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.
33 default 552 - minimum discovered Path MTU
35 ip_forward_use_pmtu - BOOLEAN
36 By default we don't trust protocol path MTUs while forwarding
37 because they could be easily forged and can lead to unwanted
38 fragmentation by the router.
39 You only need to enable this if you have user-space software
40 which tries to discover path mtus by itself and depends on the
41 kernel honoring this information. This is normally not the
48 route/max_size - INTEGER
49 Maximum number of routes allowed in the kernel. Increase
50 this when using large numbers of interfaces and/or routes.
52 neigh/default/gc_thresh1 - INTEGER
53 Minimum number of entries to keep. Garbage collector will not
54 purge entries if there are fewer than this number.
57 neigh/default/gc_thresh3 - INTEGER
58 Maximum number of neighbor entries allowed. Increase this
59 when using large numbers of interfaces and when communicating
60 with large numbers of directly-connected peers.
63 neigh/default/unres_qlen_bytes - INTEGER
64 The maximum number of bytes which may be used by packets
65 queued for each unresolved address by other network layers.
67 Setting negative value is meaningless and will return error.
68 Default: 65536 Bytes(64KB)
70 neigh/default/unres_qlen - INTEGER
71 The maximum number of packets which may be queued for each
72 unresolved address by other network layers.
73 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
74 Prior to linux 3.3, the default value is 3 which may cause
75 unexpected packet loss. The current default value is calculated
76 according to default value of unres_qlen_bytes and true size of
81 Time, in seconds, that cached PMTU information is kept.
84 The advertised MSS depends on the first hop route MTU, but will
85 never be lower than this setting.
89 ipfrag_high_thresh - INTEGER
90 Maximum memory used to reassemble IP fragments. When
91 ipfrag_high_thresh bytes of memory is allocated for this purpose,
92 the fragment handler will toss packets until ipfrag_low_thresh
95 ipfrag_low_thresh - INTEGER
96 See ipfrag_high_thresh
99 Time in seconds to keep an IP fragment in memory.
101 ipfrag_secret_interval - INTEGER
102 Regeneration interval (in seconds) of the hash secret (or lifetime
103 for the hash secret) for IP fragments.
106 ipfrag_max_dist - INTEGER
107 ipfrag_max_dist is a non-negative integer value which defines the
108 maximum "disorder" which is allowed among fragments which share a
109 common IP source address. Note that reordering of packets is
110 not unusual, but if a large number of fragments arrive from a source
111 IP address while a particular fragment queue remains incomplete, it
112 probably indicates that one or more fragments belonging to that queue
113 have been lost. When ipfrag_max_dist is positive, an additional check
114 is done on fragments before they are added to a reassembly queue - if
115 ipfrag_max_dist (or more) fragments have arrived from a particular IP
116 address between additions to any IP fragment queue using that source
117 address, it's presumed that one or more fragments in the queue are
118 lost. The existing fragment queue will be dropped, and a new one
119 started. An ipfrag_max_dist value of zero disables this check.
121 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
122 result in unnecessarily dropping fragment queues when normal
123 reordering of packets occurs, which could lead to poor application
124 performance. Using a very large value, e.g. 50000, increases the
125 likelihood of incorrectly reassembling IP fragments that originate
126 from different IP datagrams, which could result in data corruption.
131 inet_peer_threshold - INTEGER
132 The approximate size of the storage. Starting from this threshold
133 entries will be thrown aggressively. This threshold also determines
134 entries' time-to-live and time intervals between garbage collection
135 passes. More entries, less time-to-live, less GC interval.
137 inet_peer_minttl - INTEGER
138 Minimum time-to-live of entries. Should be enough to cover fragment
139 time-to-live on the reassembling side. This minimum time-to-live is
140 guaranteed if the pool size is less than inet_peer_threshold.
143 inet_peer_maxttl - INTEGER
144 Maximum time-to-live of entries. Unused entries will expire after
145 this period of time if there is no memory pressure on the pool (i.e.
146 when the number of entries in the pool is very small).
152 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
153 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
156 tcp_abort_on_overflow - BOOLEAN
157 If listening service is too slow to accept new connections,
158 reset them. Default state is FALSE. It means that if overflow
159 occurred due to a burst, connection will recover. Enable this
160 option _only_ if you are really sure that listening daemon
161 cannot be tuned to accept connections faster. Enabling this
162 option can harm clients of your server.
164 tcp_adv_win_scale - INTEGER
165 Count buffering overhead as bytes/2^tcp_adv_win_scale
166 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
168 Possible values are [-31, 31], inclusive.
171 tcp_allowed_congestion_control - STRING
172 Show/set the congestion control choices available to non-privileged
173 processes. The list is a subset of those listed in
174 tcp_available_congestion_control.
175 Default is "reno" and the default setting (tcp_congestion_control).
177 tcp_app_win - INTEGER
178 Reserve max(window/2^tcp_app_win, mss) of window for application
179 buffer. Value 0 is special, it means that nothing is reserved.
182 tcp_autocorking - BOOLEAN
183 Enable TCP auto corking :
184 When applications do consecutive small write()/sendmsg() system calls,
185 we try to coalesce these small writes as much as possible, to lower
186 total amount of sent packets. This is done if at least one prior
187 packet for the flow is waiting in Qdisc queues or device transmit
188 queue. Applications can still use TCP_CORK for optimal behavior
189 when they know how/when to uncork their sockets.
192 tcp_available_congestion_control - STRING
193 Shows the available congestion control choices that are registered.
194 More congestion control algorithms may be available as modules,
197 tcp_base_mss - INTEGER
198 The initial value of search_low to be used by the packetization layer
199 Path MTU discovery (MTU probing). If MTU probing is enabled,
200 this is the initial MSS used by the connection.
202 tcp_congestion_control - STRING
203 Set the congestion control algorithm to be used for new
204 connections. The algorithm "reno" is always available, but
205 additional choices may be available based on kernel configuration.
206 Default is set as part of kernel configuration.
207 For passive connections, the listener congestion control choice
209 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
212 Allows TCP to send "duplicate" SACKs.
214 tcp_early_retrans - INTEGER
215 Enable Early Retransmit (ER), per RFC 5827. ER lowers the threshold
216 for triggering fast retransmit when the amount of outstanding data is
217 small and when no previously unsent data can be transmitted (such
218 that limited transmit could be used). Also controls the use of
219 Tail loss probe (TLP) that converts RTOs occurring due to tail
220 losses into fast recovery (draft-dukkipati-tcpm-tcp-loss-probe-01).
224 2 enables ER but delays fast recovery and fast retransmit
225 by a fourth of RTT. This mitigates connection falsely
226 recovers when network has a small degree of reordering
227 (less than 3 packets).
228 3 enables delayed ER and TLP.
