1 .. SPDX-License-Identifier: GPL-2.0
7 /proc/sys/net/ipv4/* Variables
8 ==============================
11 - 0 - disabled (default)
14 Forward Packets between interfaces.
16 This variable is special, its change resets all configuration
17 parameters to their default state (RFC1122 for hosts, RFC1812
20 ip_default_ttl - INTEGER
21 Default value of TTL field (Time To Live) for outgoing (but not
22 forwarded) IP packets. Should be between 1 and 255 inclusive.
23 Default: 64 (as recommended by RFC1700)
25 ip_no_pmtu_disc - INTEGER
26 Disable Path MTU Discovery. If enabled in mode 1 and a
27 fragmentation-required ICMP is received, the PMTU to this
28 destination will be set to the smallest of the old MTU to
29 this destination and min_pmtu (see below). You will need
30 to raise min_pmtu to the smallest interface MTU on your system
31 manually if you want to avoid locally generated fragments.
33 In mode 2 incoming Path MTU Discovery messages will be
34 discarded. Outgoing frames are handled the same as in mode 1,
35 implicitly setting IP_PMTUDISC_DONT on every created socket.
37 Mode 3 is a hardened pmtu discover mode. The kernel will only
38 accept fragmentation-needed errors if the underlying protocol
39 can verify them besides a plain socket lookup. Current
40 protocols for which pmtu events will be honored are TCP, SCTP
41 and DCCP as they verify e.g. the sequence number or the
42 association. This mode should not be enabled globally but is
43 only intended to secure e.g. name servers in namespaces where
44 TCP path mtu must still work but path MTU information of other
45 protocols should be discarded. If enabled globally this mode
46 could break other protocols.
53 default 552 - minimum Path MTU. Unless this is changed manually,
54 each cached pmtu will never be lower than this setting.
56 ip_forward_use_pmtu - BOOLEAN
57 By default we don't trust protocol path MTUs while forwarding
58 because they could be easily forged and can lead to unwanted
59 fragmentation by the router.
60 You only need to enable this if you have user-space software
61 which tries to discover path mtus by itself and depends on the
62 kernel honoring this information. This is normally not the
72 fwmark_reflect - BOOLEAN
73 Controls the fwmark of kernel-generated IPv4 reply packets that are not
74 associated with a socket for example, TCP RSTs or ICMP echo replies).
75 If unset, these packets have a fwmark of zero. If set, they have the
76 fwmark of the packet they are replying to.
80 fib_multipath_use_neigh - BOOLEAN
81 Use status of existing neighbor entry when determining nexthop for
82 multipath routes. If disabled, neighbor information is not used and
83 packets could be directed to a failed nexthop. Only valid for kernels
84 built with CONFIG_IP_ROUTE_MULTIPATH enabled.
93 fib_multipath_hash_policy - INTEGER
94 Controls which hash policy to use for multipath routes. Only valid
95 for kernels built with CONFIG_IP_ROUTE_MULTIPATH enabled.
103 - 2 - Layer 3 or inner Layer 3 if present
104 - 3 - Custom multipath hash. Fields used for multipath hash calculation
105 are determined by fib_multipath_hash_fields sysctl
107 fib_multipath_hash_fields - UNSIGNED INTEGER
108 When fib_multipath_hash_policy is set to 3 (custom multipath hash), the
109 fields used for multipath hash calculation are determined by this
112 This value is a bitmask which enables various fields for multipath hash
117 ====== ============================
118 0x0001 Source IP address
119 0x0002 Destination IP address
121 0x0008 Unused (Flow Label)
123 0x0020 Destination port
124 0x0040 Inner source IP address
125 0x0080 Inner destination IP address
126 0x0100 Inner IP protocol
127 0x0200 Inner Flow Label
128 0x0400 Inner source port
129 0x0800 Inner destination port
130 ====== ============================
132 Default: 0x0007 (source IP, destination IP and IP protocol)
134 fib_sync_mem - UNSIGNED INTEGER
135 Amount of dirty memory from fib entries that can be backlogged before
136 synchronize_rcu is forced.
138 Default: 512kB Minimum: 64kB Maximum: 64MB
140 ip_forward_update_priority - INTEGER
141 Whether to update SKB priority from "TOS" field in IPv4 header after it
142 is forwarded. The new SKB priority is mapped from TOS field value
143 according to an rt_tos2priority table (see e.g. man tc-prio).
145 Default: 1 (Update priority.)
149 - 0 - Do not update priority.
150 - 1 - Update priority.
152 route/max_size - INTEGER
153 Maximum number of routes allowed in the kernel. Increase
154 this when using large numbers of interfaces and/or routes.
156 From linux kernel 3.6 onwards, this is deprecated for ipv4
157 as route cache is no longer used.
159 From linux kernel 6.3 onwards, this is deprecated for ipv6
160 as garbage collection manages cached route entries.
162 neigh/default/gc_thresh1 - INTEGER
163 Minimum number of entries to keep. Garbage collector will not
164 purge entries if there are fewer than this number.
168 neigh/default/gc_thresh2 - INTEGER
169 Threshold when garbage collector becomes more aggressive about
170 purging entries. Entries older than 5 seconds will be cleared
171 when over this number.
175 neigh/default/gc_thresh3 - INTEGER
176 Maximum number of non-PERMANENT neighbor entries allowed. Increase
177 this when using large numbers of interfaces and when communicating
178 with large numbers of directly-connected peers.
182 neigh/default/unres_qlen_bytes - INTEGER
183 The maximum number of bytes which may be used by packets
184 queued for each unresolved address by other network layers.
187 Setting negative value is meaningless and will return error.
189 Default: SK_WMEM_MAX, (same as net.core.wmem_default).
191 Exact value depends on architecture and kernel options,
192 but should be enough to allow queuing 256 packets
195 neigh/default/unres_qlen - INTEGER
196 The maximum number of packets which may be queued for each
197 unresolved address by other network layers.
199 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
201 Prior to linux 3.3, the default value is 3 which may cause
202 unexpected packet loss. The current default value is calculated
203 according to default value of unres_qlen_bytes and true size of
208 neigh/default/interval_probe_time_ms - INTEGER
209 The probe interval for neighbor entries with NTF_MANAGED flag,
214 mtu_expires - INTEGER
215 Time, in seconds, that cached PMTU information is kept.
217 min_adv_mss - INTEGER
218 The advertised MSS depends on the first hop route MTU, but will
219 never be lower than this setting.
221 fib_notify_on_flag_change - INTEGER
222 Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/
223 RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed.
225 After installing a route to the kernel, user space receives an
226 acknowledgment, which means the route was installed in the kernel,
227 but not necessarily in hardware.
228 It is also possible for a route already installed in hardware to change
229 its action and therefore its flags. For example, a host route that is
230 trapping packets can be "promoted" to perform decapsulation following
231 the installation of an IPinIP/VXLAN tunnel.
232 The notifications will indicate to user-space the state of the route.
234 Default: 0 (Do not emit notifications.)
238 - 0 - Do not emit notifications.
239 - 1 - Emit notifications.
240 - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change.
244 ipfrag_high_thresh - LONG INTEGER
245 Maximum memory used to reassemble IP fragments.
247 ipfrag_low_thresh - LONG INTEGER
248 (Obsolete since linux-4.17)
249 Maximum memory used to reassemble IP fragments before the kernel
250 begins to remove incomplete fragment queues to free up resources.
251 The kernel still accepts new fragments for defragmentation.
253 ipfrag_time - INTEGER
254 Time in seconds to keep an IP fragment in memory.
256 ipfrag_max_dist - INTEGER
257 ipfrag_max_dist is a non-negative integer value which defines the
258 maximum "disorder" which is allowed among fragments which share a
259 common IP source address. Note that reordering of packets is
260 not unusual, but if a large number of fragments arrive from a source
261 IP address while a particular fragment queue remains incomplete, it
262 probably indicates that one or more fragments belonging to that queue
263 have been lost. When ipfrag_max_dist is positive, an additional check
264 is done on fragments before they are added to a reassembly queue - if
265 ipfrag_max_dist (or more) fragments have arrived from a particular IP
266 address between additions to any IP fragment queue using that source
267 address, it's presumed that one or more fragments in the queue are
268 lost. The existing fragment queue will be dropped, and a new one
269 started. An ipfrag_max_dist value of zero disables this check.
271 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
272 result in unnecessarily dropping fragment queues when normal
273 reordering of packets occurs, which could lead to poor application
274 performance. Using a very large value, e.g. 50000, increases the
275 likelihood of incorrectly reassembling IP fragments that originate
276 from different IP datagrams, which could result in data corruption.
279 bc_forwarding - INTEGER
280 bc_forwarding enables the feature described in rfc1812#section-5.3.5.2
281 and rfc2644. It allows the router to forward directed broadcast.
282 To enable this feature, the 'all' entry and the input interface entry
289 inet_peer_threshold - INTEGER
290 The approximate size of the storage. Starting from this threshold
291 entries will be thrown aggressively. This threshold also determines
292 entries' time-to-live and time intervals between garbage collection
293 passes. More entries, less time-to-live, less GC interval.
295 inet_peer_minttl - INTEGER
296 Minimum time-to-live of entries. Should be enough to cover fragment
297 time-to-live on the reassembling side. This minimum time-to-live is
298 guaranteed if the pool size is less than inet_peer_threshold.
301 inet_peer_maxttl - INTEGER
302 Maximum time-to-live of entries. Unused entries will expire after
303 this period of time if there is no memory pressure on the pool (i.e.
304 when the number of entries in the pool is very small).
311 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
312 Defaults to 4096. (Was 128 before linux-5.4)
313 See also tcp_max_syn_backlog for additional tuning for TCP sockets.
315 tcp_abort_on_overflow - BOOLEAN
316 If listening service is too slow to accept new connections,
317 reset them. Default state is FALSE. It means that if overflow
318 occurred due to a burst, connection will recover. Enable this
319 option _only_ if you are really sure that listening daemon
320 cannot be tuned to accept connections faster. Enabling this
321 option can harm clients of your server.
323 tcp_adv_win_scale - INTEGER
324 Count buffering overhead as bytes/2^tcp_adv_win_scale
325 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
328 Possible values are [-31, 31], inclusive.
332 tcp_allowed_congestion_control - STRING
333 Show/set the congestion control choices available to non-privileged
334 processes. The list is a subset of those listed in
335 tcp_available_congestion_control.
337 Default is "reno" and the default setting (tcp_congestion_control).
339 tcp_app_win - INTEGER
340 Reserve max(window/2^tcp_app_win, mss) of window for application
341 buffer. Value 0 is special, it means that nothing is reserved.
345 tcp_autocorking - BOOLEAN
346 Enable TCP auto corking :
347 When applications do consecutive small write()/sendmsg() system calls,
348 we try to coalesce these small writes as much as possible, to lower
349 total amount of sent packets. This is done if at least one prior
350 packet for the flow is waiting in Qdisc queues or device transmit
351 queue. Applications can still use TCP_CORK for optimal behavior
352 when they know how/when to uncork their sockets.
356 tcp_available_congestion_control - STRING
357 Shows the available congestion control choices that are registered.
358 More congestion control algorithms may be available as modules,
361 tcp_base_mss - INTEGER
362 The initial value of search_low to be used by the packetization layer
363 Path MTU discovery (MTU probing). If MTU probing is enabled,
364 this is the initial MSS used by the connection.
366 tcp_mtu_probe_floor - INTEGER
367 If MTU probing is enabled this caps the minimum MSS used for search_low
372 tcp_min_snd_mss - INTEGER
373 TCP SYN and SYNACK messages usually advertise an ADVMSS option,
374 as described in RFC 1122 and RFC 6691.
376 If this ADVMSS option is smaller than tcp_min_snd_mss,
377 it is silently capped to tcp_min_snd_mss.
379 Default : 48 (at least 8 bytes of payload per segment)
381 tcp_congestion_control - STRING
382 Set the congestion control algorithm to be used for new
383 connections. The algorithm "reno" is always available, but
384 additional choices may be available based on kernel configuration.
385 Default is set as part of kernel configuration.
386 For passive connections, the listener congestion control choice
389 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
392 Allows TCP to send "duplicate" SACKs.
394 tcp_early_retrans - INTEGER
395 Tail loss probe (TLP) converts RTOs occurring due to tail
396 losses into fast recovery (draft-ietf-tcpm-rack). Note that
397 TLP requires RACK to function properly (see tcp_recovery below)
407 Control use of Explicit Congestion Notification (ECN) by TCP.
408 ECN is used only when both ends of the TCP connection indicate
409 support for it. This feature is useful in avoiding losses due
410 to congestion by allowing supporting routers to signal
411 congestion before having to drop packets.
415 = =====================================================
416 0 Disable ECN. Neither initiate nor accept ECN.
417 1 Enable ECN when requested by incoming connections and
418 also request ECN on outgoing connection attempts.
419 2 Enable ECN when requested by incoming connections
420 but do not request ECN on outgoing connections.
