1 /proc/sys/net/ipv4/* Variables:
7 Forward Packets between interfaces.
9 This variable is special, its change resets all configuration
10 parameters to their default state (RFC1122 for hosts, RFC1812
13 ip_default_ttl - INTEGER
14 Default value of TTL field (Time To Live) for outgoing (but not
15 forwarded) IP packets. Should be between 1 and 255 inclusive.
16 Default: 64 (as recommended by RFC1700)
18 ip_no_pmtu_disc - INTEGER
19 Disable Path MTU Discovery. If enabled in mode 1 and a
20 fragmentation-required ICMP is received, the PMTU to this
21 destination will be set to min_pmtu (see below). You will need
22 to raise min_pmtu to the smallest interface MTU on your system
23 manually if you want to avoid locally generated fragments.
25 In mode 2 incoming Path MTU Discovery messages will be
26 discarded. Outgoing frames are handled the same as in mode 1,
27 implicitly setting IP_PMTUDISC_DONT on every created socket.
29 Mode 3 is a hardend pmtu discover mode. The kernel will only
30 accept fragmentation-needed errors if the underlying protocol
31 can verify them besides a plain socket lookup. Current
32 protocols for which pmtu events will be honored are TCP, SCTP
33 and DCCP as they verify e.g. the sequence number or the
34 association. This mode should not be enabled globally but is
35 only intended to secure e.g. name servers in namespaces where
36 TCP path mtu must still work but path MTU information of other
37 protocols should be discarded. If enabled globally this mode
38 could break other protocols.
44 default 552 - minimum discovered Path MTU
46 ip_forward_use_pmtu - BOOLEAN
47 By default we don't trust protocol path MTUs while forwarding
48 because they could be easily forged and can lead to unwanted
49 fragmentation by the router.
50 You only need to enable this if you have user-space software
51 which tries to discover path mtus by itself and depends on the
52 kernel honoring this information. This is normally not the
59 fwmark_reflect - BOOLEAN
60 Controls the fwmark of kernel-generated IPv4 reply packets that are not
61 associated with a socket for example, TCP RSTs or ICMP echo replies).
62 If unset, these packets have a fwmark of zero. If set, they have the
63 fwmark of the packet they are replying to.
66 fib_multipath_use_neigh - BOOLEAN
67 Use status of existing neighbor entry when determining nexthop for
68 multipath routes. If disabled, neighbor information is not used and
69 packets could be directed to a failed nexthop. Only valid for kernels
70 built with CONFIG_IP_ROUTE_MULTIPATH enabled.
76 route/max_size - INTEGER
77 Maximum number of routes allowed in the kernel. Increase
78 this when using large numbers of interfaces and/or routes.
79 From linux kernel 3.6 onwards, this is deprecated for ipv4
80 as route cache is no longer used.
82 neigh/default/gc_thresh1 - INTEGER
83 Minimum number of entries to keep. Garbage collector will not
84 purge entries if there are fewer than this number.
87 neigh/default/gc_thresh2 - INTEGER
88 Threshold when garbage collector becomes more aggressive about
89 purging entries. Entries older than 5 seconds will be cleared
90 when over this number.
93 neigh/default/gc_thresh3 - INTEGER
94 Maximum number of neighbor entries allowed. Increase this
95 when using large numbers of interfaces and when communicating
96 with large numbers of directly-connected peers.
99 neigh/default/unres_qlen_bytes - INTEGER
100 The maximum number of bytes which may be used by packets
101 queued for each unresolved address by other network layers.
103 Setting negative value is meaningless and will return error.
104 Default: 65536 Bytes(64KB)
106 neigh/default/unres_qlen - INTEGER
107 The maximum number of packets which may be queued for each
108 unresolved address by other network layers.
109 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
110 Prior to linux 3.3, the default value is 3 which may cause
111 unexpected packet loss. The current default value is calculated
112 according to default value of unres_qlen_bytes and true size of
116 mtu_expires - INTEGER
117 Time, in seconds, that cached PMTU information is kept.
119 min_adv_mss - INTEGER
120 The advertised MSS depends on the first hop route MTU, but will
121 never be lower than this setting.
125 ipfrag_high_thresh - INTEGER
126 Maximum memory used to reassemble IP fragments. When
127 ipfrag_high_thresh bytes of memory is allocated for this purpose,
128 the fragment handler will toss packets until ipfrag_low_thresh
129 is reached. This also serves as a maximum limit to namespaces
130 different from the initial one.
132 ipfrag_low_thresh - INTEGER
133 Maximum memory used to reassemble IP fragments before the kernel
134 begins to remove incomplete fragment queues to free up resources.
135 The kernel still accepts new fragments for defragmentation.
137 ipfrag_time - INTEGER
138 Time in seconds to keep an IP fragment in memory.
140 ipfrag_max_dist - INTEGER
141 ipfrag_max_dist is a non-negative integer value which defines the
142 maximum "disorder" which is allowed among fragments which share a
143 common IP source address. Note that reordering of packets is
144 not unusual, but if a large number of fragments arrive from a source
145 IP address while a particular fragment queue remains incomplete, it
146 probably indicates that one or more fragments belonging to that queue
147 have been lost. When ipfrag_max_dist is positive, an additional check
148 is done on fragments before they are added to a reassembly queue - if
149 ipfrag_max_dist (or more) fragments have arrived from a particular IP
150 address between additions to any IP fragment queue using that source
151 address, it's presumed that one or more fragments in the queue are
152 lost. The existing fragment queue will be dropped, and a new one
153 started. An ipfrag_max_dist value of zero disables this check.
155 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
156 result in unnecessarily dropping fragment queues when normal
157 reordering of packets occurs, which could lead to poor application
158 performance. Using a very large value, e.g. 50000, increases the
159 likelihood of incorrectly reassembling IP fragments that originate
160 from different IP datagrams, which could result in data corruption.
165 inet_peer_threshold - INTEGER
166 The approximate size of the storage. Starting from this threshold
167 entries will be thrown aggressively. This threshold also determines
168 entries' time-to-live and time intervals between garbage collection
169 passes. More entries, less time-to-live, less GC interval.
171 inet_peer_minttl - INTEGER
172 Minimum time-to-live of entries. Should be enough to cover fragment
173 time-to-live on the reassembling side. This minimum time-to-live is
174 guaranteed if the pool size is less than inet_peer_threshold.
177 inet_peer_maxttl - INTEGER
178 Maximum time-to-live of entries. Unused entries will expire after
179 this period of time if there is no memory pressure on the pool (i.e.
180 when the number of entries in the pool is very small).
186 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
187 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
190 tcp_abort_on_overflow - BOOLEAN
191 If listening service is too slow to accept new connections,
192 reset them. Default state is FALSE. It means that if overflow
193 occurred due to a burst, connection will recover. Enable this
194 option _only_ if you are really sure that listening daemon
195 cannot be tuned to accept connections faster. Enabling this
196 option can harm clients of your server.
198 tcp_adv_win_scale - INTEGER
199 Count buffering overhead as bytes/2^tcp_adv_win_scale
200 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
202 Possible values are [-31, 31], inclusive.
205 tcp_allowed_congestion_control - STRING
206 Show/set the congestion control choices available to non-privileged
207 processes. The list is a subset of those listed in
208 tcp_available_congestion_control.
209 Default is "reno" and the default setting (tcp_congestion_control).
211 tcp_app_win - INTEGER
212 Reserve max(window/2^tcp_app_win, mss) of window for application
213 buffer. Value 0 is special, it means that nothing is reserved.
216 tcp_autocorking - BOOLEAN
217 Enable TCP auto corking :
218 When applications do consecutive small write()/sendmsg() system calls,
219 we try to coalesce these small writes as much as possible, to lower
220 total amount of sent packets. This is done if at least one prior
221 packet for the flow is waiting in Qdisc queues or device transmit
222 queue. Applications can still use TCP_CORK for optimal behavior
223 when they know how/when to uncork their sockets.
226 tcp_available_congestion_control - STRING
227 Shows the available congestion control choices that are registered.
228 More congestion control algorithms may be available as modules,
231 tcp_base_mss - INTEGER
232 The initial value of search_low to be used by the packetization layer
233 Path MTU discovery (MTU probing). If MTU probing is enabled,
234 this is the initial MSS used by the connection.
236 tcp_congestion_control - STRING
237 Set the congestion control algorithm to be used for new
238 connections. The algorithm "reno" is always available, but
239 additional choices may be available based on kernel configuration.
