC encryption on this interface, whatever the policy
1592 igmpv2_unsolicited_report_interval - INTEGER
1593 The interval in milliseconds in which the next unsolicited
1594 IGMPv1 or IGMPv2 report retransmit will take place.
1596 Default: 10000 (10 seconds)
1598 igmpv3_unsolicited_report_interval - INTEGER
1599 The interval in milliseconds in which the next unsolicited
1600 IGMPv3 report retransmit will take place.
1602 Default: 1000 (1 seconds)
1604 ignore_routes_with_linkdown - BOOLEAN
1605 Ignore routes whose link is down when performing a FIB lookup.
1607 promote_secondaries - BOOLEAN
1608 When a primary IP address is removed from this interface
1609 promote a corresponding secondary IP address instead of
1610 removing all the corresponding secondary IP addresses.
1612 drop_unicast_in_l2_multicast - BOOLEAN
1613 Drop any unicast IP packets that are received in link-layer
1614 multicast (or broadcast) frames.
1616 This behavior (for multicast) is actually a SHOULD in RFC
1617 1122, but is disabled by default for compatibility reasons.
1621 drop_gratuitous_arp - BOOLEAN
1622 Drop all gratuitous ARP frames, for example if there's a known
1623 good ARP proxy on the network and such frames need not be used
1624 (or in the case of 802.11, must not be used to prevent attacks.)
1630 Allows you to write a number, which can be used as required.
1634 xfrm4_gc_thresh - INTEGER
1635 (Obsolete since linux-4.14)
1636 The threshold at which we will start garbage collecting for IPv4
1637 destination cache entries. At twice this value the system will
1638 refuse new allocations.
1640 igmp_link_local_mcast_reports - BOOLEAN
1641 Enable IGMP reports for link local multicast groups in the
1647 kuznet@ms2.inr.ac.ru
1654 delon.nicolas@wanadoo.fr
1659 /proc/sys/net/ipv6/* Variables
1660 ==============================
1662 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1663 apply to IPv6 [XXX?].
1665 bindv6only - BOOLEAN
1666 Default value for IPV6_V6ONLY socket option,
1667 which restricts use of the IPv6 socket to IPv6 communication
1670 - TRUE: disable IPv4-mapped address feature
1671 - FALSE: enable IPv4-mapped address feature
1673 Default: FALSE (as specified in RFC3493)
1675 flowlabel_consistency - BOOLEAN
1676 Protect the consistency (and unicity) of flow label.
1677 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1685 auto_flowlabels - INTEGER
1686 Automatically generate flow labels based on a flow hash of the
1687 packet. This allows intermediate devices, such as routers, to
1688 identify packet flows for mechanisms like Equal Cost Multipath
1689 Routing (see RFC 6438).
1691 = ===========================================================
1692 0 automatic flow labels are completely disabled
1693 1 automatic flow labels are enabled by default, they can be
1694 disabled on a per socket basis using the IPV6_AUTOFLOWLABEL
1696 2 automatic flow labels are allowed, they may be enabled on a
1697 per socket basis using the IPV6_AUTOFLOWLABEL socket option
1698 3 automatic flow labels are enabled and enforced, they cannot
1699 be disabled by the socket option
1700 = ===========================================================
1704 flowlabel_state_ranges - BOOLEAN
1705 Split the flow label number space into two ranges. 0-0x7FFFF is
1706 reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF
1707 is reserved for stateless flow labels as described in RFC6437.
1714 flowlabel_reflect - INTEGER
1715 Control flow label reflection. Needed for Path MTU
1716 Discovery to work with Equal Cost Multipath Routing in anycast
1717 environments. See RFC 7690 and:
1718 https://tools.ietf.org/html/draft-wang-6man-flow-label-reflection-01
1722 - 1: enabled for established flows
1724 Note that this prevents automatic flowlabel changes, as done
1725 in "tcp: change IPv6 flow-label upon receiving spurious retransmission"
1726 and "tcp: Change txhash on every SYN and RTO retransmit"
1728 - 2: enabled for TCP RESET packets (no active listener)
1729 If set, a RST packet sent in response to a SYN packet on a closed
1730 port will reflect the incoming flow label.
1732 - 4: enabled for ICMPv6 echo reply messages.