421 = =====================================================
425 tcp_ecn_fallback - BOOLEAN
426 If the kernel detects that ECN connection misbehaves, enable fall
427 back to non-ECN. Currently, this knob implements the fallback
428 from RFC3168, section 6.1.1.1., but we reserve that in future,
429 additional detection mechanisms could be implemented under this
430 knob. The value is not used, if tcp_ecn or per route (or congestion
431 control) ECN settings are disabled.
433 Default: 1 (fallback enabled)
436 This is a legacy option, it has no effect anymore.
438 tcp_fin_timeout - INTEGER
439 The length of time an orphaned (no longer referenced by any
440 application) connection will remain in the FIN_WAIT_2 state
441 before it is aborted at the local end. While a perfectly
442 valid "receive only" state for an un-orphaned connection, an
443 orphaned connection in FIN_WAIT_2 state could otherwise wait
444 forever for the remote to close its end of the connection.
451 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
452 F-RTO is an enhanced recovery algorithm for TCP retransmission
453 timeouts. It is particularly beneficial in networks where the
454 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
455 modification. It does not require any support from the peer.
457 By default it's enabled with a non-zero value. 0 disables F-RTO.
459 tcp_fwmark_accept - BOOLEAN
460 If set, incoming connections to listening sockets that do not have a
461 socket mark will set the mark of the accepting socket to the fwmark of
462 the incoming SYN packet. This will cause all packets on that connection
463 (starting from the first SYNACK) to be sent with that fwmark. The
464 listening socket's mark is unchanged. Listening sockets that already
465 have a fwmark set via setsockopt(SOL_SOCKET, SO_MARK, ...) are
470 tcp_invalid_ratelimit - INTEGER
471 Limit the maximal rate for sending duplicate acknowledgments
472 in response to incoming TCP packets that are for an existing
473 connection but that are invalid due to any of these reasons:
475 (a) out-of-window sequence number,
476 (b) out-of-window acknowledgment number, or
477 (c) PAWS (Protection Against Wrapped Sequence numbers) check failure
479 This can help mitigate simple "ack loop" DoS attacks, wherein
480 a buggy or malicious middlebox or man-in-the-middle can
481 rewrite TCP header fields in manner that causes each endpoint
482 to think that the other is sending invalid TCP segments, thus
483 causing each side to send an unterminating stream of duplicate
484 acknowledgments for invalid segments.
486 Using 0 disables rate-limiting of dupacks in response to
487 invalid segments; otherwise this value specifies the minimal
488 space between sending such dupacks, in milliseconds.
490 Default: 500 (milliseconds).
492 tcp_keepalive_time - INTEGER
493 How often TCP sends out keepalive messages when keepalive is enabled.
496 tcp_keepalive_probes - INTEGER
497 How many keepalive probes TCP sends out, until it decides that the
498 connection is broken. Default value: 9.
500 tcp_keepalive_intvl - INTEGER
501 How frequently the probes are send out. Multiplied by
502 tcp_keepalive_probes it is time to kill not responding connection,
503 after probes started. Default value: 75sec i.e. connection
504 will be aborted after ~11 minutes of retries.
506 tcp_l3mdev_accept - BOOLEAN
507 Enables child sockets to inherit the L3 master device index.
508 Enabling this option allows a "global" listen socket to work
509 across L3 master domains (e.g., VRFs) with connected sockets
510 derived from the listen socket to be bound to the L3 domain in
511 which the packets originated. Only valid when the kernel was
512 compiled with CONFIG_NET_L3_MASTER_DEV.
514 Default: 0 (disabled)
516 tcp_low_latency - BOOLEAN
517 This is a legacy option, it has no effect anymore.
519 tcp_max_orphans - INTEGER
520 Maximal number of TCP sockets not attached to any user file handle,
521 held by system. If this number is exceeded orphaned connections are
522 reset immediately and warning is printed. This limit exists
523 only to prevent simple DoS attacks, you _must_ not rely on this
524 or lower the limit artificially, but rather increase it
525 (probably, after increasing installed memory),
526 if network conditions require more than default value,
527 and tune network services to linger and kill such states
528 more aggressively. Let me to remind again: each orphan eats
529 up to ~64K of unswappable memory.
531 tcp_max_syn_backlog - INTEGER
532 Maximal number of remembered connection requests (SYN_RECV),
533 which have not received an acknowledgment from connecting client.
535 This is a per-listener limit.
537 The minimal value is 128 for low memory machines, and it will
538 increase in proportion to the memory of machine.
540 If server suffers from overload, try increasing this number.
542 Remember to also check /proc/sys/net/core/somaxconn
543 A SYN_RECV request socket consumes about 304 bytes of memory.
545 tcp_max_tw_buckets - INTEGER
546 Maximal number of timewait sockets held by system simultaneously.
547 If this number is exceeded time-wait socket is immediately destroyed
548 and warning is printed. This limit exists only to prevent
549 simple DoS attacks, you _must_ not lower the limit artificially,
550 but rather increase it (probably, after increasing installed memory),
551 if network conditions require more than default value.
553 tcp_mem - vector of 3 INTEGERs: min, pressure, max
554 min: below this number of pages TCP is not bothered about its
557 pressure: when amount of memory allocated by TCP exceeds this number
558 of pages, TCP moderates its memory consumption and enters memory
559 pressure mode, which is exited when memory consumption falls
562 max: number of pages allowed for queueing by all TCP sockets.
564 Defaults are calculated at boot time from amount of available
567 tcp_min_rtt_wlen - INTEGER
568 The window length of the windowed min filter to track the minimum RTT.
569 A shorter window lets a flow more quickly pick up new (higher)
570 minimum RTT when it is moved to a longer path (e.g., due to traffic
571 engineering). A longer window makes the filter more resistant to RTT
572 inflations such as transient congestion. The unit is seconds.
574 Possible values: 0 - 86400 (1 day)
578 tcp_moderate_rcvbuf - BOOLEAN
579 If set, TCP performs receive buffer auto-tuning, attempting to
580 automatically size the buffer (no greater than tcp_rmem[2]) to
581 match the size required by the path for full throughput. Enabled by
584 tcp_mtu_probing - INTEGER
585 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
589 - 1 - Disabled by default, enabled when an ICMP black hole detected
590 - 2 - Always enabled, use initial MSS of tcp_base_mss.
592 tcp_probe_interval - UNSIGNED INTEGER
593 Controls how often to start TCP Packetization-Layer Path MTU
594 Discovery reprobe. The default is reprobing every 10 minutes as
597 tcp_probe_threshold - INTEGER
598 Controls when TCP Packetization-Layer Path MTU Discovery probing
599 will stop in respect to the width of search range in bytes. Default
602 tcp_no_metrics_save - BOOLEAN
603 By default, TCP saves various connection metrics in the route cache
604 when the connection closes, so that connections established in the
605 near future can use these to set initial conditions. Usually, this
606 increases overall performance, but may sometimes cause performance
607 degradation. If set, TCP will not cache metrics on closing
610 tcp_no_ssthresh_metrics_save - BOOLEAN
611 Controls whether TCP saves ssthresh metrics in the route cache.
613 Default is 1, which disables ssthresh metrics.
615 tcp_orphan_retries - INTEGER
616 This value influences the timeout of a locally closed TCP connection,
617 when RTO retransmissions remain unacknowledged.
618 See tcp_retries2 for more details.
620 The default value is 8.
622 If your machine is a loaded WEB server,
623 you should think about lowering this value, such sockets
624 may consume significant resources. Cf. tcp_max_orphans.
626 tcp_recovery - INTEGER
627 This value is a bitmap to enable various experimental loss recovery
630 ========= =============================================================
631 RACK: 0x1 enables the RACK loss detection for fast detection of lost
632 retransmissions and tail drops. It also subsumes and disables
633 RFC6675 recovery for SACK connections.
635 RACK: 0x2 makes RACK's reordering window static (min_rtt/4).
637 RACK: 0x4 disables RACK's DUPACK threshold heuristic
638 ========= =============================================================
642 tcp_reflect_tos - BOOLEAN
643 For listening sockets, reuse the DSCP value of the initial SYN message
644 for outgoing packets. This allows to have both directions of a TCP
645 stream to use the same DSCP value, assuming DSCP remains unchanged for
646 the lifetime of the connection.
648 This options affects both IPv4 and IPv6.
650 Default: 0 (disabled)
652 tcp_reordering - INTEGER
653 Initial reordering level of packets in a TCP stream.
654 TCP stack can then dynamically adjust flow reordering level
655 between this initial value and tcp_max_reordering
659 tcp_max_reordering - INTEGER
660 Maximal reordering level of packets in a TCP stream.
661 300 is a fairly conservative value, but you might increase it
662 if paths are using per packet load balancing (like bonding rr mode)
666 tcp_retrans_collapse - BOOLEAN
667 Bug-to-bug compatibility with some broken printers.
668 On retransmit try to send bigger packets to work around bugs in
671 tcp_retries1 - INTEGER
672 This value influences the time, after which TCP decides, that
673 something is wrong due to unacknowledged RTO retransmissions,
674 and reports this suspicion to the network layer.
675 See tcp_retries2 for more details.
677 RFC 1122 recommends at least 3 retransmissions, which is the
680 tcp_retries2 - INTEGER
681 This value influences the timeout of an alive TCP connection,
682 when RTO retransmissions remain unacknowledged.
683 Given a value of N, a hypothetical TCP connection following
684 exponential backoff with an initial RTO of TCP_RTO_MIN would
685 retransmit N times before killing the connection at the (N+1)th RTO.
687 The default value of 15 yields a hypothetical timeout of 924.6
688 seconds and is a lower bound for the effective timeout.
689 TCP will effectively time out at the first RTO which exceeds the
690 hypothetical timeout.
692 RFC 1122 recommends at least 100 seconds for the timeout,
693 which corresponds to a value of at least 8.
695 tcp_rfc1337 - BOOLEAN
696 If set, the TCP stack behaves conforming to RFC1337. If unset,
697 we are not conforming to RFC, but prevent TCP TIME_WAIT
702 tcp_rmem - vector of 3 INTEGERs: min, default, max
703 min: Minimal size of receive buffer used by TCP sockets.
704 It is guaranteed to each TCP socket, even under moderate memory
709 default: initial size of receive buffer used by TCP sockets.
710 This value overrides net.core.rmem_default used by other protocols.
711 Default: 131072 bytes.
712 This value results in initial window of 65535.
714 max: maximal size of receive buffer allowed for automatically
715 selected receiver buffers for TCP socket. This value does not override
716 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
717 automatic tuning of that socket's receive buffer size, in which
718 case this value is ignored.
719 Default: between 131072 and 6MB, depending on RAM size.
722 Enable select acknowledgments (SACKS).
724 tcp_comp_sack_delay_ns - LONG INTEGER
725 TCP tries to reduce number of SACK sent, using a timer
726 based on 5% of SRTT, capped by this sysctl, in nano seconds.
727 The default is 1ms, based on TSO autosizing period.
729 Default : 1,000,000 ns (1 ms)
731 tcp_comp_sack_slack_ns - LONG INTEGER
732 This sysctl control the slack used when arming the
733 timer used by SACK compression. This gives extra time
734 for small RTT flows, and reduces system overhead by allowing
735 opportunistic reduction of timer interrupts.
737 Default : 100,000 ns (100 us)
739 tcp_comp_sack_nr - INTEGER
740 Max number of SACK that can be compressed.
741 Using 0 disables SACK compression.
745 tcp_slow_start_after_idle - BOOLEAN
746 If set, provide RFC2861 behavior and time out the congestion
747 window after an idle period. An idle period is defined at
748 the current RTO. If unset, the congestion window will not
749 be timed out after an idle period.
754 Use the Host requirements interpretation of the TCP urgent pointer field.
755 Most hosts use the older BSD interpretation, so if you turn this on
756 Linux might not communicate correctly with them.
760 tcp_synack_retries - INTEGER
761 Number of times SYNACKs for a passive TCP connection attempt will
762 be retransmitted. Should not be higher than 255. Default value
763 is 5, which corresponds to 31seconds till the last retransmission
764 with the current initial RTO of 1second. With this the final timeout
765 for a passive TCP connection will happen after 63seconds.
767 tcp_syncookies - INTEGER
768 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
769 Send out syncookies when the syn backlog queue of a socket
770 overflows. This is to prevent against the common 'SYN flood attack'
773 Note, that syncookies is fallback facility.
774 It MUST NOT be used to help highly loaded servers to stand
775 against legal connection rate. If you see SYN flood warnings
776 in your logs, but investigation shows that they occur
777 because of overload with legal connections, you should tune
778 another parameters until this warning disappear.
779 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
781 syncookies seriously violate TCP protocol, do not allow
782 to use TCP extensions, can result in serious degradation
783 of some services (f.e. SMTP relaying), visible not by you,
784 but your clients and relays, contacting you. While you see
785 SYN flood warnings in logs not being really flooded, your server
786 is seriously misconfigured.