240 Default is set as part of kernel configuration.
241 For passive connections, the listener congestion control choice
243 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
246 Allows TCP to send "duplicate" SACKs.
248 tcp_early_retrans - INTEGER
249 Enable Early Retransmit (ER), per RFC 5827. ER lowers the threshold
250 for triggering fast retransmit when the amount of outstanding data is
251 small and when no previously unsent data can be transmitted (such
252 that limited transmit could be used). Also controls the use of
253 Tail loss probe (TLP) that converts RTOs occurring due to tail
254 losses into fast recovery (draft-dukkipati-tcpm-tcp-loss-probe-01).
258 2 enables ER but delays fast recovery and fast retransmit
259 by a fourth of RTT. This mitigates connection falsely
260 recovers when network has a small degree of reordering
261 (less than 3 packets).
262 3 enables delayed ER and TLP.
267 Control use of Explicit Congestion Notification (ECN) by TCP.
268 ECN is used only when both ends of the TCP connection indicate
269 support for it. This feature is useful in avoiding losses due
270 to congestion by allowing supporting routers to signal
271 congestion before having to drop packets.
273 0 Disable ECN. Neither initiate nor accept ECN.
274 1 Enable ECN when requested by incoming connections and
275 also request ECN on outgoing connection attempts.
276 2 Enable ECN when requested by incoming connections
277 but do not request ECN on outgoing connections.
280 tcp_ecn_fallback - BOOLEAN
281 If the kernel detects that ECN connection misbehaves, enable fall
282 back to non-ECN. Currently, this knob implements the fallback
283 from RFC3168, section 6.1.1.1., but we reserve that in future,
284 additional detection mechanisms could be implemented under this
285 knob. The value is not used, if tcp_ecn or per route (or congestion
286 control) ECN settings are disabled.
287 Default: 1 (fallback enabled)
290 Enable FACK congestion avoidance and fast retransmission.
291 The value is not used, if tcp_sack is not enabled.
293 tcp_fin_timeout - INTEGER
294 The length of time an orphaned (no longer referenced by any
295 application) connection will remain in the FIN_WAIT_2 state
296 before it is aborted at the local end. While a perfectly
297 valid "receive only" state for an un-orphaned connection, an
298 orphaned connection in FIN_WAIT_2 state could otherwise wait
299 forever for the remote to close its end of the connection.
304 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
305 F-RTO is an enhanced recovery algorithm for TCP retransmission
306 timeouts. It is particularly beneficial in networks where the
307 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
308 modification. It does not require any support from the peer.
310 By default it's enabled with a non-zero value. 0 disables F-RTO.
312 tcp_invalid_ratelimit - INTEGER
313 Limit the maximal rate for sending duplicate acknowledgments
314 in response to incoming TCP packets that are for an existing
315 connection but that are invalid due to any of these reasons:
317 (a) out-of-window sequence number,
318 (b) out-of-window acknowledgment number, or
319 (c) PAWS (Protection Against Wrapped Sequence numbers) check failure
321 This can help mitigate simple "ack loop" DoS attacks, wherein
322 a buggy or malicious middlebox or man-in-the-middle can
323 rewrite TCP header fields in manner that causes each endpoint
324 to think that the other is sending invalid TCP segments, thus
325 causing each side to send an unterminating stream of duplicate
326 acknowledgments for invalid segments.
328 Using 0 disables rate-limiting of dupacks in response to
329 invalid segments; otherwise this value specifies the minimal
330 space between sending such dupacks, in milliseconds.
332 Default: 500 (milliseconds).
334 tcp_keepalive_time - INTEGER
335 How often TCP sends out keepalive messages when keepalive is enabled.
338 tcp_keepalive_probes - INTEGER
339 How many keepalive probes TCP sends out, until it decides that the
340 connection is broken. Default value: 9.
342 tcp_keepalive_intvl - INTEGER
343 How frequently the probes are send out. Multiplied by
344 tcp_keepalive_probes it is time to kill not responding connection,
345 after probes started. Default value: 75sec i.e. connection
346 will be aborted after ~11 minutes of retries.
348 tcp_l3mdev_accept - BOOLEAN
349 Enables child sockets to inherit the L3 master device index.
350 Enabling this option allows a "global" listen socket to work
351 across L3 master domains (e.g., VRFs) with connected sockets
352 derived from the listen socket to be bound to the L3 domain in
353 which the packets originated. Only valid when the kernel was
354 compiled with CONFIG_NET_L3_MASTER_DEV.
356 tcp_low_latency - BOOLEAN
357 If set, the TCP stack makes decisions that prefer lower
358 latency as opposed to higher throughput. By default, this
359 option is not set meaning that higher throughput is preferred.
360 An example of an application where this default should be
361 changed would be a Beowulf compute cluster.
364 tcp_max_orphans - INTEGER
365 Maximal number of TCP sockets not attached to any user file handle,
366 held by system. If this number is exceeded orphaned connections are
367 reset immediately and warning is printed. This limit exists
368 only to prevent simple DoS attacks, you _must_ not rely on this
369 or lower the limit artificially, but rather increase it
370 (probably, after increasing installed memory),
371 if network conditions require more than default value,
372 and tune network services to linger and kill such states
373 more aggressively. Let me to remind again: each orphan eats
374 up to ~64K of unswappable memory.
376 tcp_max_syn_backlog - INTEGER
377 Maximal number of remembered connection requests, which have not
378 received an acknowledgment from connecting client.
379 The minimal value is 128 for low memory machines, and it will
380 increase in proportion to the memory of machine.
381 If server suffers from overload, try increasing this number.
383 tcp_max_tw_buckets - INTEGER
384 Maximal number of timewait sockets held by system simultaneously.
385 If this number is exceeded time-wait socket is immediately destroyed
386 and warning is printed. This limit exists only to prevent
387 simple DoS attacks, you _must_ not lower the limit artificially,
388 but rather increase it (probably, after increasing installed memory),
389 if network conditions require more than default value.
391 tcp_mem - vector of 3 INTEGERs: min, pressure, max
392 min: below this number of pages TCP is not bothered about its
395 pressure: when amount of memory allocated by TCP exceeds this number
396 of pages, TCP moderates its memory consumption and enters memory
397 pressure mode, which is exited when memory consumption falls
400 max: number of pages allowed for queueing by all TCP sockets.
402 Defaults are calculated at boot time from amount of available
405 tcp_min_rtt_wlen - INTEGER
406 The window length of the windowed min filter to track the minimum RTT.
407 A shorter window lets a flow more quickly pick up new (higher)
408 minimum RTT when it is moved to a longer path (e.g., due to traffic
409 engineering). A longer window makes the filter more resistant to RTT
410 inflations such as transient congestion. The unit is seconds.
413 tcp_moderate_rcvbuf - BOOLEAN
414 If set, TCP performs receive buffer auto-tuning, attempting to
415 automatically size the buffer (no greater than tcp_rmem[2]) to
416 match the size required by the path for full throughput. Enabled by
419 tcp_mtu_probing - INTEGER
420 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
423 1 - Disabled by default, enabled when an ICMP black hole detected
424 2 - Always enabled, use initial MSS of tcp_base_mss.
426 tcp_probe_interval - INTEGER
427 Controls how often to start TCP Packetization-Layer Path MTU
428 Discovery reprobe. The default is reprobing every 10 minutes as
431 tcp_probe_threshold - INTEGER
432 Controls when TCP Packetization-Layer Path MTU Discovery probing
433 will stop in respect to the width of search range in bytes. Default
436 tcp_no_metrics_save - BOOLEAN
437 By default, TCP saves various connection metrics in the route cache
438 when the connection closes, so that connections established in the
439 near future can use these to set initial conditions. Usually, this
440 increases overall performance, but may sometimes cause performance
441 degradation. If set, TCP will not cache metrics on closing
444 tcp_orphan_retries - INTEGER
445 This value influences the timeout of a locally closed TCP connection,
446 when RTO retransmissions remain unacknowledged.
447 See tcp_retries2 for more details.
449 The default value is 8.
450 If your machine is a loaded WEB server,
451 you should think about lowering this value, such sockets
452 may consume significant resources. Cf. tcp_max_orphans.
454 tcp_recovery - INTEGER
455 This value is a bitmap to enable various experimental loss recovery
458 RACK: 0x1 enables the RACK loss detection for fast detection of lost
459 retransmissions and tail drops.