1736 fib_multipath_hash_policy - INTEGER
1737 Controls which hash policy to use for multipath routes.
1739 Default: 0 (Layer 3)
1743 - 0 - Layer 3 (source and destination addresses plus flow label)
1744 - 1 - Layer 4 (standard 5-tuple)
1745 - 2 - Layer 3 or inner Layer 3 if present
1747 anycast_src_echo_reply - BOOLEAN
1748 Controls the use of anycast addresses as source addresses for ICMPv6
1756 idgen_delay - INTEGER
1757 Controls the delay in seconds after which time to retry
1758 privacy stable address generation if a DAD conflict is
1761 Default: 1 (as specified in RFC7217)
1763 idgen_retries - INTEGER
1764 Controls the number of retries to generate a stable privacy
1765 address if a DAD conflict is detected.
1767 Default: 3 (as specified in RFC7217)
1770 Controls the MLD query robustness variable (see RFC3810 9.1).
1772 Default: 2 (as specified by RFC3810 9.1)
1774 Minimum: 1 (as specified by RFC6636 4.5)
1776 max_dst_opts_number - INTEGER
1777 Maximum number of non-padding TLVs allowed in a Destination
1778 options extension header. If this value is less than zero
1779 then unknown options are disallowed and the number of known
1780 TLVs allowed is the absolute value of this number.
1784 max_hbh_opts_number - INTEGER
1785 Maximum number of non-padding TLVs allowed in a Hop-by-Hop
1786 options extension header. If this value is less than zero
1787 then unknown options are disallowed and the number of known
1788 TLVs allowed is the absolute value of this number.
1792 max_dst_opts_length - INTEGER
1793 Maximum length allowed for a Destination options extension
1796 Default: INT_MAX (unlimited)
1798 max_hbh_length - INTEGER
1799 Maximum length allowed for a Hop-by-Hop options extension
1802 Default: INT_MAX (unlimited)
1804 skip_notify_on_dev_down - BOOLEAN
1805 Controls whether an RTM_DELROUTE message is generated for routes
1806 removed when a device is taken down or deleted. IPv4 does not
1807 generate this message; IPv6 does by default. Setting this sysctl
1808 to true skips the message, making IPv4 and IPv6 on par in relying
1809 on userspace caches to track link events and evict routes.
1811 Default: false (generate message)
1813 nexthop_compat_mode - BOOLEAN
1814 New nexthop API provides a means for managing nexthops independent of
1815 prefixes. Backwards compatibilty with old route format is enabled by
1816 default which means route dumps and notifications contain the new
1817 nexthop attribute but also the full, expanded nexthop definition.
1818 Further, updates or deletes of a nexthop configuration generate route
1819 notifications for each fib entry using the nexthop. Once a system
1820 understands the new API, this sysctl can be disabled to achieve full
1821 performance benefits of the new API by disabling the nexthop expansion
1822 and extraneous notifications.
1823 Default: true (backward compat mode)
1825 fib_notify_on_flag_change - INTEGER
1826 Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/
1827 RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed.
1829 After installing a route to the kernel, user space receives an
1830 acknowledgment, which means the route was installed in the kernel,
1831 but not necessarily in hardware.
1832 It is also possible for a route already installed in hardware to change
1833 its action and therefore its flags. For example, a host route that is
1834 trapping packets can be "promoted" to perform decapsulation following
1835 the installation of an IPinIP/VXLAN tunnel.
1836 The notifications will indicate to user-space the state of the route.
1838 Default: 0 (Do not emit notifications.)
1842 - 0 - Do not emit notifications.
1843 - 1 - Emit notifications.
1844 - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change.
1848 ip6frag_high_thresh - INTEGER
1849 Maximum memory used to reassemble IPv6 fragments. When
1850 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1851 the fragment handler will toss packets until ip6frag_low_thresh
1854 ip6frag_low_thresh - INTEGER
1855 See ip6frag_high_thresh
1857 ip6frag_time - INTEGER
1858 Time in seconds to keep an IPv6 fragment in memory.
1861 Change the interface-specific default settings.
1863 These settings would be used during creating new interfaces.
1867 Change all the interface-specific settings.
1869 [XXX: Other special features than forwarding?]
1871 conf/all/disable_ipv6 - BOOLEAN
1872 Changing this value is same as changing ``conf/default/disable_ipv6``
1873 setting and also all per-interface ``disable_ipv6`` settings to the same
1876 Reading this value does not have any particular meaning. It does not say
1877 whether IPv6 support is enabled or disabled. Returned value can be 1
1878 also in the case when some interface has ``disable_ipv6`` set to 0 and
1879 has configured IPv6 addresses.
1881 conf/all/forwarding - BOOLEAN
1882 Enable global IPv6 forwarding between all interfaces.
1884 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1885 to control which interfaces may forward packets and which not.
1887 This also sets all interfaces' Host/Router setting
1888 'forwarding' to the specified value. See below for details.