788 If you want to test which effects syncookies have to your
789 network connections you can set this knob to 2 to enable
790 unconditionally generation of syncookies.
792 tcp_migrate_req - BOOLEAN
793 The incoming connection is tied to a specific listening socket when
794 the initial SYN packet is received during the three-way handshake.
795 When a listener is closed, in-flight request sockets during the
796 handshake and established sockets in the accept queue are aborted.
798 If the listener has SO_REUSEPORT enabled, other listeners on the
799 same port should have been able to accept such connections. This
800 option makes it possible to migrate such child sockets to another
801 listener after close() or shutdown().
803 The BPF_SK_REUSEPORT_SELECT_OR_MIGRATE type of eBPF program should
804 usually be used to define the policy to pick an alive listener.
805 Otherwise, the kernel will randomly pick an alive listener only if
806 this option is enabled.
808 Note that migration between listeners with different settings may
809 crash applications. Let's say migration happens from listener A to
810 B, and only B has TCP_SAVE_SYN enabled. B cannot read SYN data from
811 the requests migrated from A. To avoid such a situation, cancel
812 migration by returning SK_DROP in the type of eBPF program, or
817 tcp_fastopen - INTEGER
818 Enable TCP Fast Open (RFC7413) to send and accept data in the opening
821 The client support is enabled by flag 0x1 (on by default). The client
822 then must use sendmsg() or sendto() with the MSG_FASTOPEN flag,
823 rather than connect() to send data in SYN.
825 The server support is enabled by flag 0x2 (off by default). Then
826 either enable for all listeners with another flag (0x400) or
827 enable individual listeners via TCP_FASTOPEN socket option with
828 the option value being the length of the syn-data backlog.
830 The values (bitmap) are
832 ===== ======== ======================================================
833 0x1 (client) enables sending data in the opening SYN on the client.
834 0x2 (server) enables the server support, i.e., allowing data in
835 a SYN packet to be accepted and passed to the
836 application before 3-way handshake finishes.
837 0x4 (client) send data in the opening SYN regardless of cookie
838 availability and without a cookie option.
839 0x200 (server) accept data-in-SYN w/o any cookie option present.
840 0x400 (server) enable all listeners to support Fast Open by
841 default without explicit TCP_FASTOPEN socket option.
842 ===== ======== ======================================================
846 Note that additional client or server features are only
847 effective if the basic support (0x1 and 0x2) are enabled respectively.
849 tcp_fastopen_blackhole_timeout_sec - INTEGER
850 Initial time period in second to disable Fastopen on active TCP sockets
851 when a TFO firewall blackhole issue happens.
852 This time period will grow exponentially when more blackhole issues
853 get detected right after Fastopen is re-enabled and will reset to
854 initial value when the blackhole issue goes away.
855 0 to disable the blackhole detection.
857 By default, it is set to 0 (feature is disabled).
859 tcp_fastopen_key - list of comma separated 32-digit hexadecimal INTEGERs
860 The list consists of a primary key and an optional backup key. The
861 primary key is used for both creating and validating cookies, while the
862 optional backup key is only used for validating cookies. The purpose of
863 the backup key is to maximize TFO validation when keys are rotated.
865 A randomly chosen primary key may be configured by the kernel if
866 the tcp_fastopen sysctl is set to 0x400 (see above), or if the
867 TCP_FASTOPEN setsockopt() optname is set and a key has not been
868 previously configured via sysctl. If keys are configured via
869 setsockopt() by using the TCP_FASTOPEN_KEY optname, then those
870 per-socket keys will be used instead of any keys that are specified via
873 A key is specified as 4 8-digit hexadecimal integers which are separated
874 by a '-' as: xxxxxxxx-xxxxxxxx-xxxxxxxx-xxxxxxxx. Leading zeros may be
875 omitted. A primary and a backup key may be specified by separating them
876 by a comma. If only one key is specified, it becomes the primary key and
877 any previously configured backup keys are removed.
879 tcp_syn_retries - INTEGER
880 Number of times initial SYNs for an active TCP connection attempt
881 will be retransmitted. Should not be higher than 127. Default value
882 is 6, which corresponds to 63seconds till the last retransmission
883 with the current initial RTO of 1second. With this the final timeout
884 for an active TCP connection attempt will happen after 127seconds.
886 tcp_timestamps - INTEGER
887 Enable timestamps as defined in RFC1323.
890 - 1: Enable timestamps as defined in RFC1323 and use random offset for
891 each connection rather than only using the current time.
892 - 2: Like 1, but without random offsets.
896 tcp_min_tso_segs - INTEGER
897 Minimal number of segments per TSO frame.
899 Since linux-3.12, TCP does an automatic sizing of TSO frames,
900 depending on flow rate, instead of filling 64Kbytes packets.
901 For specific usages, it's possible to force TCP to build big
902 TSO frames. Note that TCP stack might split too big TSO packets
903 if available window is too small.
907 tcp_tso_rtt_log - INTEGER
908 Adjustment of TSO packet sizes based on min_rtt
910 Starting from linux-5.18, TCP autosizing can be tweaked
911 for flows having small RTT.
913 Old autosizing was splitting the pacing budget to send 1024 TSO
916 tso_packet_size = sk->sk_pacing_rate / 1024;
918 With the new mechanism, we increase this TSO sizing using:
920 distance = min_rtt_usec / (2^tcp_tso_rtt_log)
921 tso_packet_size += gso_max_size >> distance;
923 This means that flows between very close hosts can use bigger
924 TSO packets, reducing their cpu costs.
926 If you want to use the old autosizing, set this sysctl to 0.
928 Default: 9 (2^9 = 512 usec)
930 tcp_pacing_ss_ratio - INTEGER
931 sk->sk_pacing_rate is set by TCP stack using a ratio applied
932 to current rate. (current_rate = cwnd * mss / srtt)
933 If TCP is in slow start, tcp_pacing_ss_ratio is applied
934 to let TCP probe for bigger speeds, assuming cwnd can be
935 doubled every other RTT.
939 tcp_pacing_ca_ratio - INTEGER
940 sk->sk_pacing_rate is set by TCP stack using a ratio applied
941 to current rate. (current_rate = cwnd * mss / srtt)
942 If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio
943 is applied to conservatively probe for bigger throughput.
947 tcp_tso_win_divisor - INTEGER
948 This allows control over what percentage of the congestion window
949 can be consumed by a single TSO frame.
950 The setting of this parameter is a choice between burstiness and
951 building larger TSO frames.
955 tcp_tw_reuse - INTEGER
956 Enable reuse of TIME-WAIT sockets for new connections when it is
957 safe from protocol viewpoint.
961 - 2 - enable for loopback traffic only
963 It should not be changed without advice/request of technical
968 tcp_window_scaling - BOOLEAN
969 Enable window scaling as defined in RFC1323.
971 tcp_wmem - vector of 3 INTEGERs: min, default, max
972 min: Amount of memory reserved for send buffers for TCP sockets.
973 Each TCP socket has rights to use it due to fact of its birth.
977 default: initial size of send buffer used by TCP sockets. This
978 value overrides net.core.wmem_default used by other protocols.
980 It is usually lower than net.core.wmem_default.
984 max: Maximal amount of memory allowed for automatically tuned
985 send buffers for TCP sockets. This value does not override
986 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
987 automatic tuning of that socket's send buffer size, in which case
988 this value is ignored.
990 Default: between 64K and 4MB, depending on RAM size.
992 tcp_notsent_lowat - UNSIGNED INTEGER
993 A TCP socket can control the amount of unsent bytes in its write queue,
994 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
995 reports POLLOUT events if the amount of unsent bytes is below a per
996 socket value, and if the write queue is not full. sendmsg() will
997 also not add new buffers if the limit is hit.
999 This global variable controls the amount of unsent data for
1000 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
1001 to the global variable has immediate effect.
1003 Default: UINT_MAX (0xFFFFFFFF)
1005 tcp_workaround_signed_windows - BOOLEAN
1006 If set, assume no receipt of a window scaling option means the
1007 remote TCP is broken and treats the window as a signed quantity.
1008 If unset, assume the remote TCP is not broken even if we do
1009 not receive a window scaling option from them.
1013 tcp_thin_linear_timeouts - BOOLEAN
1014 Enable dynamic triggering of linear timeouts for thin streams.
1015 If set, a check is performed upon retransmission by timeout to
1016 determine if the stream is thin (less than 4 packets in flight).
1017 As long as the stream is found to be thin, up to 6 linear
1018 timeouts may be performed before exponential backoff mode is
1019 initiated. This improves retransmission latency for
1020 non-aggressive thin streams, often found to be time-dependent.
1021 For more information on thin streams, see
1022 Documentation/networking/tcp-thin.rst
1026 tcp_limit_output_bytes - INTEGER
1027 Controls TCP Small Queue limit per tcp socket.
1028 TCP bulk sender tends to increase packets in flight until it
1029 gets losses notifications. With SNDBUF autotuning, this can
1030 result in a large amount of packets queued on the local machine
1031 (e.g.: qdiscs, CPU backlog, or device) hurting latency of other
1032 flows, for typical pfifo_fast qdiscs. tcp_limit_output_bytes
1033 limits the number of bytes on qdisc or device to reduce artificial
1034 RTT/cwnd and reduce bufferbloat.
1036 Default: 1048576 (16 * 65536)
1038 tcp_challenge_ack_limit - INTEGER
1039 Limits number of Challenge ACK sent per second, as recommended
1040 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
1041 Note that this per netns rate limit can allow some side channel
1042 attacks and probably should not be enabled.
1043 TCP stack implements per TCP socket limits anyway.
1044 Default: INT_MAX (unlimited)
1046 tcp_ehash_entries - INTEGER
1047 Show the number of hash buckets for TCP sockets in the current
1048 networking namespace.
1050 A negative value means the networking namespace does not own its
1051 hash buckets and shares the initial networking namespace's one.
1053 tcp_child_ehash_entries - INTEGER
1054 Control the number of hash buckets for TCP sockets in the child
1055 networking namespace, which must be set before clone() or unshare().
1057 If the value is not 0, the kernel uses a value rounded up to 2^n
1058 as the actual hash bucket size. 0 is a special value, meaning
1059 the child networking namespace will share the initial networking
1060 namespace's hash buckets.
1062 Note that the child will use the global one in case the kernel
1063 fails to allocate enough memory. In addition, the global hash
1064 buckets are spread over available NUMA nodes, but the allocation
1065 of the child hash table depends on the current process's NUMA
1066 policy, which could result in performance differences.
1068 Note also that the default value of tcp_max_tw_buckets and
1069 tcp_max_syn_backlog depend on the hash bucket size.
1071 Possible values: 0, 2^n (n: 0 - 24 (16Mi))
1075 tcp_plb_enabled - BOOLEAN
1076 If set and the underlying congestion control (e.g. DCTCP) supports
1077 and enables PLB feature, TCP PLB (Protective Load Balancing) is
1078 enabled. PLB is described in the following paper:
1079 https://doi.org/10.1145/3544216.3544226. Based on PLB parameters,
1080 upon sensing sustained congestion, TCP triggers a change in
1081 flow label field for outgoing IPv6 packets. A change in flow label
1082 field potentially changes the path of outgoing packets for switches
1083 that use ECMP/WCMP for routing.
1085 PLB changes socket txhash which results in a change in IPv6 Flow Label
1086 field, and currently no-op for IPv4 headers. It is possible
1087 to apply PLB for IPv4 with other network header fields (e.g. TCP
1088 or IPv4 options) or using encapsulation where outer header is used
1089 by switches to determine next hop. In either case, further host
1090 and switch side changes will be needed.
1092 When set, PLB assumes that congestion signal (e.g. ECN) is made
1093 available and used by congestion control module to estimate a
1094 congestion measure (e.g. ce_ratio). PLB needs a congestion measure to
1095 make repathing decisions.
1099 tcp_plb_idle_rehash_rounds - INTEGER
1100 Number of consecutive congested rounds (RTT) seen after which
1101 a rehash can be performed, given there are no packets in flight.
1102 This is referred to as M in PLB paper:
1103 https://doi.org/10.1145/3544216.3544226.
1105 Possible Values: 0 - 31
1109 tcp_plb_rehash_rounds - INTEGER
1110 Number of consecutive congested rounds (RTT) seen after which
1111 a forced rehash can be performed. Be careful when setting this
1112 parameter, as a small value increases the risk of retransmissions.
1113 This is referred to as N in PLB paper:
1114 https://doi.org/10.1145/3544216.3544226.
1116 Possible Values: 0 - 31
1120 tcp_plb_suspend_rto_sec - INTEGER
1121 Time, in seconds, to suspend PLB in event of an RTO. In order to avoid
1122 having PLB repath onto a connectivity "black hole", after an RTO a TCP
1123 connection suspends PLB repathing for a random duration between 1x and
1124 2x of this parameter. Randomness is added to avoid concurrent rehashing
1125 of multiple TCP connections. This should be set corresponding to the
1126 amount of time it takes to repair a failed link.