463 tcp_reordering - INTEGER
464 Initial reordering level of packets in a TCP stream.
465 TCP stack can then dynamically adjust flow reordering level
466 between this initial value and tcp_max_reordering
469 tcp_max_reordering - INTEGER
470 Maximal reordering level of packets in a TCP stream.
471 300 is a fairly conservative value, but you might increase it
472 if paths are using per packet load balancing (like bonding rr mode)
475 tcp_retrans_collapse - BOOLEAN
476 Bug-to-bug compatibility with some broken printers.
477 On retransmit try to send bigger packets to work around bugs in
480 tcp_retries1 - INTEGER
481 This value influences the time, after which TCP decides, that
482 something is wrong due to unacknowledged RTO retransmissions,
483 and reports this suspicion to the network layer.
484 See tcp_retries2 for more details.
486 RFC 1122 recommends at least 3 retransmissions, which is the
489 tcp_retries2 - INTEGER
490 This value influences the timeout of an alive TCP connection,
491 when RTO retransmissions remain unacknowledged.
492 Given a value of N, a hypothetical TCP connection following
493 exponential backoff with an initial RTO of TCP_RTO_MIN would
494 retransmit N times before killing the connection at the (N+1)th RTO.
496 The default value of 15 yields a hypothetical timeout of 924.6
497 seconds and is a lower bound for the effective timeout.
498 TCP will effectively time out at the first RTO which exceeds the
499 hypothetical timeout.
501 RFC 1122 recommends at least 100 seconds for the timeout,
502 which corresponds to a value of at least 8.
504 tcp_rfc1337 - BOOLEAN
505 If set, the TCP stack behaves conforming to RFC1337. If unset,
506 we are not conforming to RFC, but prevent TCP TIME_WAIT
510 tcp_rmem - vector of 3 INTEGERs: min, default, max
511 min: Minimal size of receive buffer used by TCP sockets.
512 It is guaranteed to each TCP socket, even under moderate memory
516 default: initial size of receive buffer used by TCP sockets.
517 This value overrides net.core.rmem_default used by other protocols.
518 Default: 87380 bytes. This value results in window of 65535 with
519 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
520 less for default tcp_app_win. See below about these variables.
522 max: maximal size of receive buffer allowed for automatically
523 selected receiver buffers for TCP socket. This value does not override
524 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
525 automatic tuning of that socket's receive buffer size, in which
526 case this value is ignored.
527 Default: between 87380B and 6MB, depending on RAM size.
530 Enable select acknowledgments (SACKS).
532 tcp_slow_start_after_idle - BOOLEAN
533 If set, provide RFC2861 behavior and time out the congestion
534 window after an idle period. An idle period is defined at
535 the current RTO. If unset, the congestion window will not
536 be timed out after an idle period.
540 Use the Host requirements interpretation of the TCP urgent pointer field.
541 Most hosts use the older BSD interpretation, so if you turn this on
542 Linux might not communicate correctly with them.
545 tcp_synack_retries - INTEGER
546 Number of times SYNACKs for a passive TCP connection attempt will
547 be retransmitted. Should not be higher than 255. Default value
548 is 5, which corresponds to 31seconds till the last retransmission
549 with the current initial RTO of 1second. With this the final timeout
550 for a passive TCP connection will happen after 63seconds.
552 tcp_syncookies - BOOLEAN
553 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
554 Send out syncookies when the syn backlog queue of a socket
555 overflows. This is to prevent against the common 'SYN flood attack'
558 Note, that syncookies is fallback facility.
559 It MUST NOT be used to help highly loaded servers to stand
560 against legal connection rate. If you see SYN flood warnings
561 in your logs, but investigation shows that they occur
562 because of overload with legal connections, you should tune
563 another parameters until this warning disappear.
564 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
566 syncookies seriously violate TCP protocol, do not allow
567 to use TCP extensions, can result in serious degradation
568 of some services (f.e. SMTP relaying), visible not by you,
569 but your clients and relays, contacting you. While you see
570 SYN flood warnings in logs not being really flooded, your server
571 is seriously misconfigured.
573 If you want to test which effects syncookies have to your
574 network connections you can set this knob to 2 to enable
575 unconditionally generation of syncookies.
577 tcp_fastopen - INTEGER
578 Enable TCP Fast Open feature (draft-ietf-tcpm-fastopen) to send data
579 in the opening SYN packet. To use this feature, the client application
580 must use sendmsg() or sendto() with MSG_FASTOPEN flag rather than
581 connect() to perform a TCP handshake automatically.
583 The values (bitmap) are
584 1: Enables sending data in the opening SYN on the client w/ MSG_FASTOPEN.
585 2: Enables TCP Fast Open on the server side, i.e., allowing data in
586 a SYN packet to be accepted and passed to the application before
587 3-way hand shake finishes.
588 4: Send data in the opening SYN regardless of cookie availability and
589 without a cookie option.
590 0x100: Accept SYN data w/o validating the cookie.
591 0x200: Accept data-in-SYN w/o any cookie option present.
592 0x400/0x800: Enable Fast Open on all listeners regardless of the
593 TCP_FASTOPEN socket option. The two different flags designate two
594 different ways of setting max_qlen without the TCP_FASTOPEN socket
599 Note that the client & server side Fast Open flags (1 and 2
600 respectively) must be also enabled before the rest of flags can take
603 See include/net/tcp.h and the code for more details.
605 tcp_syn_retries - INTEGER
606 Number of times initial SYNs for an active TCP connection attempt
607 will be retransmitted. Should not be higher than 127. Default value
608 is 6, which corresponds to 63seconds till the last retransmission
609 with the current initial RTO of 1second. With this the final timeout
610 for an active TCP connection attempt will happen after 127seconds.
612 tcp_timestamps - BOOLEAN
613 Enable timestamps as defined in RFC1323.
615 tcp_min_tso_segs - INTEGER
616 Minimal number of segments per TSO frame.
617 Since linux-3.12, TCP does an automatic sizing of TSO frames,
618 depending on flow rate, instead of filling 64Kbytes packets.
619 For specific usages, it's possible to force TCP to build big
620 TSO frames. Note that TCP stack might split too big TSO packets
621 if available window is too small.
624 tcp_pacing_ss_ratio - INTEGER
625 sk->sk_pacing_rate is set by TCP stack using a ratio applied
626 to current rate. (current_rate = cwnd * mss / srtt)
627 If TCP is in slow start, tcp_pacing_ss_ratio is applied
628 to let TCP probe for bigger speeds, assuming cwnd can be
629 doubled every other RTT.
632 tcp_pacing_ca_ratio - INTEGER
633 sk->sk_pacing_rate is set by TCP stack using a ratio applied
634 to current rate. (current_rate = cwnd * mss / srtt)
635 If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio
636 is applied to conservatively probe for bigger throughput.
639 tcp_tso_win_divisor - INTEGER
640 This allows control over what percentage of the congestion window
641 can be consumed by a single TSO frame.
642 The setting of this parameter is a choice between burstiness and
643 building larger TSO frames.
646 tcp_tw_recycle - BOOLEAN
647 Enable fast recycling TIME-WAIT sockets. Default value is 0.
648 It should not be changed without advice/request of technical
651 tcp_tw_reuse - BOOLEAN
652 Allow to reuse TIME-WAIT sockets for new connections when it is
653 safe from protocol viewpoint. Default value is 0.
654 It should not be changed without advice/request of technical
657 tcp_window_scaling - BOOLEAN
658 Enable window scaling as defined in RFC1323.
660 tcp_wmem - vector of 3 INTEGERs: min, default, max
661 min: Amount of memory reserved for send buffers for TCP sockets.
662 Each TCP socket has rights to use it due to fact of its birth.
665 default: initial size of send buffer used by TCP sockets. This
666 value overrides net.core.wmem_default used by other protocols.
667 It is usually lower than net.core.wmem_default.
670 max: Maximal amount of memory allowed for automatically tuned
671 send buffers for TCP sockets. This value does not override
672 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
673 automatic tuning of that socket's send buffer size, in which case
674 this value is ignored.
675 Default: between 64K and 4MB, depending on RAM size.
677 tcp_notsent_lowat - UNSIGNED INTEGER
678 A TCP socket can control the amount of unsent bytes in its write queue,
679 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
680 reports POLLOUT events if the amount of unsent bytes is below a per
681 socket value, and if the write queue is not full. sendmsg() will
682 also not add new buffers if the limit is hit.
684 This global variable controls the amount of unsent data for
685 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
686 to the global variable has immediate effect.
688 Default: UINT_MAX (0xFFFFFFFF)
690 tcp_workaround_signed_windows - BOOLEAN
691 If set, assume no receipt of a window scaling option means the
692 remote TCP is broken and treats the window as a signed quantity.
693 If unset, assume the remote TCP is not broken even if we do
694 not receive a window scaling option from them.