233 Control use of Explicit Congestion Notification (ECN) by TCP.
234 ECN is used only when both ends of the TCP connection indicate
235 support for it. This feature is useful in avoiding losses due
236 to congestion by allowing supporting routers to signal
237 congestion before having to drop packets.
239 0 Disable ECN. Neither initiate nor accept ECN.
240 1 Enable ECN when requested by incoming connections and
241 also request ECN on outgoing connection attempts.
242 2 Enable ECN when requested by incoming connections
243 but do not request ECN on outgoing connections.
247 Enable FACK congestion avoidance and fast retransmission.
248 The value is not used, if tcp_sack is not enabled.
250 tcp_fin_timeout - INTEGER
251 The length of time an orphaned (no longer referenced by any
252 application) connection will remain in the FIN_WAIT_2 state
253 before it is aborted at the local end. While a perfectly
254 valid "receive only" state for an un-orphaned connection, an
255 orphaned connection in FIN_WAIT_2 state could otherwise wait
256 forever for the remote to close its end of the connection.
261 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
262 F-RTO is an enhanced recovery algorithm for TCP retransmission
263 timeouts. It is particularly beneficial in networks where the
264 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
265 modification. It does not require any support from the peer.
267 By default it's enabled with a non-zero value. 0 disables F-RTO.
269 tcp_keepalive_time - INTEGER
270 How often TCP sends out keepalive messages when keepalive is enabled.
273 tcp_keepalive_probes - INTEGER
274 How many keepalive probes TCP sends out, until it decides that the
275 connection is broken. Default value: 9.
277 tcp_keepalive_intvl - INTEGER
278 How frequently the probes are send out. Multiplied by
279 tcp_keepalive_probes it is time to kill not responding connection,
280 after probes started. Default value: 75sec i.e. connection
281 will be aborted after ~11 minutes of retries.
283 tcp_low_latency - BOOLEAN
284 If set, the TCP stack makes decisions that prefer lower
285 latency as opposed to higher throughput. By default, this
286 option is not set meaning that higher throughput is preferred.
287 An example of an application where this default should be
288 changed would be a Beowulf compute cluster.
291 tcp_max_orphans - INTEGER
292 Maximal number of TCP sockets not attached to any user file handle,
293 held by system. If this number is exceeded orphaned connections are
294 reset immediately and warning is printed. This limit exists
295 only to prevent simple DoS attacks, you _must_ not rely on this
296 or lower the limit artificially, but rather increase it
297 (probably, after increasing installed memory),
298 if network conditions require more than default value,
299 and tune network services to linger and kill such states
300 more aggressively. Let me to remind again: each orphan eats
301 up to ~64K of unswappable memory.
303 tcp_max_syn_backlog - INTEGER
304 Maximal number of remembered connection requests, which have not
305 received an acknowledgment from connecting client.
306 The minimal value is 128 for low memory machines, and it will
307 increase in proportion to the memory of machine.
308 If server suffers from overload, try increasing this number.
310 tcp_max_tw_buckets - INTEGER
311 Maximal number of timewait sockets held by system simultaneously.
312 If this number is exceeded time-wait socket is immediately destroyed
313 and warning is printed. This limit exists only to prevent
314 simple DoS attacks, you _must_ not lower the limit artificially,
315 but rather increase it (probably, after increasing installed memory),
316 if network conditions require more than default value.
318 tcp_mem - vector of 3 INTEGERs: min, pressure, max
319 min: below this number of pages TCP is not bothered about its
322 pressure: when amount of memory allocated by TCP exceeds this number
323 of pages, TCP moderates its memory consumption and enters memory
324 pressure mode, which is exited when memory consumption falls
327 max: number of pages allowed for queueing by all TCP sockets.
329 Defaults are calculated at boot time from amount of available
332 tcp_moderate_rcvbuf - BOOLEAN
333 If set, TCP performs receive buffer auto-tuning, attempting to
334 automatically size the buffer (no greater than tcp_rmem[2]) to
335 match the size required by the path for full throughput. Enabled by
338 tcp_mtu_probing - INTEGER
339 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
342 1 - Disabled by default, enabled when an ICMP black hole detected
343 2 - Always enabled, use initial MSS of tcp_base_mss.
345 tcp_no_metrics_save - BOOLEAN
346 By default, TCP saves various connection metrics in the route cache
347 when the connection closes, so that connections established in the
348 near future can use these to set initial conditions. Usually, this
349 increases overall performance, but may sometimes cause performance
350 degradation. If set, TCP will not cache metrics on closing
353 tcp_orphan_retries - INTEGER
354 This value influences the timeout of a locally closed TCP connection,
355 when RTO retransmissions remain unacknowledged.
356 See tcp_retries2 for more details.
358 The default value is 8.
359 If your machine is a loaded WEB server,
360 you should think about lowering this value, such sockets
361 may consume significant resources. Cf. tcp_max_orphans.
363 tcp_reordering - INTEGER
364 Maximal reordering of packets in a TCP stream.
367 tcp_retrans_collapse - BOOLEAN
368 Bug-to-bug compatibility with some broken printers.
369 On retransmit try to send bigger packets to work around bugs in
372 tcp_retries1 - INTEGER
373 This value influences the time, after which TCP decides, that
374 something is wrong due to unacknowledged RTO retransmissions,
375 and reports this suspicion to the network layer.
376 See tcp_retries2 for more details.
378 RFC 1122 recommends at least 3 retransmissions, which is the
381 tcp_retries2 - INTEGER
382 This value influences the timeout of an alive TCP connection,
383 when RTO retransmissions remain unacknowledged.
384 Given a value of N, a hypothetical TCP connection following
385 exponential backoff with an initial RTO of TCP_RTO_MIN would
386 retransmit N times before killing the connection at the (N+1)th RTO.
388 The default value of 15 yields a hypothetical timeout of 924.6
389 seconds and is a lower bound for the effective timeout.
390 TCP will effectively time out at the first RTO which exceeds the
391 hypothetical timeout.
393 RFC 1122 recommends at least 100 seconds for the timeout,
394 which corresponds to a value of at least 8.
396 tcp_rfc1337 - BOOLEAN
397 If set, the TCP stack behaves conforming to RFC1337. If unset,
398 we are not conforming to RFC, but prevent TCP TIME_WAIT
402 tcp_rmem - vector of 3 INTEGERs: min, default, max
403 min: Minimal size of receive buffer used by TCP sockets.
404 It is guaranteed to each TCP socket, even under moderate memory
408 default: initial size of receive buffer used by TCP sockets.
409 This value overrides net.core.rmem_default used by other protocols.
410 Default: 87380 bytes. This value results in window of 65535 with
411 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
412 less for default tcp_app_win. See below about these variables.
414 max: maximal size of receive buffer allowed for automatically
415 selected receiver buffers for TCP socket. This value does not override
416 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
417 automatic tuning of that socket's receive buffer size, in which
418 case this value is ignored.