1128 Possible Values: 0 - 255
1132 tcp_plb_cong_thresh - INTEGER
1133 Fraction of packets marked with congestion over a round (RTT) to
1134 tag that round as congested. This is referred to as K in the PLB paper:
1135 https://doi.org/10.1145/3544216.3544226.
1137 The 0-1 fraction range is mapped to 0-256 range to avoid floating
1138 point operations. For example, 128 means that if at least 50% of
1139 the packets in a round were marked as congested then the round
1140 will be tagged as congested.
1142 Setting threshold to 0 means that PLB repaths every RTT regardless
1143 of congestion. This is not intended behavior for PLB and should be
1144 used only for experimentation purpose.
1146 Possible Values: 0 - 256
1153 udp_l3mdev_accept - BOOLEAN
1154 Enabling this option allows a "global" bound socket to work
1155 across L3 master domains (e.g., VRFs) with packets capable of
1156 being received regardless of the L3 domain in which they
1157 originated. Only valid when the kernel was compiled with
1158 CONFIG_NET_L3_MASTER_DEV.
1160 Default: 0 (disabled)
1162 udp_mem - vector of 3 INTEGERs: min, pressure, max
1163 Number of pages allowed for queueing by all UDP sockets.
1165 min: Number of pages allowed for queueing by all UDP sockets.
1167 pressure: This value was introduced to follow format of tcp_mem.
1169 max: This value was introduced to follow format of tcp_mem.
1171 Default is calculated at boot time from amount of available memory.
1173 udp_rmem_min - INTEGER
1174 Minimal size of receive buffer used by UDP sockets in moderation.
1175 Each UDP socket is able to use the size for receiving data, even if
1176 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
1180 udp_wmem_min - INTEGER
1181 UDP does not have tx memory accounting and this tunable has no effect.
1183 udp_hash_entries - INTEGER
1184 Show the number of hash buckets for UDP sockets in the current
1185 networking namespace.
1187 A negative value means the networking namespace does not own its
1188 hash buckets and shares the initial networking namespace's one.
1190 udp_child_ehash_entries - INTEGER
1191 Control the number of hash buckets for UDP sockets in the child
1192 networking namespace, which must be set before clone() or unshare().
1194 If the value is not 0, the kernel uses a value rounded up to 2^n
1195 as the actual hash bucket size. 0 is a special value, meaning
1196 the child networking namespace will share the initial networking
1197 namespace's hash buckets.
1199 Note that the child will use the global one in case the kernel
1200 fails to allocate enough memory. In addition, the global hash
1201 buckets are spread over available NUMA nodes, but the allocation
1202 of the child hash table depends on the current process's NUMA
1203 policy, which could result in performance differences.
1205 Possible values: 0, 2^n (n: 7 (128) - 16 (64K))
1213 raw_l3mdev_accept - BOOLEAN
1214 Enabling this option allows a "global" bound socket to work
1215 across L3 master domains (e.g., VRFs) with packets capable of
1216 being received regardless of the L3 domain in which they
1217 originated. Only valid when the kernel was compiled with
1218 CONFIG_NET_L3_MASTER_DEV.
1220 Default: 1 (enabled)
1225 cipso_cache_enable - BOOLEAN
1226 If set, enable additions to and lookups from the CIPSO label mapping
1227 cache. If unset, additions are ignored and lookups always result in a
1228 miss. However, regardless of the setting the cache is still
1229 invalidated when required when means you can safely toggle this on and
1230 off and the cache will always be "safe".
1234 cipso_cache_bucket_size - INTEGER
1235 The CIPSO label cache consists of a fixed size hash table with each
1236 hash bucket containing a number of cache entries. This variable limits
1237 the number of entries in each hash bucket; the larger the value is, the
1238 more CIPSO label mappings that can be cached. When the number of
1239 entries in a given hash bucket reaches this limit adding new entries
1240 causes the oldest entry in the bucket to be removed to make room.
1244 cipso_rbm_optfmt - BOOLEAN
1245 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
1246 the CIPSO draft specification (see Documentation/netlabel for details).
1247 This means that when set the CIPSO tag will be padded with empty
1248 categories in order to make the packet data 32-bit aligned.
1252 cipso_rbm_structvalid - BOOLEAN
1253 If set, do a very strict check of the CIPSO option when
1254 ip_options_compile() is called. If unset, relax the checks done during
1255 ip_options_compile(). Either way is "safe" as errors are caught else
1256 where in the CIPSO processing code but setting this to 0 (False) should
1257 result in less work (i.e. it should be faster) but could cause problems
1258 with other implementations that require strict checking.
1265 ip_local_port_range - 2 INTEGERS
1266 Defines the local port range that is used by TCP and UDP to
1267 choose the local port. The first number is the first, the
1268 second the last local port number.
1269 If possible, it is better these numbers have different parity
1270 (one even and one odd value).
1271 Must be greater than or equal to ip_unprivileged_port_start.
1272 The default values are 32768 and 60999 respectively.
1274 ip_local_reserved_ports - list of comma separated ranges
1275 Specify the ports which are reserved for known third-party
1276 applications. These ports will not be used by automatic port
1277 assignments (e.g. when calling connect() or bind() with port
1278 number 0). Explicit port allocation behavior is unchanged.
1280 The format used for both input and output is a comma separated
1281 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
1282 10). Writing to the file will clear all previously reserved
1283 ports and update the current list with the one given in the
1286 Note that ip_local_port_range and ip_local_reserved_ports
1287 settings are independent and both are considered by the kernel
1288 when determining which ports are available for automatic port
1291 You can reserve ports which are not in the current
1292 ip_local_port_range, e.g.::
1294 $ cat /proc/sys/net/ipv4/ip_local_port_range
1296 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
1299 although this is redundant. However such a setting is useful
1300 if later the port range is changed to a value that will
1301 include the reserved ports. Also keep in mind, that overlapping
1302 of these ranges may affect probability of selecting ephemeral
1303 ports which are right after block of reserved ports.
1307 ip_unprivileged_port_start - INTEGER
1308 This is a per-namespace sysctl. It defines the first
1309 unprivileged port in the network namespace. Privileged ports
1310 require root or CAP_NET_BIND_SERVICE in order to bind to them.
1311 To disable all privileged ports, set this to 0. They must not
1312 overlap with the ip_local_port_range.
1316 ip_nonlocal_bind - BOOLEAN
1317 If set, allows processes to bind() to non-local IP addresses,
1318 which can be quite useful - but may break some applications.
1322 ip_autobind_reuse - BOOLEAN
1323 By default, bind() does not select the ports automatically even if
1324 the new socket and all sockets bound to the port have SO_REUSEADDR.
1325 ip_autobind_reuse allows bind() to reuse the port and this is useful
1326 when you use bind()+connect(), but may break some applications.
1327 The preferred solution is to use IP_BIND_ADDRESS_NO_PORT and this
1328 option should only be set by experts.
1331 ip_dynaddr - INTEGER
1332 If set non-zero, enables support for dynamic addresses.
1333 If set to a non-zero value larger than 1, a kernel log
1334 message will be printed when dynamic address rewriting
1339 ip_early_demux - BOOLEAN
1340 Optimize input packet processing down to one demux for
1341 certain kinds of local sockets. Currently we only do this
1342 for established TCP and connected UDP sockets.
1344 It may add an additional cost for pure routing workloads that
1345 reduces overall throughput, in such case you should disable it.
1349 ping_group_range - 2 INTEGERS
1350 Restrict ICMP_PROTO datagram sockets to users in the group range.
1351 The default is "1 0", meaning, that nobody (not even root) may
1352 create ping sockets. Setting it to "100 100" would grant permissions
1353 to the single group. "0 4294967295" would enable it for the world, "100
1354 4294967295" would enable it for the users, but not daemons.
1356 tcp_early_demux - BOOLEAN
1357 Enable early demux for established TCP sockets.
1361 udp_early_demux - BOOLEAN
1362 Enable early demux for connected UDP sockets. Disable this if
1363 your system could experience more unconnected load.
1367 icmp_echo_ignore_all - BOOLEAN
1368 If set non-zero, then the kernel will ignore all ICMP ECHO
1369 requests sent to it.
1373 icmp_echo_enable_probe - BOOLEAN
1374 If set to one, then the kernel will respond to RFC 8335 PROBE
1375 requests sent to it.
1379 icmp_echo_ignore_broadcasts - BOOLEAN
1380 If set non-zero, then the kernel will ignore all ICMP ECHO and
1381 TIMESTAMP requests sent to it via broadcast/multicast.
1385 icmp_ratelimit - INTEGER
1386 Limit the maximal rates for sending ICMP packets whose type matches
1387 icmp_ratemask (see below) to specific targets.
1388 0 to disable any limiting,
1389 otherwise the minimal space between responses in milliseconds.
1390 Note that another sysctl, icmp_msgs_per_sec limits the number
1391 of ICMP packets sent on all targets.
1395 icmp_msgs_per_sec - INTEGER
1396 Limit maximal number of ICMP packets sent per second from this host.
1397 Only messages whose type matches icmp_ratemask (see below) are
1398 controlled by this limit. For security reasons, the precise count
1399 of messages per second is randomized.
1403 icmp_msgs_burst - INTEGER
1404 icmp_msgs_per_sec controls number of ICMP packets sent per second,
1405 while icmp_msgs_burst controls the burst size of these packets.
1406 For security reasons, the precise burst size is randomized.
1410 icmp_ratemask - INTEGER
1411 Mask made of ICMP types for which rates are being limited.
1413 Significant bits: IHGFEDCBA9876543210
1415 Default mask: 0000001100000011000 (6168)
1417 Bit definitions (see include/linux/icmp.h):
1419 = =========================
1421 3 Destination Unreachable [1]_
1422 4 Source Quench [1]_
1425 B Time Exceeded [1]_
1426 C Parameter Problem [1]_
1431 H Address Mask Request
1432 I Address Mask Reply
1433 = =========================
1435 .. [1] These are rate limited by default (see default mask above)
1437 icmp_ignore_bogus_error_responses - BOOLEAN
1438 Some routers violate RFC1122 by sending bogus responses to broadcast
1439 frames. Such violations are normally logged via a kernel warning.
1440 If this is set to TRUE, the kernel will not give such warnings, which
1441 will avoid log file clutter.
1445 icmp_errors_use_inbound_ifaddr - BOOLEAN
1447 If zero, icmp error messages are sent with the primary address of
1448 the exiting interface.
1450 If non-zero, the message will be sent with the primary address of
1451 the interface that received the packet that caused the icmp error.
1452 This is the behaviour many network administrators will expect from
1453 a router. And it can make debugging complicated network layouts
1456 Note that if no primary address exists for the interface selected,
1457 then the primary address of the first non-loopback interface that
1458 has one will be used regardless of this setting.
1462 igmp_max_memberships - INTEGER
1463 Change the maximum number of multicast groups we can subscribe to.
1466 Theoretical maximum value is bounded by having to send a membership
1467 report in a single datagram (i.e. the report can't span multiple
1468 datagrams, or risk confusing the switch and leaving groups you don't
1471 The number of supported groups 'M' is bounded by the number of group
1472 report entries you can fit into a single datagram of 65535 bytes.
1474 M = 65536-sizeof (ip header)/(sizeof(Group record))
1476 Group records are variable length, with a minimum of 12 bytes.
1477 So net.ipv4.igmp_max_memberships should not be set higher than:
1479 (65536-24) / 12 = 5459
1481 The value 5459 assumes no IP header options, so in practice
1482 this number may be lower.
1484 igmp_max_msf - INTEGER
1485 Maximum number of addresses allowed in the source filter list for a
1491 Controls the IGMP query robustness variable (see RFC2236 8.1).
1493 Default: 2 (as specified by RFC2236 8.1)
1495 Minimum: 1 (as specified by RFC6636 4.5)
1497 force_igmp_version - INTEGER
1498 - 0 - (default) No enforcement of a IGMP version, IGMPv1/v2 fallback
1499 allowed. Will back to IGMPv3 mode again if all IGMPv1/v2 Querier
1500 Present timer expires.
1501 - 1 - Enforce to use IGMP version 1. Will also reply IGMPv1 report if
1502 receive IGMPv2/v3 query.
1503 - 2 - Enforce to use IGMP version 2. Will fallback to IGMPv1 if receive
1504 IGMPv1 query message. Will reply report if receive IGMPv3 query.
1505 - 3 - Enforce to use IGMP version 3. The same react with default 0.
1509 this is not the same with force_mld_version because IGMPv3 RFC3376
1510 Security Considerations does not have clear description that we could
1511 ignore other version messages completely as MLDv2 RFC3810. So make
1512 this value as default 0 is recommended.