697 tcp_thin_linear_timeouts - BOOLEAN
698 Enable dynamic triggering of linear timeouts for thin streams.
699 If set, a check is performed upon retransmission by timeout to
700 determine if the stream is thin (less than 4 packets in flight).
701 As long as the stream is found to be thin, up to 6 linear
702 timeouts may be performed before exponential backoff mode is
703 initiated. This improves retransmission latency for
704 non-aggressive thin streams, often found to be time-dependent.
705 For more information on thin streams, see
706 Documentation/networking/tcp-thin.txt
709 tcp_thin_dupack - BOOLEAN
710 Enable dynamic triggering of retransmissions after one dupACK
711 for thin streams. If set, a check is performed upon reception
712 of a dupACK to determine if the stream is thin (less than 4
713 packets in flight). As long as the stream is found to be thin,
714 data is retransmitted on the first received dupACK. This
715 improves retransmission latency for non-aggressive thin
716 streams, often found to be time-dependent.
717 For more information on thin streams, see
718 Documentation/networking/tcp-thin.txt
721 tcp_limit_output_bytes - INTEGER
722 Controls TCP Small Queue limit per tcp socket.
723 TCP bulk sender tends to increase packets in flight until it
724 gets losses notifications. With SNDBUF autotuning, this can
725 result in a large amount of packets queued in qdisc/device
726 on the local machine, hurting latency of other flows, for
727 typical pfifo_fast qdiscs.
728 tcp_limit_output_bytes limits the number of bytes on qdisc
729 or device to reduce artificial RTT/cwnd and reduce bufferbloat.
732 tcp_challenge_ack_limit - INTEGER
733 Limits number of Challenge ACK sent per second, as recommended
734 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
739 udp_mem - vector of 3 INTEGERs: min, pressure, max
740 Number of pages allowed for queueing by all UDP sockets.
742 min: Below this number of pages UDP is not bothered about its
743 memory appetite. When amount of memory allocated by UDP exceeds
744 this number, UDP starts to moderate memory usage.
746 pressure: This value was introduced to follow format of tcp_mem.
748 max: Number of pages allowed for queueing by all UDP sockets.
750 Default is calculated at boot time from amount of available memory.
752 udp_rmem_min - INTEGER
753 Minimal size of receive buffer used by UDP sockets in moderation.
754 Each UDP socket is able to use the size for receiving data, even if
755 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
758 udp_wmem_min - INTEGER
759 Minimal size of send buffer used by UDP sockets in moderation.
760 Each UDP socket is able to use the size for sending data, even if
761 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
766 cipso_cache_enable - BOOLEAN
767 If set, enable additions to and lookups from the CIPSO label mapping
768 cache. If unset, additions are ignored and lookups always result in a
769 miss. However, regardless of the setting the cache is still
770 invalidated when required when means you can safely toggle this on and
771 off and the cache will always be "safe".
774 cipso_cache_bucket_size - INTEGER
775 The CIPSO label cache consists of a fixed size hash table with each
776 hash bucket containing a number of cache entries. This variable limits
777 the number of entries in each hash bucket; the larger the value the
778 more CIPSO label mappings that can be cached. When the number of
779 entries in a given hash bucket reaches this limit adding new entries
780 causes the oldest entry in the bucket to be removed to make room.
783 cipso_rbm_optfmt - BOOLEAN
784 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
785 the CIPSO draft specification (see Documentation/netlabel for details).
786 This means that when set the CIPSO tag will be padded with empty
787 categories in order to make the packet data 32-bit aligned.
790 cipso_rbm_structvalid - BOOLEAN
791 If set, do a very strict check of the CIPSO option when
792 ip_options_compile() is called. If unset, relax the checks done during
793 ip_options_compile(). Either way is "safe" as errors are caught else
794 where in the CIPSO processing code but setting this to 0 (False) should
795 result in less work (i.e. it should be faster) but could cause problems
796 with other implementations that require strict checking.
801 ip_local_port_range - 2 INTEGERS
802 Defines the local port range that is used by TCP and UDP to
803 choose the local port. The first number is the first, the
804 second the last local port number.
805 If possible, it is better these numbers have different parity.
806 (one even and one odd values)
807 The default values are 32768 and 60999 respectively.
809 ip_local_reserved_ports - list of comma separated ranges
810 Specify the ports which are reserved for known third-party
811 applications. These ports will not be used by automatic port
812 assignments (e.g. when calling connect() or bind() with port
813 number 0). Explicit port allocation behavior is unchanged.
815 The format used for both input and output is a comma separated
816 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
817 10). Writing to the file will clear all previously reserved
818 ports and update the current list with the one given in the
821 Note that ip_local_port_range and ip_local_reserved_ports
822 settings are independent and both are considered by the kernel
823 when determining which ports are available for automatic port
826 You can reserve ports which are not in the current
827 ip_local_port_range, e.g.:
829 $ cat /proc/sys/net/ipv4/ip_local_port_range
831 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
834 although this is redundant. However such a setting is useful
835 if later the port range is changed to a value that will
836 include the reserved ports.
840 ip_nonlocal_bind - BOOLEAN
841 If set, allows processes to bind() to non-local IP addresses,
842 which can be quite useful - but may break some applications.
846 If set non-zero, enables support for dynamic addresses.
847 If set to a non-zero value larger than 1, a kernel log
848 message will be printed when dynamic address rewriting
852 ip_early_demux - BOOLEAN
853 Optimize input packet processing down to one demux for
854 certain kinds of local sockets. Currently we only do this
855 for established TCP sockets.
857 It may add an additional cost for pure routing workloads that
858 reduces overall throughput, in such case you should disable it.
861 icmp_echo_ignore_all - BOOLEAN
862 If set non-zero, then the kernel will ignore all ICMP ECHO
866 icmp_echo_ignore_broadcasts - BOOLEAN
867 If set non-zero, then the kernel will ignore all ICMP ECHO and
868 TIMESTAMP requests sent to it via broadcast/multicast.
871 icmp_ratelimit - INTEGER
872 Limit the maximal rates for sending ICMP packets whose type matches
873 icmp_ratemask (see below) to specific targets.
874 0 to disable any limiting,
875 otherwise the minimal space between responses in milliseconds.
876 Note that another sysctl, icmp_msgs_per_sec limits the number
877 of ICMP packets sent on all targets.
880 icmp_msgs_per_sec - INTEGER
881 Limit maximal number of ICMP packets sent per second from this host.
882 Only messages whose type matches icmp_ratemask (see below) are
883 controlled by this limit.
886 icmp_msgs_burst - INTEGER
887 icmp_msgs_per_sec controls number of ICMP packets sent per second,
888 while icmp_msgs_burst controls the burst size of these packets.
891 icmp_ratemask - INTEGER
892 Mask made of ICMP types for which rates are being limited.
893 Significant bits: IHGFEDCBA9876543210
894 Default mask: 0000001100000011000 (6168)
896 Bit definitions (see include/linux/icmp.h):
898 3 Destination Unreachable *
903 C Parameter Problem *
908 H Address Mask Request
911 * These are rate limited by default (see default mask above)
913 icmp_ignore_bogus_error_responses - BOOLEAN
914 Some routers violate RFC1122 by sending bogus responses to broadcast
915 frames. Such violations are normally logged via a kernel warning.
916 If this is set to TRUE, the kernel will not give such warnings, which
917 will avoid log file clutter.
920 icmp_errors_use_inbound_ifaddr - BOOLEAN
922 If zero, icmp error messages are sent with the primary address of
923 the exiting interface.
925 If non-zero, the message will be sent with the primary address of
926 the interface that received the packet that caused the icmp error.
927 This is the behaviour network many administrators will expect from
928 a router. And it can make debugging complicated network layouts
931 Note that if no primary address exists for the interface selected,
932 then the primary address of the first non-loopback interface that
933 has one will be used regardless of this setting.
937 igmp_max_memberships - INTEGER
938 Change the maximum number of multicast groups we can subscribe to.
941 Theoretical maximum value is bounded by having to send a membership
942 report in a single datagram (i.e. the report can't span multiple
943 datagrams, or risk confusing the switch and leaving groups you don't
946 The number of supported groups 'M' is bounded by the number of group
947 report entries you can fit into a single datagram of 65535 bytes.
949 M = 65536-sizeof (ip header)/(sizeof(Group record))
951 Group records are variable length, with a minimum of 12 bytes.
952 So net.ipv4.igmp_max_memberships should not be set higher than:
954 (65536-24) / 12 = 5459
956 The value 5459 assumes no IP header options, so in practice
957 this number may be lower.