419 Default: between 87380B and 6MB, depending on RAM size.
422 Enable select acknowledgments (SACKS).
424 tcp_slow_start_after_idle - BOOLEAN
425 If set, provide RFC2861 behavior and time out the congestion
426 window after an idle period. An idle period is defined at
427 the current RTO. If unset, the congestion window will not
428 be timed out after an idle period.
432 Use the Host requirements interpretation of the TCP urgent pointer field.
433 Most hosts use the older BSD interpretation, so if you turn this on
434 Linux might not communicate correctly with them.
437 tcp_synack_retries - INTEGER
438 Number of times SYNACKs for a passive TCP connection attempt will
439 be retransmitted. Should not be higher than 255. Default value
440 is 5, which corresponds to 31seconds till the last retransmission
441 with the current initial RTO of 1second. With this the final timeout
442 for a passive TCP connection will happen after 63seconds.
444 tcp_syncookies - BOOLEAN
445 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
446 Send out syncookies when the syn backlog queue of a socket
447 overflows. This is to prevent against the common 'SYN flood attack'
450 Note, that syncookies is fallback facility.
451 It MUST NOT be used to help highly loaded servers to stand
452 against legal connection rate. If you see SYN flood warnings
453 in your logs, but investigation shows that they occur
454 because of overload with legal connections, you should tune
455 another parameters until this warning disappear.
456 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
458 syncookies seriously violate TCP protocol, do not allow
459 to use TCP extensions, can result in serious degradation
460 of some services (f.e. SMTP relaying), visible not by you,
461 but your clients and relays, contacting you. While you see
462 SYN flood warnings in logs not being really flooded, your server
463 is seriously misconfigured.
465 If you want to test which effects syncookies have to your
466 network connections you can set this knob to 2 to enable
467 unconditionally generation of syncookies.
469 tcp_fastopen - INTEGER
470 Enable TCP Fast Open feature (draft-ietf-tcpm-fastopen) to send data
471 in the opening SYN packet. To use this feature, the client application
472 must use sendmsg() or sendto() with MSG_FASTOPEN flag rather than
473 connect() to perform a TCP handshake automatically.
475 The values (bitmap) are
476 1: Enables sending data in the opening SYN on the client w/ MSG_FASTOPEN.
477 2: Enables TCP Fast Open on the server side, i.e., allowing data in
478 a SYN packet to be accepted and passed to the application before
479 3-way hand shake finishes.
480 4: Send data in the opening SYN regardless of cookie availability and
481 without a cookie option.
482 0x100: Accept SYN data w/o validating the cookie.
483 0x200: Accept data-in-SYN w/o any cookie option present.
484 0x400/0x800: Enable Fast Open on all listeners regardless of the
485 TCP_FASTOPEN socket option. The two different flags designate two
486 different ways of setting max_qlen without the TCP_FASTOPEN socket
491 Note that the client & server side Fast Open flags (1 and 2
492 respectively) must be also enabled before the rest of flags can take
495 See include/net/tcp.h and the code for more details.
497 tcp_syn_retries - INTEGER
498 Number of times initial SYNs for an active TCP connection attempt
499 will be retransmitted. Should not be higher than 255. Default value
500 is 6, which corresponds to 63seconds till the last retransmission
501 with the current initial RTO of 1second. With this the final timeout
502 for an active TCP connection attempt will happen after 127seconds.
504 tcp_timestamps - BOOLEAN
505 Enable timestamps as defined in RFC1323.
507 tcp_min_tso_segs - INTEGER
508 Minimal number of segments per TSO frame.
509 Since linux-3.12, TCP does an automatic sizing of TSO frames,
510 depending on flow rate, instead of filling 64Kbytes packets.
511 For specific usages, it's possible to force TCP to build big
512 TSO frames. Note that TCP stack might split too big TSO packets
513 if available window is too small.
516 tcp_tso_win_divisor - INTEGER
517 This allows control over what percentage of the congestion window
518 can be consumed by a single TSO frame.
519 The setting of this parameter is a choice between burstiness and
520 building larger TSO frames.
523 tcp_tw_recycle - BOOLEAN
524 Enable fast recycling TIME-WAIT sockets. Default value is 0.
525 It should not be changed without advice/request of technical
528 tcp_tw_reuse - BOOLEAN
529 Allow to reuse TIME-WAIT sockets for new connections when it is
530 safe from protocol viewpoint. Default value is 0.
531 It should not be changed without advice/request of technical
534 tcp_window_scaling - BOOLEAN
535 Enable window scaling as defined in RFC1323.
537 tcp_wmem - vector of 3 INTEGERs: min, default, max
538 min: Amount of memory reserved for send buffers for TCP sockets.
539 Each TCP socket has rights to use it due to fact of its birth.
542 default: initial size of send buffer used by TCP sockets. This
543 value overrides net.core.wmem_default used by other protocols.
544 It is usually lower than net.core.wmem_default.
547 max: Maximal amount of memory allowed for automatically tuned
548 send buffers for TCP sockets. This value does not override
549 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
550 automatic tuning of that socket's send buffer size, in which case
551 this value is ignored.
552 Default: between 64K and 4MB, depending on RAM size.
554 tcp_notsent_lowat - UNSIGNED INTEGER
555 A TCP socket can control the amount of unsent bytes in its write queue,
556 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
557 reports POLLOUT events if the amount of unsent bytes is below a per
558 socket value, and if the write queue is not full. sendmsg() will
559 also not add new buffers if the limit is hit.
561 This global variable controls the amount of unsent data for
562 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
563 to the global variable has immediate effect.
565 Default: UINT_MAX (0xFFFFFFFF)
567 tcp_workaround_signed_windows - BOOLEAN
568 If set, assume no receipt of a window scaling option means the
569 remote TCP is broken and treats the window as a signed quantity.
570 If unset, assume the remote TCP is not broken even if we do
571 not receive a window scaling option from them.
574 tcp_dma_copybreak - INTEGER
575 Lower limit, in bytes, of the size of socket reads that will be
576 offloaded to a DMA copy engine, if one is present in the system
577 and CONFIG_NET_DMA is enabled.
580 tcp_thin_linear_timeouts - BOOLEAN
581 Enable dynamic triggering of linear timeouts for thin streams.
582 If set, a check is performed upon retransmission by timeout to
583 determine if the stream is thin (less than 4 packets in flight).
584 As long as the stream is found to be thin, up to 6 linear
585 timeouts may be performed before exponential backoff mode is
586 initiated. This improves retransmission latency for
587 non-aggressive thin streams, often found to be time-dependent.