1890 This referred to as global forwarding.
1895 fwmark_reflect - BOOLEAN
1896 Controls the fwmark of kernel-generated IPv6 reply packets that are not
1897 associated with a socket for example, TCP RSTs or ICMPv6 echo replies).
1898 If unset, these packets have a fwmark of zero. If set, they have the
1899 fwmark of the packet they are replying to.
1903 ``conf/interface/*``:
1904 Change special settings per interface.
1906 The functional behaviour for certain settings is different
1907 depending on whether local forwarding is enabled or not.
1910 Accept Router Advertisements; autoconfigure using them.
1912 It also determines whether or not to transmit Router
1913 Solicitations. If and only if the functional setting is to
1914 accept Router Advertisements, Router Solicitations will be
1917 Possible values are:
1919 == ===========================================================
1920 0 Do not accept Router Advertisements.
1921 1 Accept Router Advertisements if forwarding is disabled.
1922 2 Overrule forwarding behaviour. Accept Router Advertisements
1923 even if forwarding is enabled.
1924 == ===========================================================
1928 - enabled if local forwarding is disabled.
1929 - disabled if local forwarding is enabled.
1931 accept_ra_defrtr - BOOLEAN
1932 Learn default router in Router Advertisement.
1936 - enabled if accept_ra is enabled.
1937 - disabled if accept_ra is disabled.
1939 ra_defrtr_metric - UNSIGNED INTEGER
1940 Route metric for default route learned in Router Advertisement. This value
1941 will be assigned as metric for the default route learned via IPv6 Router
1942 Advertisement. Takes affect only if accept_ra_defrtr is enabled.
1947 Default: IP6_RT_PRIO_USER i.e. 1024.
1949 accept_ra_from_local - BOOLEAN
1950 Accept RA with source-address that is found on local machine
1951 if the RA is otherwise proper and able to be accepted.
1953 Default is to NOT accept these as it may be an un-intended
1958 - enabled if accept_ra_from_local is enabled
1959 on a specific interface.
1960 - disabled if accept_ra_from_local is disabled
1961 on a specific interface.
1963 accept_ra_min_hop_limit - INTEGER
1964 Minimum hop limit Information in Router Advertisement.
1966 Hop limit Information in Router Advertisement less than this
1967 variable shall be ignored.
1971 accept_ra_pinfo - BOOLEAN
1972 Learn Prefix Information in Router Advertisement.
1976 - enabled if accept_ra is enabled.
1977 - disabled if accept_ra is disabled.
1979 accept_ra_rt_info_min_plen - INTEGER
1980 Minimum prefix length of Route Information in RA.
1982 Route Information w/ prefix smaller than this variable shall
1987 * 0 if accept_ra_rtr_pref is enabled.
1988 * -1 if accept_ra_rtr_pref is disabled.
1990 accept_ra_rt_info_max_plen - INTEGER
1991 Maximum prefix length of Route Information in RA.
1993 Route Information w/ prefix larger than this variable shall
1998 * 0 if accept_ra_rtr_pref is enabled.
1999 * -1 if accept_ra_rtr_pref is disabled.
2001 accept_ra_rtr_pref - BOOLEAN
2002 Accept Router Preference in RA.
2006 - enabled if accept_ra is enabled.
2007 - disabled if accept_ra is disabled.
2009 accept_ra_mtu - BOOLEAN
2010 Apply the MTU value specified in RA option 5 (RFC4861). If
2011 disabled, the MTU specified in the RA will be ignored.
2015 - enabled if accept_ra is enabled.
2016 - disabled if accept_ra is disabled.
2018 accept_redirects - BOOLEAN
2023 - enabled if local forwarding is disabled.
2024 - disabled if local forwarding is enabled.
2026 accept_source_route - INTEGER
2027 Accept source routing (routing extension header).
2029 - >= 0: Accept only routing header type 2.
2030 - < 0: Do not accept routing header.
2035 Autoconfigure addresses using Prefix Information in Router
2040 - enabled if accept_ra_pinfo is enabled.
2041 - disabled if accept_ra_pinfo is disabled.
2043 dad_transmits - INTEGER
2044 The amount of Duplicate Address Detection probes to send.
2048 forwarding - INTEGER
2049 Configure interface-specific Host/Router behaviour.
2053 It is recommended to have the same setting on all
2054 interfaces; mixed router/host scenarios are rather uncommon.
2056 Possible values are:
2058 - 0 Forwarding disabled
2059 - 1 Forwarding enabled
2063 By default, Host behaviour is assumed. This means:
2065 1. IsRouter flag is not set in Neighbour Advertisements.