959 igmp_max_msf - INTEGER
960 Maximum number of addresses allowed in the source filter list for a
965 Controls the IGMP query robustness variable (see RFC2236 8.1).
966 Default: 2 (as specified by RFC2236 8.1)
967 Minimum: 1 (as specified by RFC6636 4.5)
969 conf/interface/* changes special settings per interface (where
970 "interface" is the name of your network interface)
972 conf/all/* is special, changes the settings for all interfaces
974 log_martians - BOOLEAN
975 Log packets with impossible addresses to kernel log.
976 log_martians for the interface will be enabled if at least one of
977 conf/{all,interface}/log_martians is set to TRUE,
978 it will be disabled otherwise
980 accept_redirects - BOOLEAN
981 Accept ICMP redirect messages.
982 accept_redirects for the interface will be enabled if:
983 - both conf/{all,interface}/accept_redirects are TRUE in the case
984 forwarding for the interface is enabled
986 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
987 case forwarding for the interface is disabled
988 accept_redirects for the interface will be disabled otherwise
993 Enable IP forwarding on this interface.
995 mc_forwarding - BOOLEAN
996 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
997 and a multicast routing daemon is required.
998 conf/all/mc_forwarding must also be set to TRUE to enable multicast
999 routing for the interface
1002 Integer value used to differentiate the devices by the medium they
1003 are attached to. Two devices can have different id values when
1004 the broadcast packets are received only on one of them.
1005 The default value 0 means that the device is the only interface
1006 to its medium, value of -1 means that medium is not known.
1008 Currently, it is used to change the proxy_arp behavior:
1009 the proxy_arp feature is enabled for packets forwarded between
1010 two devices attached to different media.
1014 proxy_arp for the interface will be enabled if at least one of
1015 conf/{all,interface}/proxy_arp is set to TRUE,
1016 it will be disabled otherwise
1018 proxy_arp_pvlan - BOOLEAN
1019 Private VLAN proxy arp.
1020 Basically allow proxy arp replies back to the same interface
1021 (from which the ARP request/solicitation was received).
1023 This is done to support (ethernet) switch features, like RFC
1024 3069, where the individual ports are NOT allowed to
1025 communicate with each other, but they are allowed to talk to
1026 the upstream router. As described in RFC 3069, it is possible
1027 to allow these hosts to communicate through the upstream
1028 router by proxy_arp'ing. Don't need to be used together with
1031 This technology is known by different names:
1032 In RFC 3069 it is called VLAN Aggregation.
1033 Cisco and Allied Telesyn call it Private VLAN.
1034 Hewlett-Packard call it Source-Port filtering or port-isolation.
1035 Ericsson call it MAC-Forced Forwarding (RFC Draft).
1037 shared_media - BOOLEAN
1038 Send(router) or accept(host) RFC1620 shared media redirects.
1039 Overrides ip_secure_redirects.
1040 shared_media for the interface will be enabled if at least one of
1041 conf/{all,interface}/shared_media is set to TRUE,
1042 it will be disabled otherwise
1045 secure_redirects - BOOLEAN
1046 Accept ICMP redirect messages only for gateways,
1047 listed in default gateway list.
1048 secure_redirects for the interface will be enabled if at least one of
1049 conf/{all,interface}/secure_redirects is set to TRUE,
1050 it will be disabled otherwise
1053 send_redirects - BOOLEAN
1054 Send redirects, if router.
1055 send_redirects for the interface will be enabled if at least one of
1056 conf/{all,interface}/send_redirects is set to TRUE,
1057 it will be disabled otherwise
1060 bootp_relay - BOOLEAN
1061 Accept packets with source address 0.b.c.d destined
1062 not to this host as local ones. It is supposed, that
1063 BOOTP relay daemon will catch and forward such packets.
1064 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
1067 Not Implemented Yet.
1069 accept_source_route - BOOLEAN
1070 Accept packets with SRR option.
1071 conf/all/accept_source_route must also be set to TRUE to accept packets
1072 with SRR option on the interface
1073 default TRUE (router)
1076 accept_local - BOOLEAN
1077 Accept packets with local source addresses. In combination with
1078 suitable routing, this can be used to direct packets between two
1079 local interfaces over the wire and have them accepted properly.
1082 route_localnet - BOOLEAN
1083 Do not consider loopback addresses as martian source or destination
1084 while routing. This enables the use of 127/8 for local routing purposes.
1088 0 - No source validation.
1089 1 - Strict mode as defined in RFC3704 Strict Reverse Path
1090 Each incoming packet is tested against the FIB and if the interface
1091 is not the best reverse path the packet check will fail.
1092 By default failed packets are discarded.
1093 2 - Loose mode as defined in RFC3704 Loose Reverse Path
1094 Each incoming packet's source address is also tested against the FIB
1095 and if the source address is not reachable via any interface
1096 the packet check will fail.
1098 Current recommended practice in RFC3704 is to enable strict mode
1099 to prevent IP spoofing from DDos attacks. If using asymmetric routing
1100 or other complicated routing, then loose mode is recommended.
1102 The max value from conf/{all,interface}/rp_filter is used
1103 when doing source validation on the {interface}.
1105 Default value is 0. Note that some distributions enable it
1108 arp_filter - BOOLEAN
1109 1 - Allows you to have multiple network interfaces on the same
1110 subnet, and have the ARPs for each interface be answered
1111 based on whether or not the kernel would route a packet from
1112 the ARP'd IP out that interface (therefore you must use source
1113 based routing for this to work). In other words it allows control
1114 of which cards (usually 1) will respond to an arp request.
1116 0 - (default) The kernel can respond to arp requests with addresses
1117 from other interfaces. This may seem wrong but it usually makes
1118 sense, because it increases the chance of successful communication.
1119 IP addresses are owned by the complete host on Linux, not by
1120 particular interfaces. Only for more complex setups like load-
1121 balancing, does this behaviour cause problems.
1123 arp_filter for the interface will be enabled if at least one of
1124 conf/{all,interface}/arp_filter is set to TRUE,
1125 it will be disabled otherwise
1127 arp_announce - INTEGER
1128 Define different restriction levels for announcing the local
1129 source IP address from IP packets in ARP requests sent on
1131 0 - (default) Use any local address, configured on any interface
1132 1 - Try to avoid local addresses that are not in the target's
1133 subnet for this interface. This mode is useful when target
1134 hosts reachable via this interface require the source IP
1135 address in ARP requests to be part of their logical network
1136 configured on the receiving interface. When we generate the
1137 request we will check all our subnets that include the
1138 target IP and will preserve the source address if it is from
1139 such subnet. If there is no such subnet we select source
1140 address according to the rules for level 2.
1141 2 - Always use the best local address for this target.
1142 In this mode we ignore the source address in the IP packet
1143 and try to select local address that we prefer for talks with
1144 the target host. Such local address is selected by looking
1145 for primary IP addresses on all our subnets on the outgoing
1146 interface that include the target IP address. If no suitable
1147 local address is found we select the first local address
1148 we have on the outgoing interface or on all other interfaces,
1149 with the hope we will receive reply for our request and
1150 even sometimes no matter the source IP address we announce.
1152 The max value from conf/{all,interface}/arp_announce is used.
1154 Increasing the restriction level gives more chance for
1155 receiving answer from the resolved target while decreasing
1156 the level announces more valid sender's information.
1158 arp_ignore - INTEGER
1159 Define different modes for sending replies in response to
1160 received ARP requests that resolve local target IP addresses:
1161 0 - (default): reply for any local target IP address, configured
1163 1 - reply only if the target IP address is local address
1164 configured on the incoming interface
1165 2 - reply only if the target IP address is local address
1166 configured on the incoming interface and both with the
1167 sender's IP address are part from same subnet on this interface
1168 3 - do not reply for local addresses configured with scope host,
1169 only resolutions for global and link addresses are replied
1171 8 - do not reply for all local addresses
1173 The max value from conf/{all,interface}/arp_ignore is used
1174 when ARP request is received on the {interface}
1176 arp_notify - BOOLEAN
1177 Define mode for notification of address and device changes.
1178 0 - (default): do nothing
1179 1 - Generate gratuitous arp requests when device is brought up
1180 or hardware address changes.
1182 arp_accept - BOOLEAN
1183 Define behavior for gratuitous ARP frames who's IP is not
1184 already present in the ARP table:
1185 0 - don't create new entries in the ARP table
1186 1 - create new entries in the ARP table
1188 Both replies and requests type gratuitous arp will trigger the
1189 ARP table to be updated, if this setting is on.
1191 If the ARP table already contains the IP address of the
1192 gratuitous arp frame, the arp table will be updated regardless
1193 if this setting is on or off.