588 For more information on thin streams, see
589 Documentation/networking/tcp-thin.txt
592 tcp_thin_dupack - BOOLEAN
593 Enable dynamic triggering of retransmissions after one dupACK
594 for thin streams. If set, a check is performed upon reception
595 of a dupACK to determine if the stream is thin (less than 4
596 packets in flight). As long as the stream is found to be thin,
597 data is retransmitted on the first received dupACK. This
598 improves retransmission latency for non-aggressive thin
599 streams, often found to be time-dependent.
600 For more information on thin streams, see
601 Documentation/networking/tcp-thin.txt
604 tcp_limit_output_bytes - INTEGER
605 Controls TCP Small Queue limit per tcp socket.
606 TCP bulk sender tends to increase packets in flight until it
607 gets losses notifications. With SNDBUF autotuning, this can
608 result in a large amount of packets queued in qdisc/device
609 on the local machine, hurting latency of other flows, for
610 typical pfifo_fast qdiscs.
611 tcp_limit_output_bytes limits the number of bytes on qdisc
612 or device to reduce artificial RTT/cwnd and reduce bufferbloat.
615 tcp_challenge_ack_limit - INTEGER
616 Limits number of Challenge ACK sent per second, as recommended
617 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
622 udp_mem - vector of 3 INTEGERs: min, pressure, max
623 Number of pages allowed for queueing by all UDP sockets.
625 min: Below this number of pages UDP is not bothered about its
626 memory appetite. When amount of memory allocated by UDP exceeds
627 this number, UDP starts to moderate memory usage.
629 pressure: This value was introduced to follow format of tcp_mem.
631 max: Number of pages allowed for queueing by all UDP sockets.
633 Default is calculated at boot time from amount of available memory.
635 udp_rmem_min - INTEGER
636 Minimal size of receive buffer used by UDP sockets in moderation.
637 Each UDP socket is able to use the size for receiving data, even if
638 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
641 udp_wmem_min - INTEGER
642 Minimal size of send buffer used by UDP sockets in moderation.
643 Each UDP socket is able to use the size for sending data, even if
644 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
649 cipso_cache_enable - BOOLEAN
650 If set, enable additions to and lookups from the CIPSO label mapping
651 cache. If unset, additions are ignored and lookups always result in a
652 miss. However, regardless of the setting the cache is still
653 invalidated when required when means you can safely toggle this on and
654 off and the cache will always be "safe".
657 cipso_cache_bucket_size - INTEGER
658 The CIPSO label cache consists of a fixed size hash table with each
659 hash bucket containing a number of cache entries. This variable limits
660 the number of entries in each hash bucket; the larger the value the
661 more CIPSO label mappings that can be cached. When the number of
662 entries in a given hash bucket reaches this limit adding new entries
663 causes the oldest entry in the bucket to be removed to make room.
666 cipso_rbm_optfmt - BOOLEAN
667 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
668 the CIPSO draft specification (see Documentation/netlabel for details).
669 This means that when set the CIPSO tag will be padded with empty
670 categories in order to make the packet data 32-bit aligned.
673 cipso_rbm_structvalid - BOOLEAN
674 If set, do a very strict check of the CIPSO option when
675 ip_options_compile() is called. If unset, relax the checks done during
676 ip_options_compile(). Either way is "safe" as errors are caught else
677 where in the CIPSO processing code but setting this to 0 (False) should
678 result in less work (i.e. it should be faster) but could cause problems
679 with other implementations that require strict checking.
684 ip_local_port_range - 2 INTEGERS
685 Defines the local port range that is used by TCP and UDP to
686 choose the local port. The first number is the first, the
687 second the last local port number. The default values are
688 32768 and 61000 respectively.
690 ip_local_reserved_ports - list of comma separated ranges
691 Specify the ports which are reserved for known third-party
692 applications. These ports will not be used by automatic port
693 assignments (e.g. when calling connect() or bind() with port
694 number 0). Explicit port allocation behavior is unchanged.
696 The format used for both input and output is a comma separated
697 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
698 10). Writing to the file will clear all previously reserved
699 ports and update the current list with the one given in the
702 Note that ip_local_port_range and ip_local_reserved_ports
703 settings are independent and both are considered by the kernel
704 when determining which ports are available for automatic port
707 You can reserve ports which are not in the current
708 ip_local_port_range, e.g.:
710 $ cat /proc/sys/net/ipv4/ip_local_port_range
712 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
715 although this is redundant. However such a setting is useful
716 if later the port range is changed to a value that will
717 include the reserved ports.
721 ip_nonlocal_bind - BOOLEAN
722 If set, allows processes to bind() to non-local IP addresses,
723 which can be quite useful - but may break some applications.
727 If set non-zero, enables support for dynamic addresses.
728 If set to a non-zero value larger than 1, a kernel log
729 message will be printed when dynamic address rewriting
733 ip_early_demux - BOOLEAN
734 Optimize input packet processing down to one demux for
735 certain kinds of local sockets. Currently we only do this
736 for established TCP sockets.
738 It may add an additional cost for pure routing workloads that
739 reduces overall throughput, in such case you should disable it.
742 icmp_echo_ignore_all - BOOLEAN
743 If set non-zero, then the kernel will ignore all ICMP ECHO
747 icmp_echo_ignore_broadcasts - BOOLEAN
748 If set non-zero, then the kernel will ignore all ICMP ECHO and
749 TIMESTAMP requests sent to it via broadcast/multicast.
752 icmp_ratelimit - INTEGER
753 Limit the maximal rates for sending ICMP packets whose type matches
754 icmp_ratemask (see below) to specific targets.
755 0 to disable any limiting,
756 otherwise the minimal space between responses in milliseconds.
759 icmp_ratemask - INTEGER
760 Mask made of ICMP types for which rates are being limited.
761 Significant bits: IHGFEDCBA9876543210
762 Default mask: 0000001100000011000 (6168)
764 Bit definitions (see include/linux/icmp.h):
766 3 Destination Unreachable *
771 C Parameter Problem *
776 H Address Mask Request
779 * These are rate limited by default (see default mask above)
781 icmp_ignore_bogus_error_responses - BOOLEAN
782 Some routers violate RFC1122 by sending bogus responses to broadcast
783 frames. Such violations are normally logged via a kernel warning.
784 If this is set to TRUE, the kernel will not give such warnings, which
785 will avoid log file clutter.
788 icmp_errors_use_inbound_ifaddr - BOOLEAN
790 If zero, icmp error messages are sent with the primary address of
791 the exiting interface.
793 If non-zero, the message will be sent with the primary address of
794 the interface that received the packet that caused the icmp error.
795 This is the behaviour network many administrators will expect from
796 a router. And it can make debugging complicated network layouts
799 Note that if no primary address exists for the interface selected,
800 then the primary address of the first non-loopback interface that
801 has one will be used regardless of this setting.