1514 ``conf/interface/*``
1515 changes special settings per interface (where
1516 interface" is the name of your network interface)
1519 is special, changes the settings for all interfaces
1521 log_martians - BOOLEAN
1522 Log packets with impossible addresses to kernel log.
1523 log_martians for the interface will be enabled if at least one of
1524 conf/{all,interface}/log_martians is set to TRUE,
1525 it will be disabled otherwise
1527 accept_redirects - BOOLEAN
1528 Accept ICMP redirect messages.
1529 accept_redirects for the interface will be enabled if:
1531 - both conf/{all,interface}/accept_redirects are TRUE in the case
1532 forwarding for the interface is enabled
1536 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
1537 case forwarding for the interface is disabled
1539 accept_redirects for the interface will be disabled otherwise
1546 forwarding - BOOLEAN
1547 Enable IP forwarding on this interface. This controls whether packets
1548 received _on_ this interface can be forwarded.
1550 mc_forwarding - BOOLEAN
1551 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
1552 and a multicast routing daemon is required.
1553 conf/all/mc_forwarding must also be set to TRUE to enable multicast
1554 routing for the interface
1557 Integer value used to differentiate the devices by the medium they
1558 are attached to. Two devices can have different id values when
1559 the broadcast packets are received only on one of them.
1560 The default value 0 means that the device is the only interface
1561 to its medium, value of -1 means that medium is not known.
1563 Currently, it is used to change the proxy_arp behavior:
1564 the proxy_arp feature is enabled for packets forwarded between
1565 two devices attached to different media.
1570 proxy_arp for the interface will be enabled if at least one of
1571 conf/{all,interface}/proxy_arp is set to TRUE,
1572 it will be disabled otherwise
1574 proxy_arp_pvlan - BOOLEAN
1575 Private VLAN proxy arp.
1577 Basically allow proxy arp replies back to the same interface
1578 (from which the ARP request/solicitation was received).
1580 This is done to support (ethernet) switch features, like RFC
1581 3069, where the individual ports are NOT allowed to
1582 communicate with each other, but they are allowed to talk to
1583 the upstream router. As described in RFC 3069, it is possible
1584 to allow these hosts to communicate through the upstream
1585 router by proxy_arp'ing. Don't need to be used together with
1588 This technology is known by different names:
1590 In RFC 3069 it is called VLAN Aggregation.
1591 Cisco and Allied Telesyn call it Private VLAN.
1592 Hewlett-Packard call it Source-Port filtering or port-isolation.
1593 Ericsson call it MAC-Forced Forwarding (RFC Draft).
1595 proxy_delay - INTEGER
1596 Delay proxy response.
1598 Delay response to a neighbor solicitation when proxy_arp
1599 or proxy_ndp is enabled. A random value between [0, proxy_delay)
1600 will be chosen, setting to zero means reply with no delay.
1601 Value in jiffies. Defaults to 80.
1603 shared_media - BOOLEAN
1604 Send(router) or accept(host) RFC1620 shared media redirects.
1605 Overrides secure_redirects.
1607 shared_media for the interface will be enabled if at least one of
1608 conf/{all,interface}/shared_media is set to TRUE,
1609 it will be disabled otherwise
1613 secure_redirects - BOOLEAN
1614 Accept ICMP redirect messages only to gateways listed in the
1615 interface's current gateway list. Even if disabled, RFC1122 redirect
1618 Overridden by shared_media.
1620 secure_redirects for the interface will be enabled if at least one of
1621 conf/{all,interface}/secure_redirects is set to TRUE,
1622 it will be disabled otherwise
1626 send_redirects - BOOLEAN
1627 Send redirects, if router.
1629 send_redirects for the interface will be enabled if at least one of
1630 conf/{all,interface}/send_redirects is set to TRUE,
1631 it will be disabled otherwise
1635 bootp_relay - BOOLEAN
1636 Accept packets with source address 0.b.c.d destined
1637 not to this host as local ones. It is supposed, that
1638 BOOTP relay daemon will catch and forward such packets.
1639 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
1644 Not Implemented Yet.
1646 accept_source_route - BOOLEAN
1647 Accept packets with SRR option.
1648 conf/all/accept_source_route must also be set to TRUE to accept packets
1649 with SRR option on the interface
1656 accept_local - BOOLEAN
1657 Accept packets with local source addresses. In combination with
1658 suitable routing, this can be used to direct packets between two
1659 local interfaces over the wire and have them accepted properly.
1662 route_localnet - BOOLEAN
1663 Do not consider loopback addresses as martian source or destination
1664 while routing. This enables the use of 127/8 for local routing purposes.
1669 - 0 - No source validation.
1670 - 1 - Strict mode as defined in RFC3704 Strict Reverse Path
1671 Each incoming packet is tested against the FIB and if the interface
1672 is not the best reverse path the packet check will fail.
1673 By default failed packets are discarded.
1674 - 2 - Loose mode as defined in RFC3704 Loose Reverse Path
1675 Each incoming packet's source address is also tested against the FIB
1676 and if the source address is not reachable via any interface
1677 the packet check will fail.
1679 Current recommended practice in RFC3704 is to enable strict mode
1680 to prevent IP spoofing from DDos attacks. If using asymmetric routing
1681 or other complicated routing, then loose mode is recommended.
1683 The max value from conf/{all,interface}/rp_filter is used
1684 when doing source validation on the {interface}.
1686 Default value is 0. Note that some distributions enable it
1689 src_valid_mark - BOOLEAN
1690 - 0 - The fwmark of the packet is not included in reverse path
1691 route lookup. This allows for asymmetric routing configurations
1692 utilizing the fwmark in only one direction, e.g., transparent
1695 - 1 - The fwmark of the packet is included in reverse path route
1696 lookup. This permits rp_filter to function when the fwmark is
1697 used for routing traffic in both directions.
1699 This setting also affects the utilization of fmwark when
1700 performing source address selection for ICMP replies, or
1701 determining addresses stored for the IPOPT_TS_TSANDADDR and
1702 IPOPT_RR IP options.
1704 The max value from conf/{all,interface}/src_valid_mark is used.
1708 arp_filter - BOOLEAN
1709 - 1 - Allows you to have multiple network interfaces on the same
1710 subnet, and have the ARPs for each interface be answered
1711 based on whether or not the kernel would route a packet from
1712 the ARP'd IP out that interface (therefore you must use source
1713 based routing for this to work). In other words it allows control
1714 of which cards (usually 1) will respond to an arp request.
1716 - 0 - (default) The kernel can respond to arp requests with addresses
1717 from other interfaces. This may seem wrong but it usually makes
1718 sense, because it increases the chance of successful communication.
1719 IP addresses are owned by the complete host on Linux, not by
1720 particular interfaces. Only for more complex setups like load-
1721 balancing, does this behaviour cause problems.
1723 arp_filter for the interface will be enabled if at least one of
1724 conf/{all,interface}/arp_filter is set to TRUE,
1725 it will be disabled otherwise
1727 arp_announce - INTEGER
1728 Define different restriction levels for announcing the local
1729 source IP address from IP packets in ARP requests sent on
1732 - 0 - (default) Use any local address, configured on any interface
1733 - 1 - Try to avoid local addresses that are not in the target's
1734 subnet for this interface. This mode is useful when target
1735 hosts reachable via this interface require the source IP
1736 address in ARP requests to be part of their logical network
1737 configured on the receiving interface. When we generate the
1738 request we will check all our subnets that include the
1739 target IP and will preserve the source address if it is from
1740 such subnet. If there is no such subnet we select source
1741 address according to the rules for level 2.
1742 - 2 - Always use the best local address for this target.
1743 In this mode we ignore the source address in the IP packet
1744 and try to select local address that we prefer for talks with
1745 the target host. Such local address is selected by looking
1746 for primary IP addresses on all our subnets on the outgoing
1747 interface that include the target IP address. If no suitable
1748 local address is found we select the first local address
1749 we have on the outgoing interface or on all other interfaces,
1750 with the hope we will receive reply for our request and
1751 even sometimes no matter the source IP address we announce.
1753 The max value from conf/{all,interface}/arp_announce is used.
1755 Increasing the restriction level gives more chance for
1756 receiving answer from the resolved target while decreasing
1757 the level announces more valid sender's information.
1759 arp_ignore - INTEGER
1760 Define different modes for sending replies in response to
1761 received ARP requests that resolve local target IP addresses:
1763 - 0 - (default): reply for any local target IP address, configured
1765 - 1 - reply only if the target IP address is local address
1766 configured on the incoming interface
1767 - 2 - reply only if the target IP address is local address
1768 configured on the incoming interface and both with the
1769 sender's IP address are part from same subnet on this interface
1770 - 3 - do not reply for local addresses configured with scope host,
1771 only resolutions for global and link addresses are replied
1773 - 8 - do not reply for all local addresses
1775 The max value from conf/{all,interface}/arp_ignore is used
1776 when ARP request is received on the {interface}
1778 arp_notify - BOOLEAN
1779 Define mode for notification of address and device changes.
1781 == ==========================================================
1782 0 (default): do nothing
1783 1 Generate gratuitous arp requests when device is brought up
1784 or hardware address changes.
1785 == ==========================================================
1787 arp_accept - INTEGER
1788 Define behavior for accepting gratuitous ARP (garp) frames from devices
1789 that are not already present in the ARP table:
1791 - 0 - don't create new entries in the ARP table
1792 - 1 - create new entries in the ARP table
1793 - 2 - create new entries only if the source IP address is in the same
1794 subnet as an address configured on the interface that received the
1797 Both replies and requests type gratuitous arp will trigger the
1798 ARP table to be updated, if this setting is on.
1800 If the ARP table already contains the IP address of the
1801 gratuitous arp frame, the arp table will be updated regardless
1802 if this setting is on or off.
1804 arp_evict_nocarrier - BOOLEAN
1805 Clears the ARP cache on NOCARRIER events. This option is important for
1806 wireless devices where the ARP cache should not be cleared when roaming
1807 between access points on the same network. In most cases this should
1808 remain as the default (1).
1810 - 1 - (default): Clear the ARP cache on NOCARRIER events
1811 - 0 - Do not clear ARP cache on NOCARRIER events
1813 mcast_solicit - INTEGER
1814 The maximum number of multicast probes in INCOMPLETE state,
1815 when the associated hardware address is unknown. Defaults
1818 ucast_solicit - INTEGER
1819 The maximum number of unicast probes in PROBE state, when
1820 the hardware address is being reconfirmed. Defaults to 3.
1822 app_solicit - INTEGER
1823 The maximum number of probes to send to the user space ARP daemon
1824 via netlink before dropping back to multicast probes (see
1825 mcast_resolicit). Defaults to 0.
1827 mcast_resolicit - INTEGER
1828 The maximum number of multicast probes after unicast and
1829 app probes in PROBE state. Defaults to 0.
1831 disable_policy - BOOLEAN
1832 Disable IPSEC policy (SPD) for this interface
1834 disable_xfrm - BOOLEAN
1835 Disable IPSEC encryption on this interface, whatever the policy
1837 igmpv2_unsolicited_report_interval - INTEGER
1838 The interval in milliseconds in which the next unsolicited
1839 IGMPv1 or IGMPv2 report retransmit will take place.
1841 Default: 10000 (10 seconds)
1843 igmpv3_unsolicited_report_interval - INTEGER
1844 The interval in milliseconds in which the next unsolicited
1845 IGMPv3 report retransmit will take place.
1847 Default: 1000 (1 seconds)
1849 ignore_routes_with_linkdown - BOOLEAN
1850 Ignore routes whose link is down when performing a FIB lookup.
1852 promote_secondaries - BOOLEAN
1853 When a primary IP address is removed from this interface
1854 promote a corresponding secondary IP address instead of
1855 removing all the corresponding secondary IP addresses.
1857 drop_unicast_in_l2_multicast - BOOLEAN
1858 Drop any unicast IP packets that are received in link-layer
1859 multicast (or broadcast) frames.
1861 This behavior (for multicast) is actually a SHOULD in RFC
1862 1122, but is disabled by default for compatibility reasons.
1866 drop_gratuitous_arp - BOOLEAN
1867 Drop all gratuitous ARP frames, for example if there's a known
1868 good ARP proxy on the network and such frames need not be used
1869 (or in the case of 802.11, must not be used to prevent attacks.)
1875 Allows you to write a number, which can be used as required.
1879 xfrm4_gc_thresh - INTEGER
1880 (Obsolete since linux-4.14)
1881 The threshold at which we will start garbage collecting for IPv4
1882 destination cache entries. At twice this value the system will
1883 refuse new allocations.
1885 igmp_link_local_mcast_reports - BOOLEAN
1886 Enable IGMP reports for link local multicast groups in the
1892 kuznet@ms2.inr.ac.ru
1899 delon.nicolas@wanadoo.fr
1904 /proc/sys/net/ipv6/* Variables
1905 ==============================
1907 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1908 apply to IPv6 [XXX?].