1195 mcast_solicit - INTEGER
1196 The maximum number of multicast probes in INCOMPLETE state,
1197 when the associated hardware address is unknown. Defaults
1200 ucast_solicit - INTEGER
1201 The maximum number of unicast probes in PROBE state, when
1202 the hardware address is being reconfirmed. Defaults to 3.
1204 app_solicit - INTEGER
1205 The maximum number of probes to send to the user space ARP daemon
1206 via netlink before dropping back to multicast probes (see
1207 mcast_resolicit). Defaults to 0.
1209 mcast_resolicit - INTEGER
1210 The maximum number of multicast probes after unicast and
1211 app probes in PROBE state. Defaults to 0.
1213 disable_policy - BOOLEAN
1214 Disable IPSEC policy (SPD) for this interface
1216 disable_xfrm - BOOLEAN
1217 Disable IPSEC encryption on this interface, whatever the policy
1219 igmpv2_unsolicited_report_interval - INTEGER
1220 The interval in milliseconds in which the next unsolicited
1221 IGMPv1 or IGMPv2 report retransmit will take place.
1222 Default: 10000 (10 seconds)
1224 igmpv3_unsolicited_report_interval - INTEGER
1225 The interval in milliseconds in which the next unsolicited
1226 IGMPv3 report retransmit will take place.
1227 Default: 1000 (1 seconds)
1229 promote_secondaries - BOOLEAN
1230 When a primary IP address is removed from this interface
1231 promote a corresponding secondary IP address instead of
1232 removing all the corresponding secondary IP addresses.
1234 drop_unicast_in_l2_multicast - BOOLEAN
1235 Drop any unicast IP packets that are received in link-layer
1236 multicast (or broadcast) frames.
1237 This behavior (for multicast) is actually a SHOULD in RFC
1238 1122, but is disabled by default for compatibility reasons.
1241 drop_gratuitous_arp - BOOLEAN
1242 Drop all gratuitous ARP frames, for example if there's a known
1243 good ARP proxy on the network and such frames need not be used
1244 (or in the case of 802.11, must not be used to prevent attacks.)
1249 Allows you to write a number, which can be used as required.
1252 xfrm4_gc_thresh - INTEGER
1253 The threshold at which we will start garbage collecting for IPv4
1254 destination cache entries. At twice this value the system will
1255 refuse new allocations. The value must be set below the flowcache
1256 limit (4096 * number of online cpus) to take effect.
1258 igmp_link_local_mcast_reports - BOOLEAN
1259 Enable IGMP reports for link local multicast groups in the
1264 kuznet@ms2.inr.ac.ru
1270 delon.nicolas@wanadoo.fr
1275 /proc/sys/net/ipv6/* Variables:
1277 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1278 apply to IPv6 [XXX?].
1280 bindv6only - BOOLEAN
1281 Default value for IPV6_V6ONLY socket option,
1282 which restricts use of the IPv6 socket to IPv6 communication
1284 TRUE: disable IPv4-mapped address feature
1285 FALSE: enable IPv4-mapped address feature
1287 Default: FALSE (as specified in RFC3493)
1289 flowlabel_consistency - BOOLEAN
1290 Protect the consistency (and unicity) of flow label.
1291 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1297 auto_flowlabels - INTEGER
1298 Automatically generate flow labels based on a flow hash of the
1299 packet. This allows intermediate devices, such as routers, to
1300 identify packet flows for mechanisms like Equal Cost Multipath
1301 Routing (see RFC 6438).
1302 0: automatic flow labels are completely disabled
1303 1: automatic flow labels are enabled by default, they can be
1304 disabled on a per socket basis using the IPV6_AUTOFLOWLABEL
1306 2: automatic flow labels are allowed, they may be enabled on a
1307 per socket basis using the IPV6_AUTOFLOWLABEL socket option
1308 3: automatic flow labels are enabled and enforced, they cannot
1309 be disabled by the socket option
1312 flowlabel_state_ranges - BOOLEAN
1313 Split the flow label number space into two ranges. 0-0x7FFFF is
1314 reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF
1315 is reserved for stateless flow labels as described in RFC6437.
1320 anycast_src_echo_reply - BOOLEAN
1321 Controls the use of anycast addresses as source addresses for ICMPv6
1327 idgen_delay - INTEGER
1328 Controls the delay in seconds after which time to retry
1329 privacy stable address generation if a DAD conflict is
1331 Default: 1 (as specified in RFC7217)
1333 idgen_retries - INTEGER
1334 Controls the number of retries to generate a stable privacy
1335 address if a DAD conflict is detected.
1336 Default: 3 (as specified in RFC7217)
1339 Controls the MLD query robustness variable (see RFC3810 9.1).
1340 Default: 2 (as specified by RFC3810 9.1)
1341 Minimum: 1 (as specified by RFC6636 4.5)
1345 ip6frag_high_thresh - INTEGER
1346 Maximum memory used to reassemble IPv6 fragments. When
1347 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1348 the fragment handler will toss packets until ip6frag_low_thresh
1351 ip6frag_low_thresh - INTEGER
1352 See ip6frag_high_thresh
1354 ip6frag_time - INTEGER
1355 Time in seconds to keep an IPv6 fragment in memory.
1358 Change the interface-specific default settings.
1362 Change all the interface-specific settings.
1364 [XXX: Other special features than forwarding?]
1366 conf/all/forwarding - BOOLEAN
1367 Enable global IPv6 forwarding between all interfaces.
1369 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1370 to control which interfaces may forward packets and which not.
1372 This also sets all interfaces' Host/Router setting
1373 'forwarding' to the specified value. See below for details.
1375 This referred to as global forwarding.
1380 fwmark_reflect - BOOLEAN
1381 Controls the fwmark of kernel-generated IPv6 reply packets that are not
1382 associated with a socket for example, TCP RSTs or ICMPv6 echo replies).
1383 If unset, these packets have a fwmark of zero. If set, they have the
1384 fwmark of the packet they are replying to.
1388 Change special settings per interface.
1390 The functional behaviour for certain settings is different
1391 depending on whether local forwarding is enabled or not.
1394 Accept Router Advertisements; autoconfigure using them.
1396 It also determines whether or not to transmit Router
1397 Solicitations. If and only if the functional setting is to
1398 accept Router Advertisements, Router Solicitations will be
1401 Possible values are:
1402 0 Do not accept Router Advertisements.
1403 1 Accept Router Advertisements if forwarding is disabled.
1404 2 Overrule forwarding behaviour. Accept Router Advertisements
1405 even if forwarding is enabled.
1407 Functional default: enabled if local forwarding is disabled.
1408 disabled if local forwarding is enabled.
1410 accept_ra_defrtr - BOOLEAN
1411 Learn default router in Router Advertisement.
1413 Functional default: enabled if accept_ra is enabled.
1414 disabled if accept_ra is disabled.
1416 accept_ra_from_local - BOOLEAN
1417 Accept RA with source-address that is found on local machine
1418 if the RA is otherwise proper and able to be accepted.
1419 Default is to NOT accept these as it may be an un-intended
1423 enabled if accept_ra_from_local is enabled
1424 on a specific interface.
1425 disabled if accept_ra_from_local is disabled
1426 on a specific interface.
1428 accept_ra_min_hop_limit - INTEGER
1429 Minimum hop limit Information in Router Advertisement.
1431 Hop limit Information in Router Advertisement less than this
1432 variable shall be ignored.
1436 accept_ra_pinfo - BOOLEAN
1437 Learn Prefix Information in Router Advertisement.
1439 Functional default: enabled if accept_ra is enabled.
1440 disabled if accept_ra is disabled.
1442 accept_ra_rt_info_max_plen - INTEGER
1443 Maximum prefix length of Route Information in RA.
1445 Route Information w/ prefix larger than or equal to this
1446 variable shall be ignored.
1448 Functional default: 0 if accept_ra_rtr_pref is enabled.
1449 -1 if accept_ra_rtr_pref is disabled.
1451 accept_ra_rtr_pref - BOOLEAN
1452 Accept Router Preference in RA.
1454 Functional default: enabled if accept_ra is enabled.
1455 disabled if accept_ra is disabled.
1457 accept_ra_mtu - BOOLEAN
1458 Apply the MTU value specified in RA option 5 (RFC4861). If
1459 disabled, the MTU specified in the RA will be ignored.
1461 Functional default: enabled if accept_ra is enabled.
1462 disabled if accept_ra is disabled.
1464 accept_redirects - BOOLEAN
1467 Functional default: enabled if local forwarding is disabled.
1468 disabled if local forwarding is enabled.