805 igmp_max_memberships - INTEGER
806 Change the maximum number of multicast groups we can subscribe to.
809 Theoretical maximum value is bounded by having to send a membership
810 report in a single datagram (i.e. the report can't span multiple
811 datagrams, or risk confusing the switch and leaving groups you don't
814 The number of supported groups 'M' is bounded by the number of group
815 report entries you can fit into a single datagram of 65535 bytes.
817 M = 65536-sizeof (ip header)/(sizeof(Group record))
819 Group records are variable length, with a minimum of 12 bytes.
820 So net.ipv4.igmp_max_memberships should not be set higher than:
822 (65536-24) / 12 = 5459
824 The value 5459 assumes no IP header options, so in practice
825 this number may be lower.
827 conf/interface/* changes special settings per interface (where
828 "interface" is the name of your network interface)
830 conf/all/* is special, changes the settings for all interfaces
832 log_martians - BOOLEAN
833 Log packets with impossible addresses to kernel log.
834 log_martians for the interface will be enabled if at least one of
835 conf/{all,interface}/log_martians is set to TRUE,
836 it will be disabled otherwise
838 accept_redirects - BOOLEAN
839 Accept ICMP redirect messages.
840 accept_redirects for the interface will be enabled if:
841 - both conf/{all,interface}/accept_redirects are TRUE in the case
842 forwarding for the interface is enabled
844 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
845 case forwarding for the interface is disabled
846 accept_redirects for the interface will be disabled otherwise
851 Enable IP forwarding on this interface.
853 mc_forwarding - BOOLEAN
854 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
855 and a multicast routing daemon is required.
856 conf/all/mc_forwarding must also be set to TRUE to enable multicast
857 routing for the interface
860 Integer value used to differentiate the devices by the medium they
861 are attached to. Two devices can have different id values when
862 the broadcast packets are received only on one of them.
863 The default value 0 means that the device is the only interface
864 to its medium, value of -1 means that medium is not known.
866 Currently, it is used to change the proxy_arp behavior:
867 the proxy_arp feature is enabled for packets forwarded between
868 two devices attached to different media.
872 proxy_arp for the interface will be enabled if at least one of
873 conf/{all,interface}/proxy_arp is set to TRUE,
874 it will be disabled otherwise
876 proxy_arp_pvlan - BOOLEAN
877 Private VLAN proxy arp.
878 Basically allow proxy arp replies back to the same interface
879 (from which the ARP request/solicitation was received).
881 This is done to support (ethernet) switch features, like RFC
882 3069, where the individual ports are NOT allowed to
883 communicate with each other, but they are allowed to talk to
884 the upstream router. As described in RFC 3069, it is possible
885 to allow these hosts to communicate through the upstream
886 router by proxy_arp'ing. Don't need to be used together with
889 This technology is known by different names:
890 In RFC 3069 it is called VLAN Aggregation.
891 Cisco and Allied Telesyn call it Private VLAN.
892 Hewlett-Packard call it Source-Port filtering or port-isolation.
893 Ericsson call it MAC-Forced Forwarding (RFC Draft).
895 shared_media - BOOLEAN
896 Send(router) or accept(host) RFC1620 shared media redirects.
897 Overrides ip_secure_redirects.
898 shared_media for the interface will be enabled if at least one of
899 conf/{all,interface}/shared_media is set to TRUE,
900 it will be disabled otherwise
903 secure_redirects - BOOLEAN
904 Accept ICMP redirect messages only for gateways,
905 listed in default gateway list.
906 secure_redirects for the interface will be enabled if at least one of
907 conf/{all,interface}/secure_redirects is set to TRUE,
908 it will be disabled otherwise
911 send_redirects - BOOLEAN
912 Send redirects, if router.
913 send_redirects for the interface will be enabled if at least one of
914 conf/{all,interface}/send_redirects is set to TRUE,
915 it will be disabled otherwise
918 bootp_relay - BOOLEAN
919 Accept packets with source address 0.b.c.d destined
920 not to this host as local ones. It is supposed, that
921 BOOTP relay daemon will catch and forward such packets.
922 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
927 accept_source_route - BOOLEAN
928 Accept packets with SRR option.
929 conf/all/accept_source_route must also be set to TRUE to accept packets
930 with SRR option on the interface
931 default TRUE (router)
934 accept_local - BOOLEAN
935 Accept packets with local source addresses. In combination
936 with suitable routing, this can be used to direct packets
937 between two local interfaces over the wire and have them
940 rp_filter must be set to a non-zero value in order for
941 accept_local to have an effect.
945 route_localnet - BOOLEAN
946 Do not consider loopback addresses as martian source or destination
947 while routing. This enables the use of 127/8 for local routing purposes.
951 0 - No source validation.
952 1 - Strict mode as defined in RFC3704 Strict Reverse Path
953 Each incoming packet is tested against the FIB and if the interface
954 is not the best reverse path the packet check will fail.
955 By default failed packets are discarded.
956 2 - Loose mode as defined in RFC3704 Loose Reverse Path
957 Each incoming packet's source address is also tested against the FIB
958 and if the source address is not reachable via any interface
959 the packet check will fail.
961 Current recommended practice in RFC3704 is to enable strict mode
962 to prevent IP spoofing from DDos attacks. If using asymmetric routing
963 or other complicated routing, then loose mode is recommended.
965 The max value from conf/{all,interface}/rp_filter is used
966 when doing source validation on the {interface}.
968 Default value is 0. Note that some distributions enable it
972 1 - Allows you to have multiple network interfaces on the same
973 subnet, and have the ARPs for each interface be answered
974 based on whether or not the kernel would route a packet from
975 the ARP'd IP out that interface (therefore you must use source
976 based routing for this to work). In other words it allows control
977 of which cards (usually 1) will respond to an arp request.
979 0 - (default) The kernel can respond to arp requests with addresses
980 from other interfaces. This may seem wrong but it usually makes
981 sense, because it increases the chance of successful communication.
982 IP addresses are owned by the complete host on Linux, not by
983 particular interfaces. Only for more complex setups like load-
984 balancing, does this behaviour cause problems.
986 arp_filter for the interface will be enabled if at least one of
987 conf/{all,interface}/arp_filter is set to TRUE,
988 it will be disabled otherwise
990 arp_announce - INTEGER
991 Define different restriction levels for announcing the local
992 source IP address from IP packets in ARP requests sent on
994 0 - (default) Use any local address, configured on any interface
995 1 - Try to avoid local addresses that are not in the target's
996 subnet for this interface. This mode is useful when target
997 hosts reachable via this interface require the source IP
998 address in ARP requests to be part of their logical network
999 configured on the receiving interface. When we generate the
1000 request we will check all our subnets that include the
1001 target IP and will preserve the source address if it is from
1002 such subnet. If there is no such subnet we select source
1003 address according to the rules for level 2.
1004 2 - Always use the best local address for this target.