2066 2. If accept_ra is TRUE (default), transmit Router
2068 3. If accept_ra is TRUE (default), accept Router
2069 Advertisements (and do autoconfiguration).
2070 4. If accept_redirects is TRUE (default), accept Redirects.
2074 If local forwarding is enabled, Router behaviour is assumed.
2075 This means exactly the reverse from the above:
2077 1. IsRouter flag is set in Neighbour Advertisements.
2078 2. Router Solicitations are not sent unless accept_ra is 2.
2079 3. Router Advertisements are ignored unless accept_ra is 2.
2080 4. Redirects are ignored.
2082 Default: 0 (disabled) if global forwarding is disabled (default),
2083 otherwise 1 (enabled).
2086 Default Hop Limit to set.
2091 Default Maximum Transfer Unit
2093 Default: 1280 (IPv6 required minimum)
2095 ip_nonlocal_bind - BOOLEAN
2096 If set, allows processes to bind() to non-local IPv6 addresses,
2097 which can be quite useful - but may break some applications.
2101 router_probe_interval - INTEGER
2102 Minimum interval (in seconds) between Router Probing described
2107 router_solicitation_delay - INTEGER
2108 Number of seconds to wait after interface is brought up
2109 before sending Router Solicitations.
2113 router_solicitation_interval - INTEGER
2114 Number of seconds to wait between Router Solicitations.
2118 router_solicitations - INTEGER
2119 Number of Router Solicitations to send until assuming no
2120 routers are present.
2124 use_oif_addrs_only - BOOLEAN
2125 When enabled, the candidate source addresses for destinations
2126 routed via this interface are restricted to the set of addresses
2127 configured on this interface (vis. RFC 6724, section 4).
2131 use_tempaddr - INTEGER
2132 Preference for Privacy Extensions (RFC3041).
2134 * <= 0 : disable Privacy Extensions
2135 * == 1 : enable Privacy Extensions, but prefer public
2136 addresses over temporary addresses.
2137 * > 1 : enable Privacy Extensions and prefer temporary
2138 addresses over public addresses.
2142 * 0 (for most devices)
2143 * -1 (for point-to-point devices and loopback devices)
2145 temp_valid_lft - INTEGER
2146 valid lifetime (in seconds) for temporary addresses.
2148 Default: 172800 (2 days)
2150 temp_prefered_lft - INTEGER
2151 Preferred lifetime (in seconds) for temporary addresses.
2153 Default: 86400 (1 day)
2155 keep_addr_on_down - INTEGER
2156 Keep all IPv6 addresses on an interface down event. If set static
2157 global addresses with no expiration time are not flushed.
2160 * 0 : system default
2163 Default: 0 (addresses are removed)
2165 max_desync_factor - INTEGER
2166 Maximum value for DESYNC_FACTOR, which is a random value
2167 that ensures that clients don't synchronize with each
2168 other and generate new addresses at exactly the same time.
2169 value is in seconds.
2173 regen_max_retry - INTEGER
2174 Number of attempts before give up attempting to generate
2175 valid temporary addresses.
2179 max_addresses - INTEGER
2180 Maximum number of autoconfigured addresses per interface. Setting
2181 to zero disables the limitation. It is not recommended to set this
2182 value too large (or to zero) because it would be an easy way to
2183 crash the kernel by allowing too many addresses to be created.
2187 disable_ipv6 - BOOLEAN
2188 Disable IPv6 operation. If accept_dad is set to 2, this value
2189 will be dynamically set to TRUE if DAD fails for the link-local
2192 Default: FALSE (enable IPv6 operation)
2194 When this value is changed from 1 to 0 (IPv6 is being enabled),
2195 it will dynamically create a link-local address on the given
2196 interface and start Duplicate Address Detection, if necessary.
2198 When this value is changed from 0 to 1 (IPv6 is being disabled),
2199 it will dynamically delete all addresses and routes on the given
2200 interface. From now on it will not possible to add addresses/routes
2201 to the selected interface.
2203 accept_dad - INTEGER
2204 Whether to accept DAD (Duplicate Address Detection).
2206 == ==============================================================
2208 1 Enable DAD (default)
2209 2 Enable DAD, and disable IPv6 operation if MAC-based duplicate
2210 link-local address has been found.
2211 == ==============================================================
2213 DAD operation and mode on a given interface will be selected according
2214 to the maximum value of conf/{all,interface}/accept_dad.
2216 force_tllao - BOOLEAN
2217 Enable sending the target link-layer address option even when
2218 responding to a unicast neighbor solicitation.