1470 accept_source_route - INTEGER
1471 Accept source routing (routing extension header).
1473 >= 0: Accept only routing header type 2.
1474 < 0: Do not accept routing header.
1479 Autoconfigure addresses using Prefix Information in Router
1482 Functional default: enabled if accept_ra_pinfo is enabled.
1483 disabled if accept_ra_pinfo is disabled.
1485 dad_transmits - INTEGER
1486 The amount of Duplicate Address Detection probes to send.
1489 forwarding - INTEGER
1490 Configure interface-specific Host/Router behaviour.
1492 Note: It is recommended to have the same setting on all
1493 interfaces; mixed router/host scenarios are rather uncommon.
1495 Possible values are:
1496 0 Forwarding disabled
1497 1 Forwarding enabled
1501 By default, Host behaviour is assumed. This means:
1503 1. IsRouter flag is not set in Neighbour Advertisements.
1504 2. If accept_ra is TRUE (default), transmit Router
1506 3. If accept_ra is TRUE (default), accept Router
1507 Advertisements (and do autoconfiguration).
1508 4. If accept_redirects is TRUE (default), accept Redirects.
1512 If local forwarding is enabled, Router behaviour is assumed.
1513 This means exactly the reverse from the above:
1515 1. IsRouter flag is set in Neighbour Advertisements.
1516 2. Router Solicitations are not sent unless accept_ra is 2.
1517 3. Router Advertisements are ignored unless accept_ra is 2.
1518 4. Redirects are ignored.
1520 Default: 0 (disabled) if global forwarding is disabled (default),
1521 otherwise 1 (enabled).
1524 Default Hop Limit to set.
1528 Default Maximum Transfer Unit
1529 Default: 1280 (IPv6 required minimum)
1531 ip_nonlocal_bind - BOOLEAN
1532 If set, allows processes to bind() to non-local IPv6 addresses,
1533 which can be quite useful - but may break some applications.
1536 router_probe_interval - INTEGER
1537 Minimum interval (in seconds) between Router Probing described
1542 router_solicitation_delay - INTEGER
1543 Number of seconds to wait after interface is brought up
1544 before sending Router Solicitations.
1547 router_solicitation_interval - INTEGER
1548 Number of seconds to wait between Router Solicitations.
1551 router_solicitations - INTEGER
1552 Number of Router Solicitations to send until assuming no
1553 routers are present.
1556 use_oif_addrs_only - BOOLEAN
1557 When enabled, the candidate source addresses for destinations
1558 routed via this interface are restricted to the set of addresses
1559 configured on this interface (vis. RFC 6724, section 4).
1563 use_tempaddr - INTEGER
1564 Preference for Privacy Extensions (RFC3041).
1565 <= 0 : disable Privacy Extensions
1566 == 1 : enable Privacy Extensions, but prefer public
1567 addresses over temporary addresses.
1568 > 1 : enable Privacy Extensions and prefer temporary
1569 addresses over public addresses.
1570 Default: 0 (for most devices)
1571 -1 (for point-to-point devices and loopback devices)
1573 temp_valid_lft - INTEGER
1574 valid lifetime (in seconds) for temporary addresses.
1575 Default: 604800 (7 days)
1577 temp_prefered_lft - INTEGER
1578 Preferred lifetime (in seconds) for temporary addresses.
1579 Default: 86400 (1 day)
1581 keep_addr_on_down - INTEGER
1582 Keep all IPv6 addresses on an interface down event. If set static
1583 global addresses with no expiration time are not flushed.
1588 Default: 0 (addresses are removed)
1590 max_desync_factor - INTEGER
1591 Maximum value for DESYNC_FACTOR, which is a random value
1592 that ensures that clients don't synchronize with each
1593 other and generate new addresses at exactly the same time.
1594 value is in seconds.
1597 regen_max_retry - INTEGER
1598 Number of attempts before give up attempting to generate
1599 valid temporary addresses.
1602 max_addresses - INTEGER
1603 Maximum number of autoconfigured addresses per interface. Setting
1604 to zero disables the limitation. It is not recommended to set this
1605 value too large (or to zero) because it would be an easy way to
1606 crash the kernel by allowing too many addresses to be created.
1609 disable_ipv6 - BOOLEAN
1610 Disable IPv6 operation. If accept_dad is set to 2, this value
1611 will be dynamically set to TRUE if DAD fails for the link-local
1613 Default: FALSE (enable IPv6 operation)
1615 When this value is changed from 1 to 0 (IPv6 is being enabled),
1616 it will dynamically create a link-local address on the given
1617 interface and start Duplicate Address Detection, if necessary.
1619 When this value is changed from 0 to 1 (IPv6 is being disabled),
1620 it will dynamically delete all address on the given interface.
1622 accept_dad - INTEGER
1623 Whether to accept DAD (Duplicate Address Detection).
1625 1: Enable DAD (default)
1626 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1627 link-local address has been found.
1629 force_tllao - BOOLEAN
1630 Enable sending the target link-layer address option even when
1631 responding to a unicast neighbor solicitation.
1634 Quoting from RFC 2461, section 4.4, Target link-layer address:
1636 "The option MUST be included for multicast solicitations in order to
1637 avoid infinite Neighbor Solicitation "recursion" when the peer node
1638 does not have a cache entry to return a Neighbor Advertisements
1639 message. When responding to unicast solicitations, the option can be
1640 omitted since the sender of the solicitation has the correct link-
1641 layer address; otherwise it would not have be able to send the unicast
1642 solicitation in the first place. However, including the link-layer
1643 address in this case adds little overhead and eliminates a potential
1644 race condition where the sender deletes the cached link-layer address
1645 prior to receiving a response to a previous solicitation."
1647 ndisc_notify - BOOLEAN
1648 Define mode for notification of address and device changes.
1649 0 - (default): do nothing
1650 1 - Generate unsolicited neighbour advertisements when device is brought
1651 up or hardware address changes.
1653 mldv1_unsolicited_report_interval - INTEGER
1654 The interval in milliseconds in which the next unsolicited
1655 MLDv1 report retransmit will take place.
1656 Default: 10000 (10 seconds)
1658 mldv2_unsolicited_report_interval - INTEGER
1659 The interval in milliseconds in which the next unsolicited
1660 MLDv2 report retransmit will take place.
1661 Default: 1000 (1 second)
1663 force_mld_version - INTEGER
1664 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
1665 1 - Enforce to use MLD version 1
1666 2 - Enforce to use MLD version 2
1668 suppress_frag_ndisc - INTEGER
1669 Control RFC 6980 (Security Implications of IPv6 Fragmentation
1670 with IPv6 Neighbor Discovery) behavior:
1671 1 - (default) discard fragmented neighbor discovery packets
1672 0 - allow fragmented neighbor discovery packets
1674 optimistic_dad - BOOLEAN
1675 Whether to perform Optimistic Duplicate Address Detection (RFC 4429).
1676 0: disabled (default)
1679 use_optimistic - BOOLEAN
1680 If enabled, do not classify optimistic addresses as deprecated during
1681 source address selection. Preferred addresses will still be chosen
1682 before optimistic addresses, subject to other ranking in the source
1683 address selection algorithm.
1684 0: disabled (default)
1687 stable_secret - IPv6 address
1688 This IPv6 address will be used as a secret to generate IPv6
1689 addresses for link-local addresses and autoconfigured
1690 ones. All addresses generated after setting this secret will
1691 be stable privacy ones by default. This can be changed via the
1692 addrgenmode ip-link. conf/default/stable_secret is used as the
1693 secret for the namespace, the interface specific ones can
1694 overwrite that. Writes to conf/all/stable_secret are refused.
1696 It is recommended to generate this secret during installation
1697 of a system and keep it stable after that.
1699 By default the stable secret is unset.
1701 drop_unicast_in_l2_multicast - BOOLEAN
1702 Drop any unicast IPv6 packets that are received in link-layer
1703 multicast (or broadcast) frames.
1705 By default this is turned off.
1707 drop_unsolicited_na - BOOLEAN
1708 Drop all unsolicited neighbor advertisements, for example if there's
1709 a known good NA proxy on the network and such frames need not be used
1710 (or in the case of 802.11, must not be used to prevent attacks.)
1712 By default this is turned off.
1716 Limit the maximal rates for sending ICMPv6 packets.
1717 0 to disable any limiting,
1718 otherwise the minimal space between responses in milliseconds.