1005 In this mode we ignore the source address in the IP packet
1006 and try to select local address that we prefer for talks with
1007 the target host. Such local address is selected by looking
1008 for primary IP addresses on all our subnets on the outgoing
1009 interface that include the target IP address. If no suitable
1010 local address is found we select the first local address
1011 we have on the outgoing interface or on all other interfaces,
1012 with the hope we will receive reply for our request and
1013 even sometimes no matter the source IP address we announce.
1015 The max value from conf/{all,interface}/arp_announce is used.
1017 Increasing the restriction level gives more chance for
1018 receiving answer from the resolved target while decreasing
1019 the level announces more valid sender's information.
1021 arp_ignore - INTEGER
1022 Define different modes for sending replies in response to
1023 received ARP requests that resolve local target IP addresses:
1024 0 - (default): reply for any local target IP address, configured
1026 1 - reply only if the target IP address is local address
1027 configured on the incoming interface
1028 2 - reply only if the target IP address is local address
1029 configured on the incoming interface and both with the
1030 sender's IP address are part from same subnet on this interface
1031 3 - do not reply for local addresses configured with scope host,
1032 only resolutions for global and link addresses are replied
1034 8 - do not reply for all local addresses
1036 The max value from conf/{all,interface}/arp_ignore is used
1037 when ARP request is received on the {interface}
1039 arp_notify - BOOLEAN
1040 Define mode for notification of address and device changes.
1041 0 - (default): do nothing
1042 1 - Generate gratuitous arp requests when device is brought up
1043 or hardware address changes.
1045 arp_accept - BOOLEAN
1046 Define behavior for gratuitous ARP frames who's IP is not
1047 already present in the ARP table:
1048 0 - don't create new entries in the ARP table
1049 1 - create new entries in the ARP table
1051 Both replies and requests type gratuitous arp will trigger the
1052 ARP table to be updated, if this setting is on.
1054 If the ARP table already contains the IP address of the
1055 gratuitous arp frame, the arp table will be updated regardless
1056 if this setting is on or off.
1059 app_solicit - INTEGER
1060 The maximum number of probes to send to the user space ARP daemon
1061 via netlink before dropping back to multicast probes (see
1062 mcast_solicit). Defaults to 0.
1064 disable_policy - BOOLEAN
1065 Disable IPSEC policy (SPD) for this interface
1067 disable_xfrm - BOOLEAN
1068 Disable IPSEC encryption on this interface, whatever the policy
1070 igmpv2_unsolicited_report_interval - INTEGER
1071 The interval in milliseconds in which the next unsolicited
1072 IGMPv1 or IGMPv2 report retransmit will take place.
1073 Default: 10000 (10 seconds)
1075 igmpv3_unsolicited_report_interval - INTEGER
1076 The interval in milliseconds in which the next unsolicited
1077 IGMPv3 report retransmit will take place.
1078 Default: 1000 (1 seconds)
1081 Allows you to write a number, which can be used as required.
1085 kuznet@ms2.inr.ac.ru
1091 delon.nicolas@wanadoo.fr
1096 /proc/sys/net/ipv6/* Variables:
1098 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1099 apply to IPv6 [XXX?].
1101 bindv6only - BOOLEAN
1102 Default value for IPV6_V6ONLY socket option,
1103 which restricts use of the IPv6 socket to IPv6 communication
1105 TRUE: disable IPv4-mapped address feature
1106 FALSE: enable IPv4-mapped address feature
1108 Default: FALSE (as specified in RFC3493)
1110 anycast_src_echo_reply - BOOLEAN
1111 Controls the use of anycast addresses as source addresses for ICMPv6
1119 ip6frag_high_thresh - INTEGER
1120 Maximum memory used to reassemble IPv6 fragments. When
1121 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1122 the fragment handler will toss packets until ip6frag_low_thresh
1125 ip6frag_low_thresh - INTEGER
1126 See ip6frag_high_thresh
1128 ip6frag_time - INTEGER
1129 Time in seconds to keep an IPv6 fragment in memory.
1131 ip6frag_secret_interval - INTEGER
1132 Regeneration interval (in seconds) of the hash secret (or lifetime
1133 for the hash secret) for IPv6 fragments.
1137 Change the interface-specific default settings.
1141 Change all the interface-specific settings.
1143 [XXX: Other special features than forwarding?]
1145 conf/all/forwarding - BOOLEAN
1146 Enable global IPv6 forwarding between all interfaces.
1148 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1149 to control which interfaces may forward packets and which not.
1151 This also sets all interfaces' Host/Router setting
1152 'forwarding' to the specified value. See below for details.
1154 This referred to as global forwarding.
1160 Change special settings per interface.
1162 The functional behaviour for certain settings is different
1163 depending on whether local forwarding is enabled or not.
1166 Accept Router Advertisements; autoconfigure using them.
1168 It also determines whether or not to transmit Router
1169 Solicitations. If and only if the functional setting is to
1170 accept Router Advertisements, Router Solicitations will be
1173 Possible values are:
1174 0 Do not accept Router Advertisements.
1175 1 Accept Router Advertisements if forwarding is disabled.
1176 2 Overrule forwarding behaviour. Accept Router Advertisements
1177 even if forwarding is enabled.
1179 Functional default: enabled if local forwarding is disabled.
1180 disabled if local forwarding is enabled.
1182 accept_ra_defrtr - BOOLEAN
1183 Learn default router in Router Advertisement.
1185 Functional default: enabled if accept_ra is enabled.
1186 disabled if accept_ra is disabled.
1188 accept_ra_pinfo - BOOLEAN
1189 Learn Prefix Information in Router Advertisement.
1191 Functional default: enabled if accept_ra is enabled.
1192 disabled if accept_ra is disabled.
1194 accept_ra_rt_info_max_plen - INTEGER
1195 Maximum prefix length of Route Information in RA.
1197 Route Information w/ prefix larger than or equal to this
1198 variable shall be ignored.
1200 Functional default: 0 if accept_ra_rtr_pref is enabled.
1201 -1 if accept_ra_rtr_pref is disabled.
1203 accept_ra_rtr_pref - BOOLEAN
1204 Accept Router Preference in RA.
1206 Functional default: enabled if accept_ra is enabled.
1207 disabled if accept_ra is disabled.
1209 accept_redirects - BOOLEAN
1212 Functional default: enabled if local forwarding is disabled.
1213 disabled if local forwarding is enabled.
1215 accept_source_route - INTEGER
1216 Accept source routing (routing extension header).
1218 >= 0: Accept only routing header type 2.
1219 < 0: Do not accept routing header.
1224 Autoconfigure addresses using Prefix Information in Router
1227 Functional default: enabled if accept_ra_pinfo is enabled.
1228 disabled if accept_ra_pinfo is disabled.