2222 Quoting from RFC 2461, section 4.4, Target link-layer address:
2224 "The option MUST be included for multicast solicitations in order to
2225 avoid infinite Neighbor Solicitation "recursion" when the peer node
2226 does not have a cache entry to return a Neighbor Advertisements
2227 message. When responding to unicast solicitations, the option can be
2228 omitted since the sender of the solicitation has the correct link-
2229 layer address; otherwise it would not have be able to send the unicast
2230 solicitation in the first place. However, including the link-layer
2231 address in this case adds little overhead and eliminates a potential
2232 race condition where the sender deletes the cached link-layer address
2233 prior to receiving a response to a previous solicitation."
2235 ndisc_notify - BOOLEAN
2236 Define mode for notification of address and device changes.
2238 * 0 - (default): do nothing
2239 * 1 - Generate unsolicited neighbour advertisements when device is brought
2240 up or hardware address changes.
2242 ndisc_tclass - INTEGER
2243 The IPv6 Traffic Class to use by default when sending IPv6 Neighbor
2244 Discovery (Router Solicitation, Router Advertisement, Neighbor
2245 Solicitation, Neighbor Advertisement, Redirect) messages.
2246 These 8 bits can be interpreted as 6 high order bits holding the DSCP
2247 value and 2 low order bits representing ECN (which you probably want
2252 mldv1_unsolicited_report_interval - INTEGER
2253 The interval in milliseconds in which the next unsolicited
2254 MLDv1 report retransmit will take place.
2256 Default: 10000 (10 seconds)
2258 mldv2_unsolicited_report_interval - INTEGER
2259 The interval in milliseconds in which the next unsolicited
2260 MLDv2 report retransmit will take place.
2262 Default: 1000 (1 second)
2264 force_mld_version - INTEGER
2265 * 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
2266 * 1 - Enforce to use MLD version 1
2267 * 2 - Enforce to use MLD version 2
2269 suppress_frag_ndisc - INTEGER
2270 Control RFC 6980 (Security Implications of IPv6 Fragmentation
2271 with IPv6 Neighbor Discovery) behavior:
2273 * 1 - (default) discard fragmented neighbor discovery packets
2274 * 0 - allow fragmented neighbor discovery packets
2276 optimistic_dad - BOOLEAN
2277 Whether to perform Optimistic Duplicate Address Detection (RFC 4429).
2279 * 0: disabled (default)
2282 Optimistic Duplicate Address Detection for the interface will be enabled
2283 if at least one of conf/{all,interface}/optimistic_dad is set to 1,
2284 it will be disabled otherwise.
2286 use_optimistic - BOOLEAN
2287 If enabled, do not classify optimistic addresses as deprecated during
2288 source address selection. Preferred addresses will still be chosen
2289 before optimistic addresses, subject to other ranking in the source
2290 address selection algorithm.
2292 * 0: disabled (default)
2295 This will be enabled if at least one of
2296 conf/{all,interface}/use_optimistic is set to 1, disabled otherwise.
2298 stable_secret - IPv6 address
2299 This IPv6 address will be used as a secret to generate IPv6
2300 addresses for link-local addresses and autoconfigured
2301 ones. All addresses generated after setting this secret will
2302 be stable privacy ones by default. This can be changed via the
2303 addrgenmode ip-link. conf/default/stable_secret is used as the
2304 secret for the namespace, the interface specific ones can
2305 overwrite that. Writes to conf/all/stable_secret are refused.
2307 It is recommended to generate this secret during installation
2308 of a system and keep it stable after that.
2310 By default the stable secret is unset.
2312 addr_gen_mode - INTEGER
2313 Defines how link-local and autoconf addresses are generated.
2315 = =================================================================
2316 0 generate address based on EUI64 (default)
2317 1 do no generate a link-local address, use EUI64 for addresses
2318 generated from autoconf
2319 2 generate stable privacy addresses, using the secret from
2320 stable_secret (RFC7217)
2321 3 generate stable privacy addresses, using a random secret if unset
2322 = =================================================================
2324 drop_unicast_in_l2_multicast - BOOLEAN
2325 Drop any unicast IPv6 packets that are received in link-layer
2326 multicast (or broadcast) frames.
2328 By default this is turned off.
2330 drop_unsolicited_na - BOOLEAN
2331 Drop all unsolicited neighbor advertisements, for example if there's
2332 a known good NA proxy on the network and such frames need not be used
2333 (or in the case of 802.11, must not be used to prevent attacks.)