1910 bindv6only - BOOLEAN
1911 Default value for IPV6_V6ONLY socket option,
1912 which restricts use of the IPv6 socket to IPv6 communication
1915 - TRUE: disable IPv4-mapped address feature
1916 - FALSE: enable IPv4-mapped address feature
1918 Default: FALSE (as specified in RFC3493)
1920 flowlabel_consistency - BOOLEAN
1921 Protect the consistency (and unicity) of flow label.
1922 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1930 auto_flowlabels - INTEGER
1931 Automatically generate flow labels based on a flow hash of the
1932 packet. This allows intermediate devices, such as routers, to
1933 identify packet flows for mechanisms like Equal Cost Multipath
1934 Routing (see RFC 6438).
1936 = ===========================================================
1937 0 automatic flow labels are completely disabled
1938 1 automatic flow labels are enabled by default, they can be
1939 disabled on a per socket basis using the IPV6_AUTOFLOWLABEL
1941 2 automatic flow labels are allowed, they may be enabled on a
1942 per socket basis using the IPV6_AUTOFLOWLABEL socket option
1943 3 automatic flow labels are enabled and enforced, they cannot
1944 be disabled by the socket option
1945 = ===========================================================
1949 flowlabel_state_ranges - BOOLEAN
1950 Split the flow label number space into two ranges. 0-0x7FFFF is
1951 reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF
1952 is reserved for stateless flow labels as described in RFC6437.
1959 flowlabel_reflect - INTEGER
1960 Control flow label reflection. Needed for Path MTU
1961 Discovery to work with Equal Cost Multipath Routing in anycast
1962 environments. See RFC 7690 and:
1963 https://tools.ietf.org/html/draft-wang-6man-flow-label-reflection-01
1967 - 1: enabled for established flows
1969 Note that this prevents automatic flowlabel changes, as done
1970 in "tcp: change IPv6 flow-label upon receiving spurious retransmission"
1971 and "tcp: Change txhash on every SYN and RTO retransmit"
1973 - 2: enabled for TCP RESET packets (no active listener)
1974 If set, a RST packet sent in response to a SYN packet on a closed
1975 port will reflect the incoming flow label.
1977 - 4: enabled for ICMPv6 echo reply messages.
1981 fib_multipath_hash_policy - INTEGER
1982 Controls which hash policy to use for multipath routes.
1984 Default: 0 (Layer 3)
1988 - 0 - Layer 3 (source and destination addresses plus flow label)
1989 - 1 - Layer 4 (standard 5-tuple)
1990 - 2 - Layer 3 or inner Layer 3 if present
1991 - 3 - Custom multipath hash. Fields used for multipath hash calculation
1992 are determined by fib_multipath_hash_fields sysctl
1994 fib_multipath_hash_fields - UNSIGNED INTEGER
1995 When fib_multipath_hash_policy is set to 3 (custom multipath hash), the
1996 fields used for multipath hash calculation are determined by this
1999 This value is a bitmask which enables various fields for multipath hash
2002 Possible fields are:
2004 ====== ============================
2005 0x0001 Source IP address
2006 0x0002 Destination IP address
2010 0x0020 Destination port
2011 0x0040 Inner source IP address
2012 0x0080 Inner destination IP address
2013 0x0100 Inner IP protocol
2014 0x0200 Inner Flow Label
2015 0x0400 Inner source port
2016 0x0800 Inner destination port
2017 ====== ============================
2019 Default: 0x0007 (source IP, destination IP and IP protocol)
2021 anycast_src_echo_reply - BOOLEAN
2022 Controls the use of anycast addresses as source addresses for ICMPv6
2030 idgen_delay - INTEGER
2031 Controls the delay in seconds after which time to retry
2032 privacy stable address generation if a DAD conflict is
2035 Default: 1 (as specified in RFC7217)
2037 idgen_retries - INTEGER
2038 Controls the number of retries to generate a stable privacy
2039 address if a DAD conflict is detected.
2041 Default: 3 (as specified in RFC7217)
2044 Controls the MLD query robustness variable (see RFC3810 9.1).
2046 Default: 2 (as specified by RFC3810 9.1)
2048 Minimum: 1 (as specified by RFC6636 4.5)
2050 max_dst_opts_number - INTEGER
2051 Maximum number of non-padding TLVs allowed in a Destination
2052 options extension header. If this value is less than zero
2053 then unknown options are disallowed and the number of known
2054 TLVs allowed is the absolute value of this number.
2058 max_hbh_opts_number - INTEGER
2059 Maximum number of non-padding TLVs allowed in a Hop-by-Hop
2060 options extension header. If this value is less than zero
2061 then unknown options are disallowed and the number of known
2062 TLVs allowed is the absolute value of this number.
2066 max_dst_opts_length - INTEGER
2067 Maximum length allowed for a Destination options extension
2070 Default: INT_MAX (unlimited)
2072 max_hbh_length - INTEGER
2073 Maximum length allowed for a Hop-by-Hop options extension
2076 Default: INT_MAX (unlimited)
2078 skip_notify_on_dev_down - BOOLEAN
2079 Controls whether an RTM_DELROUTE message is generated for routes
2080 removed when a device is taken down or deleted. IPv4 does not
2081 generate this message; IPv6 does by default. Setting this sysctl
2082 to true skips the message, making IPv4 and IPv6 on par in relying
2083 on userspace caches to track link events and evict routes.
2085 Default: false (generate message)
2087 nexthop_compat_mode - BOOLEAN
2088 New nexthop API provides a means for managing nexthops independent of
2089 prefixes. Backwards compatibility with old route format is enabled by
2090 default which means route dumps and notifications contain the new
2091 nexthop attribute but also the full, expanded nexthop definition.
2092 Further, updates or deletes of a nexthop configuration generate route
2093 notifications for each fib entry using the nexthop. Once a system
2094 understands the new API, this sysctl can be disabled to achieve full
2095 performance benefits of the new API by disabling the nexthop expansion
2096 and extraneous notifications.
2097 Default: true (backward compat mode)
2099 fib_notify_on_flag_change - INTEGER
2100 Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/
2101 RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed.
2103 After installing a route to the kernel, user space receives an
2104 acknowledgment, which means the route was installed in the kernel,
2105 but not necessarily in hardware.
2106 It is also possible for a route already installed in hardware to change
2107 its action and therefore its flags. For example, a host route that is
2108 trapping packets can be "promoted" to perform decapsulation following
2109 the installation of an IPinIP/VXLAN tunnel.
2110 The notifications will indicate to user-space the state of the route.
2112 Default: 0 (Do not emit notifications.)
2116 - 0 - Do not emit notifications.
2117 - 1 - Emit notifications.
2118 - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change.
2121 Define the IOAM id of this node. Uses only 24 bits out of 32 in total.
2128 ioam6_id_wide - LONG INTEGER
2129 Define the wide IOAM id of this node. Uses only 56 bits out of 64 in
2130 total. Can be different from ioam6_id.
2133 Max: 0xFFFFFFFFFFFFFF
2135 Default: 0xFFFFFFFFFFFFFF
2139 ip6frag_high_thresh - INTEGER
2140 Maximum memory used to reassemble IPv6 fragments. When
2141 ip6frag_high_thresh bytes of memory is allocated for this purpose,
2142 the fragment handler will toss packets until ip6frag_low_thresh
2145 ip6frag_low_thresh - INTEGER
2146 See ip6frag_high_thresh
2148 ip6frag_time - INTEGER
2149 Time in seconds to keep an IPv6 fragment in memory.
2152 Change the interface-specific default settings.
2154 These settings would be used during creating new interfaces.
2158 Change all the interface-specific settings.
2160 [XXX: Other special features than forwarding?]
2162 conf/all/disable_ipv6 - BOOLEAN
2163 Changing this value is same as changing ``conf/default/disable_ipv6``
2164 setting and also all per-interface ``disable_ipv6`` settings to the same
2167 Reading this value does not have any particular meaning. It does not say
2168 whether IPv6 support is enabled or disabled. Returned value can be 1
2169 also in the case when some interface has ``disable_ipv6`` set to 0 and
2170 has configured IPv6 addresses.
2172 conf/all/forwarding - BOOLEAN
2173 Enable global IPv6 forwarding between all interfaces.
2175 IPv4 and IPv6 work differently here; e.g. netfilter must be used
2176 to control which interfaces may forward packets and which not.
2178 This also sets all interfaces' Host/Router setting
2179 'forwarding' to the specified value. See below for details.
2181 This referred to as global forwarding.
2186 fwmark_reflect - BOOLEAN
2187 Controls the fwmark of kernel-generated IPv6 reply packets that are not
2188 associated with a socket for example, TCP RSTs or ICMPv6 echo replies).
2189 If unset, these packets have a fwmark of zero. If set, they have the
2190 fwmark of the packet they are replying to.
2194 ``conf/interface/*``:
2195 Change special settings per interface.
2197 The functional behaviour for certain settings is different
2198 depending on whether local forwarding is enabled or not.
2201 Accept Router Advertisements; autoconfigure using them.
2203 It also determines whether or not to transmit Router
2204 Solicitations. If and only if the functional setting is to
2205 accept Router Advertisements, Router Solicitations will be
2208 Possible values are:
2210 == ===========================================================
2211 0 Do not accept Router Advertisements.
2212 1 Accept Router Advertisements if forwarding is disabled.
2213 2 Overrule forwarding behaviour. Accept Router Advertisements
2214 even if forwarding is enabled.
2215 == ===========================================================
2219 - enabled if local forwarding is disabled.
2220 - disabled if local forwarding is enabled.
2222 accept_ra_defrtr - BOOLEAN
2223 Learn default router in Router Advertisement.
2227 - enabled if accept_ra is enabled.
2228 - disabled if accept_ra is disabled.
2230 ra_defrtr_metric - UNSIGNED INTEGER
2231 Route metric for default route learned in Router Advertisement. This value
2232 will be assigned as metric for the default route learned via IPv6 Router
2233 Advertisement. Takes affect only if accept_ra_defrtr is enabled.
2238 Default: IP6_RT_PRIO_USER i.e. 1024.
2240 accept_ra_from_local - BOOLEAN
2241 Accept RA with source-address that is found on local machine
2242 if the RA is otherwise proper and able to be accepted.
2244 Default is to NOT accept these as it may be an un-intended
2249 - enabled if accept_ra_from_local is enabled
2250 on a specific interface.
2251 - disabled if accept_ra_from_local is disabled
2252 on a specific interface.
2254 accept_ra_min_hop_limit - INTEGER
2255 Minimum hop limit Information in Router Advertisement.
2257 Hop limit Information in Router Advertisement less than this
2258 variable shall be ignored.
2262 accept_ra_pinfo - BOOLEAN
2263 Learn Prefix Information in Router Advertisement.
2267 - enabled if accept_ra is enabled.
2268 - disabled if accept_ra is disabled.
2270 accept_ra_rt_info_min_plen - INTEGER
2271 Minimum prefix length of Route Information in RA.
2273 Route Information w/ prefix smaller than this variable shall
2278 * 0 if accept_ra_rtr_pref is enabled.
2279 * -1 if accept_ra_rtr_pref is disabled.
2281 accept_ra_rt_info_max_plen - INTEGER
2282 Maximum prefix length of Route Information in RA.
2284 Route Information w/ prefix larger than this variable shall
2289 * 0 if accept_ra_rtr_pref is enabled.
2290 * -1 if accept_ra_rtr_pref is disabled.
2292 accept_ra_rtr_pref - BOOLEAN
2293 Accept Router Preference in RA.
2297 - enabled if accept_ra is enabled.
2298 - disabled if accept_ra is disabled.
2300 accept_ra_mtu - BOOLEAN
2301 Apply the MTU value specified in RA option 5 (RFC4861). If
2302 disabled, the MTU specified in the RA will be ignored.
2306 - enabled if accept_ra is enabled.
2307 - disabled if accept_ra is disabled.
2309 accept_redirects - BOOLEAN
2314 - enabled if local forwarding is disabled.
2315 - disabled if local forwarding is enabled.
2317 accept_source_route - INTEGER
2318 Accept source routing (routing extension header).
2320 - >= 0: Accept only routing header type 2.
2321 - < 0: Do not accept routing header.
2326 Autoconfigure addresses using Prefix Information in Router
2331 - enabled if accept_ra_pinfo is enabled.
2332 - disabled if accept_ra_pinfo is disabled.
2334 dad_transmits - INTEGER
2335 The amount of Duplicate Address Detection probes to send.
2339 forwarding - INTEGER
2340 Configure interface-specific Host/Router behaviour.
2344 It is recommended to have the same setting on all
2345 interfaces; mixed router/host scenarios are rather uncommon.
2347 Possible values are:
2349 - 0 Forwarding disabled
2350 - 1 Forwarding enabled
2354 By default, Host behaviour is assumed. This means:
2356 1. IsRouter flag is not set in Neighbour Advertisements.
2357 2. If accept_ra is TRUE (default), transmit Router
2359 3. If accept_ra is TRUE (default), accept Router
2360 Advertisements (and do autoconfiguration).
2361 4. If accept_redirects is TRUE (default), accept Redirects.