1230 dad_transmits - INTEGER
1231 The amount of Duplicate Address Detection probes to send.
1234 forwarding - INTEGER
1235 Configure interface-specific Host/Router behaviour.
1237 Note: It is recommended to have the same setting on all
1238 interfaces; mixed router/host scenarios are rather uncommon.
1240 Possible values are:
1241 0 Forwarding disabled
1242 1 Forwarding enabled
1246 By default, Host behaviour is assumed. This means:
1248 1. IsRouter flag is not set in Neighbour Advertisements.
1249 2. If accept_ra is TRUE (default), transmit Router
1251 3. If accept_ra is TRUE (default), accept Router
1252 Advertisements (and do autoconfiguration).
1253 4. If accept_redirects is TRUE (default), accept Redirects.
1257 If local forwarding is enabled, Router behaviour is assumed.
1258 This means exactly the reverse from the above:
1260 1. IsRouter flag is set in Neighbour Advertisements.
1261 2. Router Solicitations are not sent unless accept_ra is 2.
1262 3. Router Advertisements are ignored unless accept_ra is 2.
1263 4. Redirects are ignored.
1265 Default: 0 (disabled) if global forwarding is disabled (default),
1266 otherwise 1 (enabled).
1269 Default Hop Limit to set.
1273 Default Maximum Transfer Unit
1274 Default: 1280 (IPv6 required minimum)
1276 router_probe_interval - INTEGER
1277 Minimum interval (in seconds) between Router Probing described
1282 router_solicitation_delay - INTEGER
1283 Number of seconds to wait after interface is brought up
1284 before sending Router Solicitations.
1287 router_solicitation_interval - INTEGER
1288 Number of seconds to wait between Router Solicitations.
1291 router_solicitations - INTEGER
1292 Number of Router Solicitations to send until assuming no
1293 routers are present.
1296 use_tempaddr - INTEGER
1297 Preference for Privacy Extensions (RFC3041).
1298 <= 0 : disable Privacy Extensions
1299 == 1 : enable Privacy Extensions, but prefer public
1300 addresses over temporary addresses.
1301 > 1 : enable Privacy Extensions and prefer temporary
1302 addresses over public addresses.
1303 Default: 0 (for most devices)
1304 -1 (for point-to-point devices and loopback devices)
1306 temp_valid_lft - INTEGER
1307 valid lifetime (in seconds) for temporary addresses.
1308 Default: 604800 (7 days)
1310 temp_prefered_lft - INTEGER
1311 Preferred lifetime (in seconds) for temporary addresses.
1312 Default: 86400 (1 day)
1314 max_desync_factor - INTEGER
1315 Maximum value for DESYNC_FACTOR, which is a random value
1316 that ensures that clients don't synchronize with each
1317 other and generate new addresses at exactly the same time.
1318 value is in seconds.
1321 regen_max_retry - INTEGER
1322 Number of attempts before give up attempting to generate
1323 valid temporary addresses.
1326 max_addresses - INTEGER
1327 Maximum number of autoconfigured addresses per interface. Setting
1328 to zero disables the limitation. It is not recommended to set this
1329 value too large (or to zero) because it would be an easy way to
1330 crash the kernel by allowing too many addresses to be created.
1333 disable_ipv6 - BOOLEAN
1334 Disable IPv6 operation. If accept_dad is set to 2, this value
1335 will be dynamically set to TRUE if DAD fails for the link-local
1337 Default: FALSE (enable IPv6 operation)
1339 When this value is changed from 1 to 0 (IPv6 is being enabled),
1340 it will dynamically create a link-local address on the given
1341 interface and start Duplicate Address Detection, if necessary.
1343 When this value is changed from 0 to 1 (IPv6 is being disabled),
1344 it will dynamically delete all address on the given interface.
1346 accept_dad - INTEGER
1347 Whether to accept DAD (Duplicate Address Detection).
1349 1: Enable DAD (default)
1350 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1351 link-local address has been found.
1353 force_tllao - BOOLEAN
1354 Enable sending the target link-layer address option even when
1355 responding to a unicast neighbor solicitation.
1358 Quoting from RFC 2461, section 4.4, Target link-layer address:
1360 "The option MUST be included for multicast solicitations in order to
1361 avoid infinite Neighbor Solicitation "recursion" when the peer node
1362 does not have a cache entry to return a Neighbor Advertisements
1363 message. When responding to unicast solicitations, the option can be
1364 omitted since the sender of the solicitation has the correct link-
1365 layer address; otherwise it would not have be able to send the unicast
1366 solicitation in the first place. However, including the link-layer
1367 address in this case adds little overhead and eliminates a potential
1368 race condition where the sender deletes the cached link-layer address
1369 prior to receiving a response to a previous solicitation."
1371 ndisc_notify - BOOLEAN
1372 Define mode for notification of address and device changes.
1373 0 - (default): do nothing
1374 1 - Generate unsolicited neighbour advertisements when device is brought
1375 up or hardware address changes.
1377 mldv1_unsolicited_report_interval - INTEGER
1378 The interval in milliseconds in which the next unsolicited
1379 MLDv1 report retransmit will take place.
1380 Default: 10000 (10 seconds)
1382 mldv2_unsolicited_report_interval - INTEGER
1383 The interval in milliseconds in which the next unsolicited
1384 MLDv2 report retransmit will take place.
1385 Default: 1000 (1 second)
1387 force_mld_version - INTEGER
1388 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
1389 1 - Enforce to use MLD version 1
1390 2 - Enforce to use MLD version 2
1392 suppress_frag_ndisc - INTEGER
1393 Control RFC 6980 (Security Implications of IPv6 Fragmentation
1394 with IPv6 Neighbor Discovery) behavior:
1395 1 - (default) discard fragmented neighbor discovery packets
1396 0 - allow fragmented neighbor discovery packets
1400 Limit the maximal rates for sending ICMPv6 packets.
1401 0 to disable any limiting,
1402 otherwise the minimal space between responses in milliseconds.
1407 Pekka Savola <pekkas@netcore.fi>
1408 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1411 /proc/sys/net/bridge/* Variables:
1413 bridge-nf-call-arptables - BOOLEAN
1414 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1418 bridge-nf-call-iptables - BOOLEAN
1419 1 : pass bridged IPv4 traffic to iptables' chains.
1423 bridge-nf-call-ip6tables - BOOLEAN
1424 1 : pass bridged IPv6 traffic to ip6tables' chains.
1428 bridge-nf-filter-vlan-tagged - BOOLEAN
1429 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1433 bridge-nf-filter-pppoe-tagged - BOOLEAN
1434 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1438 bridge-nf-pass-vlan-input-dev - BOOLEAN
1439 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
1440 interface on the bridge and set the netfilter input device to the vlan.
1441 This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
1442 target work with vlan-on-top-of-bridge interfaces. When no matching
1443 vlan interface is found, or this switch is off, the input device is
1444 set to the bridge interface.