2335 By default this is turned off.
2337 enhanced_dad - BOOLEAN
2338 Include a nonce option in the IPv6 neighbor solicitation messages used for
2339 duplicate address detection per RFC7527. A received DAD NS will only signal
2340 a duplicate address if the nonce is different. This avoids any false
2341 detection of duplicates due to loopback of the NS messages that we send.
2342 The nonce option will be sent on an interface unless both of
2343 conf/{all,interface}/enhanced_dad are set to FALSE.
2351 Limit the maximal rates for sending ICMPv6 messages.
2353 0 to disable any limiting,
2354 otherwise the minimal space between responses in milliseconds.
2358 ratemask - list of comma separated ranges
2359 For ICMPv6 message types matching the ranges in the ratemask, limit
2360 the sending of the message according to ratelimit parameter.
2362 The format used for both input and output is a comma separated
2363 list of ranges (e.g. "0-127,129" for ICMPv6 message type 0 to 127 and
2364 129). Writing to the file will clear all previous ranges of ICMPv6
2365 message types and update the current list with the input.
2367 Refer to: https://www.iana.org/assignments/icmpv6-parameters/icmpv6-parameters.xhtml
2368 for numerical values of ICMPv6 message types, e.g. echo request is 128
2369 and echo reply is 129.
2371 Default: 0-1,3-127 (rate limit ICMPv6 errors except Packet Too Big)
2373 echo_ignore_all - BOOLEAN
2374 If set non-zero, then the kernel will ignore all ICMP ECHO
2375 requests sent to it over the IPv6 protocol.
2379 echo_ignore_multicast - BOOLEAN
2380 If set non-zero, then the kernel will ignore all ICMP ECHO
2381 requests sent to it over the IPv6 protocol via multicast.
2385 echo_ignore_anycast - BOOLEAN
2386 If set non-zero, then the kernel will ignore all ICMP ECHO
2387 requests sent to it over the IPv6 protocol destined to anycast address.
2391 xfrm6_gc_thresh - INTEGER
2392 (Obsolete since linux-4.14)
2393 The threshold at which we will start garbage collecting for IPv6
2394 destination cache entries. At twice this value the system will
2395 refuse new allocations.
2399 Pekka Savola <pekkas@netcore.fi>
2400 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
2403 /proc/sys/net/bridge/* Variables:
2404 =================================
2406 bridge-nf-call-arptables - BOOLEAN
2407 - 1 : pass bridged ARP traffic to arptables' FORWARD chain.
2412 bridge-nf-call-iptables - BOOLEAN
2413 - 1 : pass bridged IPv4 traffic to iptables' chains.
2418 bridge-nf-call-ip6tables - BOOLEAN
2419 - 1 : pass bridged IPv6 traffic to ip6tables' chains.
2424 bridge-nf-filter-vlan-tagged - BOOLEAN
2425 - 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
2430 bridge-nf-filter-pppoe-tagged - BOOLEAN
2431 - 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
2436 bridge-nf-pass-vlan-input-dev - BOOLEAN
2437 - 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
2438 interface on the bridge and set the netfilter input device to the
2439 vlan. This allows use of e.g. "iptables -i br0.1" and makes the
2440 REDIRECT target work with vlan-on-top-of-bridge interfaces. When no
2441 matching vlan interface is found, or this switch is off, the input
2442 device is set to the bridge interface.
2444 - 0: disable bridge netfilter vlan interface lookup.
2448 ``proc/sys/net/sctp/*`` Variables:
2449 ==================================
2451 addip_enable - BOOLEAN
2452 Enable or disable extension of Dynamic Address Reconfiguration
2453 (ADD-IP) functionality specified in RFC5061. This extension provides
2454 the ability to dynamically add and remove new addresses for the SCTP
2457 1: Enable extension.
2459 0: Disable extension.
2464 Enable or disable pf (pf is short for potentially failed) state. A value
2465 of pf_retrans > path_max_retrans also disables pf state. That is, one of
2466 both pf_enable and pf_retrans > path_max_retrans can disable pf state.