2365 If local forwarding is enabled, Router behaviour is assumed.
2366 This means exactly the reverse from the above:
2368 1. IsRouter flag is set in Neighbour Advertisements.
2369 2. Router Solicitations are not sent unless accept_ra is 2.
2370 3. Router Advertisements are ignored unless accept_ra is 2.
2371 4. Redirects are ignored.
2373 Default: 0 (disabled) if global forwarding is disabled (default),
2374 otherwise 1 (enabled).
2377 Default Hop Limit to set.
2382 Default Maximum Transfer Unit
2384 Default: 1280 (IPv6 required minimum)
2386 ip_nonlocal_bind - BOOLEAN
2387 If set, allows processes to bind() to non-local IPv6 addresses,
2388 which can be quite useful - but may break some applications.
2392 router_probe_interval - INTEGER
2393 Minimum interval (in seconds) between Router Probing described
2398 router_solicitation_delay - INTEGER
2399 Number of seconds to wait after interface is brought up
2400 before sending Router Solicitations.
2404 router_solicitation_interval - INTEGER
2405 Number of seconds to wait between Router Solicitations.
2409 router_solicitations - INTEGER
2410 Number of Router Solicitations to send until assuming no
2411 routers are present.
2415 use_oif_addrs_only - BOOLEAN
2416 When enabled, the candidate source addresses for destinations
2417 routed via this interface are restricted to the set of addresses
2418 configured on this interface (vis. RFC 6724, section 4).
2422 use_tempaddr - INTEGER
2423 Preference for Privacy Extensions (RFC3041).
2425 * <= 0 : disable Privacy Extensions
2426 * == 1 : enable Privacy Extensions, but prefer public
2427 addresses over temporary addresses.
2428 * > 1 : enable Privacy Extensions and prefer temporary
2429 addresses over public addresses.
2433 * 0 (for most devices)
2434 * -1 (for point-to-point devices and loopback devices)
2436 temp_valid_lft - INTEGER
2437 valid lifetime (in seconds) for temporary addresses.
2439 Default: 172800 (2 days)
2441 temp_prefered_lft - INTEGER
2442 Preferred lifetime (in seconds) for temporary addresses.
2444 Default: 86400 (1 day)
2446 keep_addr_on_down - INTEGER
2447 Keep all IPv6 addresses on an interface down event. If set static
2448 global addresses with no expiration time are not flushed.
2451 * 0 : system default
2454 Default: 0 (addresses are removed)
2456 max_desync_factor - INTEGER
2457 Maximum value for DESYNC_FACTOR, which is a random value
2458 that ensures that clients don't synchronize with each
2459 other and generate new addresses at exactly the same time.
2460 value is in seconds.
2464 regen_max_retry - INTEGER
2465 Number of attempts before give up attempting to generate
2466 valid temporary addresses.
2470 max_addresses - INTEGER
2471 Maximum number of autoconfigured addresses per interface. Setting
2472 to zero disables the limitation. It is not recommended to set this
2473 value too large (or to zero) because it would be an easy way to
2474 crash the kernel by allowing too many addresses to be created.
2478 disable_ipv6 - BOOLEAN
2479 Disable IPv6 operation. If accept_dad is set to 2, this value
2480 will be dynamically set to TRUE if DAD fails for the link-local
2483 Default: FALSE (enable IPv6 operation)
2485 When this value is changed from 1 to 0 (IPv6 is being enabled),
2486 it will dynamically create a link-local address on the given
2487 interface and start Duplicate Address Detection, if necessary.
2489 When this value is changed from 0 to 1 (IPv6 is being disabled),
2490 it will dynamically delete all addresses and routes on the given
2491 interface. From now on it will not possible to add addresses/routes
2492 to the selected interface.
2494 accept_dad - INTEGER
2495 Whether to accept DAD (Duplicate Address Detection).
2497 == ==============================================================
2499 1 Enable DAD (default)
2500 2 Enable DAD, and disable IPv6 operation if MAC-based duplicate
2501 link-local address has been found.
2502 == ==============================================================
2504 DAD operation and mode on a given interface will be selected according
2505 to the maximum value of conf/{all,interface}/accept_dad.
2507 force_tllao - BOOLEAN
2508 Enable sending the target link-layer address option even when
2509 responding to a unicast neighbor solicitation.
2513 Quoting from RFC 2461, section 4.4, Target link-layer address:
2515 "The option MUST be included for multicast solicitations in order to
2516 avoid infinite Neighbor Solicitation "recursion" when the peer node
2517 does not have a cache entry to return a Neighbor Advertisements
2518 message. When responding to unicast solicitations, the option can be
2519 omitted since the sender of the solicitation has the correct link-
2520 layer address; otherwise it would not have be able to send the unicast
2521 solicitation in the first place. However, including the link-layer
2522 address in this case adds little overhead and eliminates a potential
2523 race condition where the sender deletes the cached link-layer address
2524 prior to receiving a response to a previous solicitation."
2526 ndisc_notify - BOOLEAN
2527 Define mode for notification of address and device changes.
2529 * 0 - (default): do nothing
2530 * 1 - Generate unsolicited neighbour advertisements when device is brought
2531 up or hardware address changes.
2533 ndisc_tclass - INTEGER
2534 The IPv6 Traffic Class to use by default when sending IPv6 Neighbor
2535 Discovery (Router Solicitation, Router Advertisement, Neighbor
2536 Solicitation, Neighbor Advertisement, Redirect) messages.
2537 These 8 bits can be interpreted as 6 high order bits holding the DSCP
2538 value and 2 low order bits representing ECN (which you probably want
2543 ndisc_evict_nocarrier - BOOLEAN
2544 Clears the neighbor discovery table on NOCARRIER events. This option is
2545 important for wireless devices where the neighbor discovery cache should
2546 not be cleared when roaming between access points on the same network.
2547 In most cases this should remain as the default (1).
2549 - 1 - (default): Clear neighbor discover cache on NOCARRIER events.
2550 - 0 - Do not clear neighbor discovery cache on NOCARRIER events.
2552 mldv1_unsolicited_report_interval - INTEGER
2553 The interval in milliseconds in which the next unsolicited
2554 MLDv1 report retransmit will take place.
2556 Default: 10000 (10 seconds)
2558 mldv2_unsolicited_report_interval - INTEGER
2559 The interval in milliseconds in which the next unsolicited
2560 MLDv2 report retransmit will take place.
2562 Default: 1000 (1 second)
2564 force_mld_version - INTEGER
2565 * 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
2566 * 1 - Enforce to use MLD version 1
2567 * 2 - Enforce to use MLD version 2
2569 suppress_frag_ndisc - INTEGER
2570 Control RFC 6980 (Security Implications of IPv6 Fragmentation
2571 with IPv6 Neighbor Discovery) behavior:
2573 * 1 - (default) discard fragmented neighbor discovery packets
2574 * 0 - allow fragmented neighbor discovery packets
2576 optimistic_dad - BOOLEAN
2577 Whether to perform Optimistic Duplicate Address Detection (RFC 4429).
2579 * 0: disabled (default)
2582 Optimistic Duplicate Address Detection for the interface will be enabled
2583 if at least one of conf/{all,interface}/optimistic_dad is set to 1,
2584 it will be disabled otherwise.
2586 use_optimistic - BOOLEAN
2587 If enabled, do not classify optimistic addresses as deprecated during
2588 source address selection. Preferred addresses will still be chosen
2589 before optimistic addresses, subject to other ranking in the source
2590 address selection algorithm.
2592 * 0: disabled (default)
2595 This will be enabled if at least one of
2596 conf/{all,interface}/use_optimistic is set to 1, disabled otherwise.
2598 stable_secret - IPv6 address
2599 This IPv6 address will be used as a secret to generate IPv6
2600 addresses for link-local addresses and autoconfigured
2601 ones. All addresses generated after setting this secret will
2602 be stable privacy ones by default. This can be changed via the
2603 addrgenmode ip-link. conf/default/stable_secret is used as the
2604 secret for the namespace, the interface specific ones can
2605 overwrite that. Writes to conf/all/stable_secret are refused.
2607 It is recommended to generate this secret during installation
2608 of a system and keep it stable after that.
2610 By default the stable secret is unset.
2612 addr_gen_mode - INTEGER
2613 Defines how link-local and autoconf addresses are generated.
2615 = =================================================================
2616 0 generate address based on EUI64 (default)
2617 1 do no generate a link-local address, use EUI64 for addresses
2618 generated from autoconf
2619 2 generate stable privacy addresses, using the secret from
2620 stable_secret (RFC7217)
2621 3 generate stable privacy addresses, using a random secret if unset
2622 = =================================================================
2624 drop_unicast_in_l2_multicast - BOOLEAN
2625 Drop any unicast IPv6 packets that are received in link-layer
2626 multicast (or broadcast) frames.
2628 By default this is turned off.
2630 drop_unsolicited_na - BOOLEAN
2631 Drop all unsolicited neighbor advertisements, for example if there's
2632 a known good NA proxy on the network and such frames need not be used
2633 (or in the case of 802.11, must not be used to prevent attacks.)
2635 By default this is turned off.
2637 accept_untracked_na - INTEGER
2638 Define behavior for accepting neighbor advertisements from devices that
2639 are absent in the neighbor cache:
2641 - 0 - (default) Do not accept unsolicited and untracked neighbor
2644 - 1 - Add a new neighbor cache entry in STALE state for routers on
2645 receiving a neighbor advertisement (either solicited or unsolicited)
2646 with target link-layer address option specified if no neighbor entry
2647 is already present for the advertised IPv6 address. Without this knob,
2648 NAs received for untracked addresses (absent in neighbor cache) are
2651 This is as per router-side behavior documented in RFC9131.
2653 This has lower precedence than drop_unsolicited_na.
2655 This will optimize the return path for the initial off-link
2656 communication that is initiated by a directly connected host, by
2657 ensuring that the first-hop router which turns on this setting doesn't
2658 have to buffer the initial return packets to do neighbor-solicitation.
2659 The prerequisite is that the host is configured to send unsolicited
2660 neighbor advertisements on interface bringup. This setting should be
2661 used in conjunction with the ndisc_notify setting on the host to
2662 satisfy this prerequisite.
2664 - 2 - Extend option (1) to add a new neighbor cache entry only if the
2665 source IP address is in the same subnet as an address configured on
2666 the interface that received the neighbor advertisement.
2668 enhanced_dad - BOOLEAN
2669 Include a nonce option in the IPv6 neighbor solicitation messages used for
2670 duplicate address detection per RFC7527. A received DAD NS will only signal
2671 a duplicate address if the nonce is different. This avoids any false
2672 detection of duplicates due to loopback of the NS messages that we send.
2673 The nonce option will be sent on an interface unless both of
2674 conf/{all,interface}/enhanced_dad are set to FALSE.
2682 Limit the maximal rates for sending ICMPv6 messages.
2684 0 to disable any limiting,
2685 otherwise the minimal space between responses in milliseconds.
2689 ratemask - list of comma separated ranges
2690 For ICMPv6 message types matching the ranges in the ratemask, limit
2691 the sending of the message according to ratelimit parameter.
2693 The format used for both input and output is a comma separated
2694 list of ranges (e.g. "0-127,129" for ICMPv6 message type 0 to 127 and
2695 129). Writing to the file will clear all previous ranges of ICMPv6
2696 message types and update the current list with the input.
2698 Refer to: https://www.iana.org/assignments/icmpv6-parameters/icmpv6-parameters.xhtml
2699 for numerical values of ICMPv6 message types, e.g. echo request is 128
2700 and echo reply is 129.
2702 Default: 0-1,3-127 (rate limit ICMPv6 errors except Packet Too Big)
2704 echo_ignore_all - BOOLEAN
2705 If set non-zero, then the kernel will ignore all ICMP ECHO
2706 requests sent to it over the IPv6 protocol.
2710 echo_ignore_multicast - BOOLEAN
2711 If set non-zero, then the kernel will ignore all ICMP ECHO
2712 requests sent to it over the IPv6 protocol via multicast.
2716 echo_ignore_anycast - BOOLEAN
2717 If set non-zero, then the kernel will ignore all ICMP ECHO
2718 requests sent to it over the IPv6 protocol destined to anycast address.
2722 xfrm6_gc_thresh - INTEGER
2723 (Obsolete since linux-4.14)
2724 The threshold at which we will start garbage collecting for IPv6
2725 destination cache entries. At twice this value the system will
2726 refuse new allocations.
2730 Pekka Savola <pekkas@netcore.fi>
2731 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
2734 /proc/sys/net/bridge/* Variables:
2735 =================================
2737 bridge-nf-call-arptables - BOOLEAN
2738 - 1 : pass bridged ARP traffic to arptables' FORWARD chain.
2743 bridge-nf-call-iptables - BOOLEAN
2744 - 1 : pass bridged IPv4 traffic to iptables' chains.
2749 bridge-nf-call-ip6tables - BOOLEAN
2750 - 1 : pass bridged IPv6 traffic to ip6tables' chains.