1721 xfrm6_gc_thresh - INTEGER
1722 The threshold at which we will start garbage collecting for IPv6
1723 destination cache entries. At twice this value the system will
1724 refuse new allocations. The value must be set below the flowcache
1725 limit (4096 * number of online cpus) to take effect.
1729 Pekka Savola <pekkas@netcore.fi>
1730 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1733 /proc/sys/net/bridge/* Variables:
1735 bridge-nf-call-arptables - BOOLEAN
1736 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1740 bridge-nf-call-iptables - BOOLEAN
1741 1 : pass bridged IPv4 traffic to iptables' chains.
1745 bridge-nf-call-ip6tables - BOOLEAN
1746 1 : pass bridged IPv6 traffic to ip6tables' chains.
1750 bridge-nf-filter-vlan-tagged - BOOLEAN
1751 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1755 bridge-nf-filter-pppoe-tagged - BOOLEAN
1756 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1760 bridge-nf-pass-vlan-input-dev - BOOLEAN
1761 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
1762 interface on the bridge and set the netfilter input device to the vlan.
1763 This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
1764 target work with vlan-on-top-of-bridge interfaces. When no matching
1765 vlan interface is found, or this switch is off, the input device is
1766 set to the bridge interface.
1767 0: disable bridge netfilter vlan interface lookup.
1770 proc/sys/net/sctp/* Variables:
1772 addip_enable - BOOLEAN
1773 Enable or disable extension of Dynamic Address Reconfiguration
1774 (ADD-IP) functionality specified in RFC5061. This extension provides
1775 the ability to dynamically add and remove new addresses for the SCTP
1778 1: Enable extension.
1780 0: Disable extension.
1785 Enable or disable pf (pf is short for potentially failed) state. A value
1786 of pf_retrans > path_max_retrans also disables pf state. That is, one of
1787 both pf_enable and pf_retrans > path_max_retrans can disable pf state.
1788 Since pf_retrans and path_max_retrans can be changed by userspace
1789 application, sometimes user expects to disable pf state by the value of
1790 pf_retrans > path_max_retrans, but occasionally the value of pf_retrans
1791 or path_max_retrans is changed by the user application, this pf state is
1792 enabled. As such, it is necessary to add this to dynamically enable
1793 and disable pf state. See:
1794 https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for
1803 addip_noauth_enable - BOOLEAN
1804 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1805 authentication to protect the operations of adding or removing new
1806 addresses. This requirement is mandated so that unauthorized hosts
1807 would not be able to hijack associations. However, older
1808 implementations may not have implemented this requirement while
1809 allowing the ADD-IP extension. For reasons of interoperability,
1810 we provide this variable to control the enforcement of the
1811 authentication requirement.
1813 1: Allow ADD-IP extension to be used without authentication. This
1814 should only be set in a closed environment for interoperability
1815 with older implementations.
1817 0: Enforce the authentication requirement
1821 auth_enable - BOOLEAN
1822 Enable or disable Authenticated Chunks extension. This extension
1823 provides the ability to send and receive authenticated chunks and is
1824 required for secure operation of Dynamic Address Reconfiguration
1827 1: Enable this extension.
1828 0: Disable this extension.
1832 prsctp_enable - BOOLEAN
1833 Enable or disable the Partial Reliability extension (RFC3758) which
1834 is used to notify peers that a given DATA should no longer be expected.
1842 The limit of the number of new packets that can be initially sent. It
1843 controls how bursty the generated traffic can be.
1847 association_max_retrans - INTEGER
1848 Set the maximum number for retransmissions that an association can
1849 attempt deciding that the remote end is unreachable. If this value
1850 is exceeded, the association is terminated.
1854 max_init_retransmits - INTEGER
1855 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1856 that an association will attempt before declaring the destination
1857 unreachable and terminating.
1861 path_max_retrans - INTEGER
1862 The maximum number of retransmissions that will be attempted on a given
1863 path. Once this threshold is exceeded, the path is considered
1864 unreachable, and new traffic will use a different path when the
1865 association is multihomed.
1869 pf_retrans - INTEGER
1870 The number of retransmissions that will be attempted on a given path
1871 before traffic is redirected to an alternate transport (should one
1872 exist). Note this is distinct from path_max_retrans, as a path that
1873 passes the pf_retrans threshold can still be used. Its only
1874 deprioritized when a transmission path is selected by the stack. This
1875 setting is primarily used to enable fast failover mechanisms without
1876 having to reduce path_max_retrans to a very low value. See:
1877 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
1878 for details. Note also that a value of pf_retrans > path_max_retrans
1879 disables this feature. Since both pf_retrans and path_max_retrans can
1880 be changed by userspace application, a variable pf_enable is used to
1885 rto_initial - INTEGER
1886 The initial round trip timeout value in milliseconds that will be used
1887 in calculating round trip times. This is the initial time interval
1888 for retransmissions.
1893 The maximum value (in milliseconds) of the round trip timeout. This
1894 is the largest time interval that can elapse between retransmissions.
1899 The minimum value (in milliseconds) of the round trip timeout. This
1900 is the smallest time interval the can elapse between retransmissions.
1904 hb_interval - INTEGER
1905 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
1906 are sent at the specified interval on idle paths to probe the state of
1907 a given path between 2 associations.
1911 sack_timeout - INTEGER
1912 The amount of time (in milliseconds) that the implementation will wait
1917 valid_cookie_life - INTEGER
1918 The default lifetime of the SCTP cookie (in milliseconds). The cookie
1919 is used during association establishment.
1923 cookie_preserve_enable - BOOLEAN
1924 Enable or disable the ability to extend the lifetime of the SCTP cookie
1925 that is used during the establishment phase of SCTP association
1927 1: Enable cookie lifetime extension.
1932 cookie_hmac_alg - STRING
1933 Select the hmac algorithm used when generating the cookie value sent by
1934 a listening sctp socket to a connecting client in the INIT-ACK chunk.
1939 Ability to assign md5 or sha1 as the selected alg is predicated on the
1940 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
1941 CONFIG_CRYPTO_SHA1).
1943 Default: Dependent on configuration. MD5 if available, else SHA1 if
1944 available, else none.
1946 rcvbuf_policy - INTEGER
1947 Determines if the receive buffer is attributed to the socket or to
1948 association. SCTP supports the capability to create multiple
1949 associations on a single socket. When using this capability, it is
1950 possible that a single stalled association that's buffering a lot
1951 of data may block other associations from delivering their data by
1952 consuming all of the receive buffer space. To work around this,
1953 the rcvbuf_policy could be set to attribute the receiver buffer space
1954 to each association instead of the socket. This prevents the described
1957 1: rcvbuf space is per association
1958 0: rcvbuf space is per socket
1962 sndbuf_policy - INTEGER
1963 Similar to rcvbuf_policy above, this applies to send buffer space.
1965 1: Send buffer is tracked per association
1966 0: Send buffer is tracked per socket.
1970 sctp_mem - vector of 3 INTEGERs: min, pressure, max
1971 Number of pages allowed for queueing by all SCTP sockets.
1973 min: Below this number of pages SCTP is not bothered about its
1974 memory appetite. When amount of memory allocated by SCTP exceeds
1975 this number, SCTP starts to moderate memory usage.
1977 pressure: This value was introduced to follow format of tcp_mem.
1979 max: Number of pages allowed for queueing by all SCTP sockets.
1981 Default is calculated at boot time from amount of available memory.
1983 sctp_rmem - vector of 3 INTEGERs: min, default, max
1984 Only the first value ("min") is used, "default" and "max" are
1987 min: Minimal size of receive buffer used by SCTP socket.
1988 It is guaranteed to each SCTP socket (but not association) even
1989 under moderate memory pressure.
1993 sctp_wmem - vector of 3 INTEGERs: min, default, max
1994 Currently this tunable has no effect.
1996 addr_scope_policy - INTEGER
1997 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
1999 0 - Disable IPv4 address scoping
2000 1 - Enable IPv4 address scoping
2001 2 - Follow draft but allow IPv4 private addresses
2002 3 - Follow draft but allow IPv4 link local addresses
2007 /proc/sys/net/core/*
2008 Please see: Documentation/sysctl/net.txt for descriptions of these entries.
2011 /proc/sys/net/unix/*
2012 max_dgram_qlen - INTEGER
2013 The maximum length of dgram socket receive queue
2020 /proc/sys/net/irda/*
2021 fast_poll_increase FIXME
2022 warn_noreply_time FIXME
2023 discovery_slots FIXME
2026 discovery_timeout FIXME
2027 lap_keepalive_time FIXME
2028 max_noreply_time FIXME
2029 max_tx_data_size FIXME
2031 min_tx_turn_time FIXME