2467 Since pf_retrans and path_max_retrans can be changed by userspace
2468 application, sometimes user expects to disable pf state by the value of
2469 pf_retrans > path_max_retrans, but occasionally the value of pf_retrans
2470 or path_max_retrans is changed by the user application, this pf state is
2471 enabled. As such, it is necessary to add this to dynamically enable
2472 and disable pf state. See:
2473 https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for
2483 Unset or enable/disable pf (pf is short for potentially failed) state
2484 exposure. Applications can control the exposure of the PF path state
2485 in the SCTP_PEER_ADDR_CHANGE event and the SCTP_GET_PEER_ADDR_INFO
2486 sockopt. When it's unset, no SCTP_PEER_ADDR_CHANGE event with
2487 SCTP_ADDR_PF state will be sent and a SCTP_PF-state transport info
2488 can be got via SCTP_GET_PEER_ADDR_INFO sockopt; When it's enabled,
2489 a SCTP_PEER_ADDR_CHANGE event will be sent for a transport becoming
2490 SCTP_PF state and a SCTP_PF-state transport info can be got via
2491 SCTP_GET_PEER_ADDR_INFO sockopt; When it's diabled, no
2492 SCTP_PEER_ADDR_CHANGE event will be sent and it returns -EACCES when
2493 trying to get a SCTP_PF-state transport info via SCTP_GET_PEER_ADDR_INFO
2496 0: Unset pf state exposure, Compatible with old applications.
2498 1: Disable pf state exposure.
2500 2: Enable pf state exposure.
2504 addip_noauth_enable - BOOLEAN
2505 Dynamic Address Reconfiguration (ADD-IP) requires the use of
2506 authentication to protect the operations of adding or removing new
2507 addresses. This requirement is mandated so that unauthorized hosts
2508 would not be able to hijack associations. However, older
2509 implementations may not have implemented this requirement while
2510 allowing the ADD-IP extension. For reasons of interoperability,
2511 we provide this variable to control the enforcement of the
2512 authentication requirement.
2514 == ===============================================================
2515 1 Allow ADD-IP extension to be used without authentication. This
2516 should only be set in a closed environment for interoperability
2517 with older implementations.
2519 0 Enforce the authentication requirement
2520 == ===============================================================
2524 auth_enable - BOOLEAN
2525 Enable or disable Authenticated Chunks extension. This extension
2526 provides the ability to send and receive authenticated chunks and is
2527 required for secure operation of Dynamic Address Reconfiguration
2530 - 1: Enable this extension.
2531 - 0: Disable this extension.
2535 prsctp_enable - BOOLEAN
2536 Enable or disable the Partial Reliability extension (RFC3758) which
2537 is used to notify peers that a given DATA should no longer be expected.
2539 - 1: Enable extension
2545 The limit of the number of new packets that can be initially sent. It
2546 controls how bursty the generated traffic can be.
2550 association_max_retrans - INTEGER
2551 Set the maximum number for retransmissions that an association can
2552 attempt deciding that the remote end is unreachable. If this value
2553 is exceeded, the association is terminated.
2557 max_init_retransmits - INTEGER
2558 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
2559 that an association will attempt before declaring the destination
2560 unreachable and terminating.
2564 path_max_retrans - INTEGER
2565 The maximum number of retransmissions that will be attempted on a given
2566 path. Once this threshold is exceeded, the path is considered
2567 unreachable, and new traffic will use a different path when the
2568 association is multihomed.
2572 pf_retrans - INTEGER
2573 The number of retransmissions that will be attempted on a given path
2574 before traffic is redirected to an alternate transport (should one
2575 exist). Note this is distinct from path_max_retrans, as a path that
2576 passes the pf_retrans threshold can still be used. Its only
2577 deprioritized when a transmission path is selected by the stack. This
2578 setting is primarily used to enable fast failover mechanisms without
2579 having to reduce path_max_retrans to a very low value. See:
2580 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
2581 for details. Note also that a value of pf_retrans > path_max_retrans
2582 disables this feature. Since both pf_retrans and path_max_retrans can
2583 be changed by userspace application, a variable pf_enable is used to
2588 ps_retrans - INTEGER
2589 Primary.Switchover.Max.Retrans (PSMR), it's a tunable parameter coming
2590 from section-5 "Primary Path Switchover" in rfc7829. The primary path
2591 will be changed to another active path when the path error counter on
2592 the old primary path exceeds PSMR, so that "the SCTP sender is allowed
2593 to continue data transmission on a new working path even when the old
2594 primary destination address becomes active again". Note this feature
2595 is disabled by initializing 'ps_retrans' per netns as 0xffff by default,
2596 and its value can't be less than 'pf_retrans' when changing by sysctl.
2600 rto_initial - INTEGER
2601 The initial round trip timeout value in milliseconds that will be used
2602 in calculating round trip times. This is the initial time interval
2603 for retransmissions.
2608 The maximum value (in milliseconds) of the round trip timeout. This
2609 is the largest time interval that can elapse between retransmissions.
2614 The minimum value (in milliseconds) of the round trip timeout. This
2615 is the smallest time interval the can elapse between retransmissions.