2755 bridge-nf-filter-vlan-tagged - BOOLEAN
2756 - 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
2761 bridge-nf-filter-pppoe-tagged - BOOLEAN
2762 - 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
2767 bridge-nf-pass-vlan-input-dev - BOOLEAN
2768 - 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
2769 interface on the bridge and set the netfilter input device to the
2770 vlan. This allows use of e.g. "iptables -i br0.1" and makes the
2771 REDIRECT target work with vlan-on-top-of-bridge interfaces. When no
2772 matching vlan interface is found, or this switch is off, the input
2773 device is set to the bridge interface.
2775 - 0: disable bridge netfilter vlan interface lookup.
2779 ``proc/sys/net/sctp/*`` Variables:
2780 ==================================
2782 addip_enable - BOOLEAN
2783 Enable or disable extension of Dynamic Address Reconfiguration
2784 (ADD-IP) functionality specified in RFC5061. This extension provides
2785 the ability to dynamically add and remove new addresses for the SCTP
2788 1: Enable extension.
2790 0: Disable extension.
2795 Enable or disable pf (pf is short for potentially failed) state. A value
2796 of pf_retrans > path_max_retrans also disables pf state. That is, one of
2797 both pf_enable and pf_retrans > path_max_retrans can disable pf state.
2798 Since pf_retrans and path_max_retrans can be changed by userspace
2799 application, sometimes user expects to disable pf state by the value of
2800 pf_retrans > path_max_retrans, but occasionally the value of pf_retrans
2801 or path_max_retrans is changed by the user application, this pf state is
2802 enabled. As such, it is necessary to add this to dynamically enable
2803 and disable pf state. See:
2804 https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for
2814 Unset or enable/disable pf (pf is short for potentially failed) state
2815 exposure. Applications can control the exposure of the PF path state
2816 in the SCTP_PEER_ADDR_CHANGE event and the SCTP_GET_PEER_ADDR_INFO
2817 sockopt. When it's unset, no SCTP_PEER_ADDR_CHANGE event with
2818 SCTP_ADDR_PF state will be sent and a SCTP_PF-state transport info
2819 can be got via SCTP_GET_PEER_ADDR_INFO sockopt; When it's enabled,
2820 a SCTP_PEER_ADDR_CHANGE event will be sent for a transport becoming
2821 SCTP_PF state and a SCTP_PF-state transport info can be got via
2822 SCTP_GET_PEER_ADDR_INFO sockopt; When it's disabled, no
2823 SCTP_PEER_ADDR_CHANGE event will be sent and it returns -EACCES when
2824 trying to get a SCTP_PF-state transport info via SCTP_GET_PEER_ADDR_INFO
2827 0: Unset pf state exposure, Compatible with old applications.
2829 1: Disable pf state exposure.
2831 2: Enable pf state exposure.
2835 addip_noauth_enable - BOOLEAN
2836 Dynamic Address Reconfiguration (ADD-IP) requires the use of
2837 authentication to protect the operations of adding or removing new
2838 addresses. This requirement is mandated so that unauthorized hosts
2839 would not be able to hijack associations. However, older
2840 implementations may not have implemented this requirement while
2841 allowing the ADD-IP extension. For reasons of interoperability,
2842 we provide this variable to control the enforcement of the
2843 authentication requirement.
2845 == ===============================================================
2846 1 Allow ADD-IP extension to be used without authentication. This
2847 should only be set in a closed environment for interoperability
2848 with older implementations.
2850 0 Enforce the authentication requirement
2851 == ===============================================================
2855 auth_enable - BOOLEAN
2856 Enable or disable Authenticated Chunks extension. This extension
2857 provides the ability to send and receive authenticated chunks and is
2858 required for secure operation of Dynamic Address Reconfiguration
2861 - 1: Enable this extension.
2862 - 0: Disable this extension.
2866 prsctp_enable - BOOLEAN
2867 Enable or disable the Partial Reliability extension (RFC3758) which
2868 is used to notify peers that a given DATA should no longer be expected.
2870 - 1: Enable extension
2876 The limit of the number of new packets that can be initially sent. It
2877 controls how bursty the generated traffic can be.
2881 association_max_retrans - INTEGER
2882 Set the maximum number for retransmissions that an association can
2883 attempt deciding that the remote end is unreachable. If this value
2884 is exceeded, the association is terminated.
2888 max_init_retransmits - INTEGER
2889 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
2890 that an association will attempt before declaring the destination
2891 unreachable and terminating.
2895 path_max_retrans - INTEGER
2896 The maximum number of retransmissions that will be attempted on a given
2897 path. Once this threshold is exceeded, the path is considered
2898 unreachable, and new traffic will use a different path when the
2899 association is multihomed.
2903 pf_retrans - INTEGER
2904 The number of retransmissions that will be attempted on a given path
2905 before traffic is redirected to an alternate transport (should one
2906 exist). Note this is distinct from path_max_retrans, as a path that
2907 passes the pf_retrans threshold can still be used. Its only
2908 deprioritized when a transmission path is selected by the stack. This
2909 setting is primarily used to enable fast failover mechanisms without
2910 having to reduce path_max_retrans to a very low value. See:
2911 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
2912 for details. Note also that a value of pf_retrans > path_max_retrans
2913 disables this feature. Since both pf_retrans and path_max_retrans can
2914 be changed by userspace application, a variable pf_enable is used to
2919 ps_retrans - INTEGER
2920 Primary.Switchover.Max.Retrans (PSMR), it's a tunable parameter coming
2921 from section-5 "Primary Path Switchover" in rfc7829. The primary path
2922 will be changed to another active path when the path error counter on
2923 the old primary path exceeds PSMR, so that "the SCTP sender is allowed
2924 to continue data transmission on a new working path even when the old
2925 primary destination address becomes active again". Note this feature
2926 is disabled by initializing 'ps_retrans' per netns as 0xffff by default,
2927 and its value can't be less than 'pf_retrans' when changing by sysctl.
2931 rto_initial - INTEGER
2932 The initial round trip timeout value in milliseconds that will be used
2933 in calculating round trip times. This is the initial time interval
2934 for retransmissions.
2939 The maximum value (in milliseconds) of the round trip timeout. This
2940 is the largest time interval that can elapse between retransmissions.
2945 The minimum value (in milliseconds) of the round trip timeout. This
2946 is the smallest time interval the can elapse between retransmissions.
2950 hb_interval - INTEGER
2951 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
2952 are sent at the specified interval on idle paths to probe the state of
2953 a given path between 2 associations.
2957 sack_timeout - INTEGER
2958 The amount of time (in milliseconds) that the implementation will wait
2963 valid_cookie_life - INTEGER
2964 The default lifetime of the SCTP cookie (in milliseconds). The cookie
2965 is used during association establishment.
2969 cookie_preserve_enable - BOOLEAN
2970 Enable or disable the ability to extend the lifetime of the SCTP cookie
2971 that is used during the establishment phase of SCTP association
2973 - 1: Enable cookie lifetime extension.
2978 cookie_hmac_alg - STRING
2979 Select the hmac algorithm used when generating the cookie value sent by
2980 a listening sctp socket to a connecting client in the INIT-ACK chunk.
2987 Ability to assign md5 or sha1 as the selected alg is predicated on the
2988 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
2989 CONFIG_CRYPTO_SHA1).
2991 Default: Dependent on configuration. MD5 if available, else SHA1 if
2992 available, else none.
2994 rcvbuf_policy - INTEGER
2995 Determines if the receive buffer is attributed to the socket or to
2996 association. SCTP supports the capability to create multiple
2997 associations on a single socket. When using this capability, it is
2998 possible that a single stalled association that's buffering a lot
2999 of data may block other associations from delivering their data by
3000 consuming all of the receive buffer space. To work around this,
3001 the rcvbuf_policy could be set to attribute the receiver buffer space
3002 to each association instead of the socket. This prevents the described
3005 - 1: rcvbuf space is per association
3006 - 0: rcvbuf space is per socket
3010 sndbuf_policy - INTEGER
3011 Similar to rcvbuf_policy above, this applies to send buffer space.
3013 - 1: Send buffer is tracked per association
3014 - 0: Send buffer is tracked per socket.
3018 sctp_mem - vector of 3 INTEGERs: min, pressure, max
3019 Number of pages allowed for queueing by all SCTP sockets.
3021 min: Below this number of pages SCTP is not bothered about its
3022 memory appetite. When amount of memory allocated by SCTP exceeds
3023 this number, SCTP starts to moderate memory usage.
3025 pressure: This value was introduced to follow format of tcp_mem.
3027 max: Number of pages allowed for queueing by all SCTP sockets.
3029 Default is calculated at boot time from amount of available memory.
3031 sctp_rmem - vector of 3 INTEGERs: min, default, max
3032 Only the first value ("min") is used, "default" and "max" are
3035 min: Minimal size of receive buffer used by SCTP socket.
3036 It is guaranteed to each SCTP socket (but not association) even
3037 under moderate memory pressure.
3041 sctp_wmem - vector of 3 INTEGERs: min, default, max
3042 Only the first value ("min") is used, "default" and "max" are
3045 min: Minimum size of send buffer that can be used by SCTP sockets.
3046 It is guaranteed to each SCTP socket (but not association) even
3047 under moderate memory pressure.
3051 addr_scope_policy - INTEGER
3052 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
3054 - 0 - Disable IPv4 address scoping
3055 - 1 - Enable IPv4 address scoping
3056 - 2 - Follow draft but allow IPv4 private addresses
3057 - 3 - Follow draft but allow IPv4 link local addresses
3062 The listening port for the local UDP tunneling sock. Normally it's
3063 using the IANA-assigned UDP port number 9899 (sctp-tunneling).
3065 This UDP sock is used for processing the incoming UDP-encapsulated
3066 SCTP packets (from RFC6951), and shared by all applications in the
3067 same net namespace. This UDP sock will be closed when the value is
3070 The value will also be used to set the src port of the UDP header
3071 for the outgoing UDP-encapsulated SCTP packets. For the dest port,
3072 please refer to 'encap_port' below.
3076 encap_port - INTEGER
3077 The default remote UDP encapsulation port.
3079 This value is used to set the dest port of the UDP header for the
3080 outgoing UDP-encapsulated SCTP packets by default. Users can also
3081 change the value for each sock/asoc/transport by using setsockopt.
3082 For further information, please refer to RFC6951.
3084 Note that when connecting to a remote server, the client should set
3085 this to the port that the UDP tunneling sock on the peer server is
3086 listening to and the local UDP tunneling sock on the client also
3087 must be started. On the server, it would get the encap_port from
3088 the incoming packet's source port.
3092 plpmtud_probe_interval - INTEGER
3093 The time interval (in milliseconds) for the PLPMTUD probe timer,
3094 which is configured to expire after this period to receive an
3095 acknowledgment to a probe packet. This is also the time interval
3096 between the probes for the current pmtu when the probe search
3099 PLPMTUD will be disabled when 0 is set, and other values for it
3104 reconf_enable - BOOLEAN
3105 Enable or disable extension of Stream Reconfiguration functionality
3106 specified in RFC6525. This extension provides the ability to "reset"
3107 a stream, and it includes the Parameters of "Outgoing/Incoming SSN
3108 Reset", "SSN/TSN Reset" and "Add Outgoing/Incoming Streams".
3110 - 1: Enable extension.
3111 - 0: Disable extension.
3115 intl_enable - BOOLEAN
3116 Enable or disable extension of User Message Interleaving functionality
3117 specified in RFC8260. This extension allows the interleaving of user
3118 messages sent on different streams. With this feature enabled, I-DATA
3119 chunk will replace DATA chunk to carry user messages if also supported
3120 by the peer. Note that to use this feature, one needs to set this option
3121 to 1 and also needs to set socket options SCTP_FRAGMENT_INTERLEAVE to 2
3122 and SCTP_INTERLEAVING_SUPPORTED to 1.
3124 - 1: Enable extension.
3125 - 0: Disable extension.
3129 ecn_enable - BOOLEAN
3130 Control use of Explicit Congestion Notification (ECN) by SCTP.
3131 Like in TCP, ECN is used only when both ends of the SCTP connection
3132 indicate support for it. This feature is useful in avoiding losses
3133 due to congestion by allowing supporting routers to signal congestion
3134 before having to drop packets.
3141 l3mdev_accept - BOOLEAN
3142 Enabling this option allows a "global" bound socket to work
3143 across L3 master domains (e.g., VRFs) with packets capable of
3144 being received regardless of the L3 domain in which they
3145 originated. Only valid when the kernel was compiled with
3146 CONFIG_NET_L3_MASTER_DEV.
3148 Default: 1 (enabled)
3151 ``/proc/sys/net/core/*``
3152 ========================
3154 Please see: Documentation/admin-guide/sysctl/net.rst for descriptions of these entries.
3157 ``/proc/sys/net/unix/*``
3158 ========================
3160 max_dgram_qlen - INTEGER
3161 The maximum length of dgram socket receive queue