1445 0: disable bridge netfilter vlan interface lookup.
1448 proc/sys/net/sctp/* Variables:
1450 addip_enable - BOOLEAN
1451 Enable or disable extension of Dynamic Address Reconfiguration
1452 (ADD-IP) functionality specified in RFC5061. This extension provides
1453 the ability to dynamically add and remove new addresses for the SCTP
1456 1: Enable extension.
1458 0: Disable extension.
1462 addip_noauth_enable - BOOLEAN
1463 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1464 authentication to protect the operations of adding or removing new
1465 addresses. This requirement is mandated so that unauthorized hosts
1466 would not be able to hijack associations. However, older
1467 implementations may not have implemented this requirement while
1468 allowing the ADD-IP extension. For reasons of interoperability,
1469 we provide this variable to control the enforcement of the
1470 authentication requirement.
1472 1: Allow ADD-IP extension to be used without authentication. This
1473 should only be set in a closed environment for interoperability
1474 with older implementations.
1476 0: Enforce the authentication requirement
1480 auth_enable - BOOLEAN
1481 Enable or disable Authenticated Chunks extension. This extension
1482 provides the ability to send and receive authenticated chunks and is
1483 required for secure operation of Dynamic Address Reconfiguration
1486 1: Enable this extension.
1487 0: Disable this extension.
1491 prsctp_enable - BOOLEAN
1492 Enable or disable the Partial Reliability extension (RFC3758) which
1493 is used to notify peers that a given DATA should no longer be expected.
1501 The limit of the number of new packets that can be initially sent. It
1502 controls how bursty the generated traffic can be.
1506 association_max_retrans - INTEGER
1507 Set the maximum number for retransmissions that an association can
1508 attempt deciding that the remote end is unreachable. If this value
1509 is exceeded, the association is terminated.
1513 max_init_retransmits - INTEGER
1514 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1515 that an association will attempt before declaring the destination
1516 unreachable and terminating.
1520 path_max_retrans - INTEGER
1521 The maximum number of retransmissions that will be attempted on a given
1522 path. Once this threshold is exceeded, the path is considered
1523 unreachable, and new traffic will use a different path when the
1524 association is multihomed.
1528 pf_retrans - INTEGER
1529 The number of retransmissions that will be attempted on a given path
1530 before traffic is redirected to an alternate transport (should one
1531 exist). Note this is distinct from path_max_retrans, as a path that
1532 passes the pf_retrans threshold can still be used. Its only
1533 deprioritized when a transmission path is selected by the stack. This
1534 setting is primarily used to enable fast failover mechanisms without
1535 having to reduce path_max_retrans to a very low value. See:
1536 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
1537 for details. Note also that a value of pf_retrans > path_max_retrans
1538 disables this feature
1542 rto_initial - INTEGER
1543 The initial round trip timeout value in milliseconds that will be used
1544 in calculating round trip times. This is the initial time interval
1545 for retransmissions.
1550 The maximum value (in milliseconds) of the round trip timeout. This
1551 is the largest time interval that can elapse between retransmissions.
1556 The minimum value (in milliseconds) of the round trip timeout. This
1557 is the smallest time interval the can elapse between retransmissions.
1561 hb_interval - INTEGER
1562 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
1563 are sent at the specified interval on idle paths to probe the state of
1564 a given path between 2 associations.
1568 sack_timeout - INTEGER
1569 The amount of time (in milliseconds) that the implementation will wait
1574 valid_cookie_life - INTEGER
1575 The default lifetime of the SCTP cookie (in milliseconds). The cookie
1576 is used during association establishment.
1580 cookie_preserve_enable - BOOLEAN
1581 Enable or disable the ability to extend the lifetime of the SCTP cookie
1582 that is used during the establishment phase of SCTP association
1584 1: Enable cookie lifetime extension.
1589 cookie_hmac_alg - STRING
1590 Select the hmac algorithm used when generating the cookie value sent by
1591 a listening sctp socket to a connecting client in the INIT-ACK chunk.
1596 Ability to assign md5 or sha1 as the selected alg is predicated on the
1597 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
1598 CONFIG_CRYPTO_SHA1).
1600 Default: Dependent on configuration. MD5 if available, else SHA1 if
1601 available, else none.
1603 rcvbuf_policy - INTEGER
1604 Determines if the receive buffer is attributed to the socket or to
1605 association. SCTP supports the capability to create multiple
1606 associations on a single socket. When using this capability, it is
1607 possible that a single stalled association that's buffering a lot
1608 of data may block other associations from delivering their data by
1609 consuming all of the receive buffer space. To work around this,
1610 the rcvbuf_policy could be set to attribute the receiver buffer space
1611 to each association instead of the socket. This prevents the described
1614 1: rcvbuf space is per association
1615 0: rcvbuf space is per socket
1619 sndbuf_policy - INTEGER
1620 Similar to rcvbuf_policy above, this applies to send buffer space.
1622 1: Send buffer is tracked per association
1623 0: Send buffer is tracked per socket.
1627 sctp_mem - vector of 3 INTEGERs: min, pressure, max
1628 Number of pages allowed for queueing by all SCTP sockets.
1630 min: Below this number of pages SCTP is not bothered about its
1631 memory appetite. When amount of memory allocated by SCTP exceeds
1632 this number, SCTP starts to moderate memory usage.
1634 pressure: This value was introduced to follow format of tcp_mem.
1636 max: Number of pages allowed for queueing by all SCTP sockets.
1638 Default is calculated at boot time from amount of available memory.
1640 sctp_rmem - vector of 3 INTEGERs: min, default, max
1641 Only the first value ("min") is used, "default" and "max" are
1644 min: Minimal size of receive buffer used by SCTP socket.
1645 It is guaranteed to each SCTP socket (but not association) even
1646 under moderate memory pressure.
1650 sctp_wmem - vector of 3 INTEGERs: min, default, max
1651 Currently this tunable has no effect.
1653 addr_scope_policy - INTEGER
1654 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
1656 0 - Disable IPv4 address scoping
1657 1 - Enable IPv4 address scoping
1658 2 - Follow draft but allow IPv4 private addresses
1659 3 - Follow draft but allow IPv4 link local addresses
1664 /proc/sys/net/core/*
1665 Please see: Documentation/sysctl/net.txt for descriptions of these entries.
1668 /proc/sys/net/unix/*
1669 max_dgram_qlen - INTEGER
1670 The maximum length of dgram socket receive queue
1677 /proc/sys/net/irda/*
1678 fast_poll_increase FIXME
1679 warn_noreply_time FIXME
1680 discovery_slots FIXME
1683 discovery_timeout FIXME
1684 lap_keepalive_time FIXME
1685 max_noreply_time FIXME
1686 max_tx_data_size FIXME
1688 min_tx_turn_time FIXME