2619 hb_interval - INTEGER
2620 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
2621 are sent at the specified interval on idle paths to probe the state of
2622 a given path between 2 associations.
2626 sack_timeout - INTEGER
2627 The amount of time (in milliseconds) that the implementation will wait
2632 valid_cookie_life - INTEGER
2633 The default lifetime of the SCTP cookie (in milliseconds). The cookie
2634 is used during association establishment.
2638 cookie_preserve_enable - BOOLEAN
2639 Enable or disable the ability to extend the lifetime of the SCTP cookie
2640 that is used during the establishment phase of SCTP association
2642 - 1: Enable cookie lifetime extension.
2647 cookie_hmac_alg - STRING
2648 Select the hmac algorithm used when generating the cookie value sent by
2649 a listening sctp socket to a connecting client in the INIT-ACK chunk.
2656 Ability to assign md5 or sha1 as the selected alg is predicated on the
2657 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
2658 CONFIG_CRYPTO_SHA1).
2660 Default: Dependent on configuration. MD5 if available, else SHA1 if
2661 available, else none.
2663 rcvbuf_policy - INTEGER
2664 Determines if the receive buffer is attributed to the socket or to
2665 association. SCTP supports the capability to create multiple
2666 associations on a single socket. When using this capability, it is
2667 possible that a single stalled association that's buffering a lot
2668 of data may block other associations from delivering their data by
2669 consuming all of the receive buffer space. To work around this,
2670 the rcvbuf_policy could be set to attribute the receiver buffer space
2671 to each association instead of the socket. This prevents the described
2674 - 1: rcvbuf space is per association
2675 - 0: rcvbuf space is per socket
2679 sndbuf_policy - INTEGER
2680 Similar to rcvbuf_policy above, this applies to send buffer space.
2682 - 1: Send buffer is tracked per association
2683 - 0: Send buffer is tracked per socket.
2687 sctp_mem - vector of 3 INTEGERs: min, pressure, max
2688 Number of pages allowed for queueing by all SCTP sockets.
2690 min: Below this number of pages SCTP is not bothered about its
2691 memory appetite. When amount of memory allocated by SCTP exceeds
2692 this number, SCTP starts to moderate memory usage.
2694 pressure: This value was introduced to follow format of tcp_mem.
2696 max: Number of pages allowed for queueing by all SCTP sockets.
2698 Default is calculated at boot time from amount of available memory.
2700 sctp_rmem - vector of 3 INTEGERs: min, default, max
2701 Only the first value ("min") is used, "default" and "max" are
2704 min: Minimal size of receive buffer used by SCTP socket.
2705 It is guaranteed to each SCTP socket (but not association) even
2706 under moderate memory pressure.
2710 sctp_wmem - vector of 3 INTEGERs: min, default, max
2711 Currently this tunable has no effect.
2713 addr_scope_policy - INTEGER
2714 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
2716 - 0 - Disable IPv4 address scoping
2717 - 1 - Enable IPv4 address scoping
2718 - 2 - Follow draft but allow IPv4 private addresses
2719 - 3 - Follow draft but allow IPv4 link local addresses
2724 The listening port for the local UDP tunneling sock. Normally it's
2725 using the IANA-assigned UDP port number 9899 (sctp-tunneling).
2727 This UDP sock is used for processing the incoming UDP-encapsulated
2728 SCTP packets (from RFC6951), and shared by all applications in the
2729 same net namespace. This UDP sock will be closed when the value is
2732 The value will also be used to set the src port of the UDP header
2733 for the outgoing UDP-encapsulated SCTP packets. For the dest port,
2734 please refer to 'encap_port' below.
2738 encap_port - INTEGER
2739 The default remote UDP encapsulation port.
2741 This value is used to set the dest port of the UDP header for the
2742 outgoing UDP-encapsulated SCTP packets by default. Users can also
2743 change the value for each sock/asoc/transport by using setsockopt.
2744 For further information, please refer to RFC6951.
2746 Note that when connecting to a remote server, the client should set
2747 this to the port that the UDP tunneling sock on the peer server is
2748 listening to and the local UDP tunneling sock on the client also
2749 must be started. On the server, it would get the encap_port from
2750 the incoming packet's source port.
2755 ``/proc/sys/net/core/*``
2756 ========================
2758 Please see: Documentation/admin-guide/sysctl/net.rst for descriptions of these entries.
2761 ``/proc/sys/net/unix/*``
2762 ========================
2764 max_dgram_qlen - INTEGER
2765 The maximum length of dgram socket receive queue