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 - BOOLEAN
19 Disable Path MTU Discovery.
23 default 552 - minimum discovered Path MTU
25 route/max_size - INTEGER
26 Maximum number of routes allowed in the kernel. Increase
27 this when using large numbers of interfaces and/or routes.
29 neigh/default/gc_thresh1 - INTEGER
30 Minimum number of entries to keep. Garbage collector will not
31 purge entries if there are fewer than this number.
34 neigh/default/gc_thresh3 - INTEGER
35 Maximum number of neighbor entries allowed. Increase this
36 when using large numbers of interfaces and when communicating
37 with large numbers of directly-connected peers.
40 neigh/default/unres_qlen_bytes - INTEGER
41 The maximum number of bytes which may be used by packets
42 queued for each unresolved address by other network layers.
44 Setting negative value is meaningless and will return error.
45 Default: 65536 Bytes(64KB)
47 neigh/default/unres_qlen - INTEGER
48 The maximum number of packets which may be queued for each
49 unresolved address by other network layers.
50 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
51 Prior to linux 3.3, the default value is 3 which may cause
52 unexpected packet loss. The current default value is calculated
53 according to default value of unres_qlen_bytes and true size of
58 Time, in seconds, that cached PMTU information is kept.
61 The advertised MSS depends on the first hop route MTU, but will
62 never be lower than this setting.
66 ipfrag_high_thresh - INTEGER
67 Maximum memory used to reassemble IP fragments. When
68 ipfrag_high_thresh bytes of memory is allocated for this purpose,
69 the fragment handler will toss packets until ipfrag_low_thresh
72 ipfrag_low_thresh - INTEGER
73 See ipfrag_high_thresh
76 Time in seconds to keep an IP fragment in memory.
78 ipfrag_secret_interval - INTEGER
79 Regeneration interval (in seconds) of the hash secret (or lifetime
80 for the hash secret) for IP fragments.
83 ipfrag_max_dist - INTEGER
84 ipfrag_max_dist is a non-negative integer value which defines the
85 maximum "disorder" which is allowed among fragments which share a
86 common IP source address. Note that reordering of packets is
87 not unusual, but if a large number of fragments arrive from a source
88 IP address while a particular fragment queue remains incomplete, it
89 probably indicates that one or more fragments belonging to that queue
90 have been lost. When ipfrag_max_dist is positive, an additional check
91 is done on fragments before they are added to a reassembly queue - if
92 ipfrag_max_dist (or more) fragments have arrived from a particular IP
93 address between additions to any IP fragment queue using that source
94 address, it's presumed that one or more fragments in the queue are
95 lost. The existing fragment queue will be dropped, and a new one
96 started. An ipfrag_max_dist value of zero disables this check.
98 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
99 result in unnecessarily dropping fragment queues when normal
100 reordering of packets occurs, which could lead to poor application
101 performance. Using a very large value, e.g. 50000, increases the
102 likelihood of incorrectly reassembling IP fragments that originate
103 from different IP datagrams, which could result in data corruption.
108 inet_peer_threshold - INTEGER
109 The approximate size of the storage. Starting from this threshold
110 entries will be thrown aggressively. This threshold also determines
111 entries' time-to-live and time intervals between garbage collection
112 passes. More entries, less time-to-live, less GC interval.
114 inet_peer_minttl - INTEGER
115 Minimum time-to-live of entries. Should be enough to cover fragment
116 time-to-live on the reassembling side. This minimum time-to-live is
117 guaranteed if the pool size is less than inet_peer_threshold.
120 inet_peer_maxttl - INTEGER
121 Maximum time-to-live of entries. Unused entries will expire after
122 this period of time if there is no memory pressure on the pool (i.e.
123 when the number of entries in the pool is very small).
129 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
130 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
134 Controls Appropriate Byte Count (ABC) defined in RFC3465.
135 ABC is a way of increasing congestion window (cwnd) more slowly
136 in response to partial acknowledgments.
138 0 increase cwnd once per acknowledgment (no ABC)
139 1 increase cwnd once per acknowledgment of full sized segment
140 2 allow increase cwnd by two if acknowledgment is
141 of two segments to compensate for delayed acknowledgments.
144 tcp_abort_on_overflow - BOOLEAN
145 If listening service is too slow to accept new connections,
146 reset them. Default state is FALSE. It means that if overflow
147 occurred due to a burst, connection will recover. Enable this
148 option _only_ if you are really sure that listening daemon
149 cannot be tuned to accept connections faster. Enabling this
150 option can harm clients of your server.
152 tcp_adv_win_scale - INTEGER
153 Count buffering overhead as bytes/2^tcp_adv_win_scale
154 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
156 Possible values are [-31, 31], inclusive.
159 tcp_allowed_congestion_control - STRING
160 Show/set the congestion control choices available to non-privileged
161 processes. The list is a subset of those listed in
162 tcp_available_congestion_control.
163 Default is "reno" and the default setting (tcp_congestion_control).
165 tcp_app_win - INTEGER
166 Reserve max(window/2^tcp_app_win, mss) of window for application
167 buffer. Value 0 is special, it means that nothing is reserved.
170 tcp_available_congestion_control - STRING
171 Shows the available congestion control choices that are registered.
172 More congestion control algorithms may be available as modules,
175 tcp_base_mss - INTEGER
176 The initial value of search_low to be used by the packetization layer
177 Path MTU discovery (MTU probing). If MTU probing is enabled,
178 this is the initial MSS used by the connection.
180 tcp_congestion_control - STRING
181 Set the congestion control algorithm to be used for new
182 connections. The algorithm "reno" is always available, but
183 additional choices may be available based on kernel configuration.
184 Default is set as part of kernel configuration.
185 For passive connections, the listener congestion control choice
187 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
189 tcp_cookie_size - INTEGER
190 Default size of TCP Cookie Transactions (TCPCT) option, that may be
191 overridden on a per socket basis by the TCPCT socket option.
192 Values greater than the maximum (16) are interpreted as the maximum.
193 Values greater than zero and less than the minimum (8) are interpreted
194 as the minimum. Odd values are interpreted as the next even value.
198 Allows TCP to send "duplicate" SACKs.
200 tcp_early_retrans - INTEGER
201 Enable Early Retransmit (ER), per RFC 5827. ER lowers the threshold
202 for triggering fast retransmit when the amount of outstanding data is
203 small and when no previously unsent data can be transmitted (such
204 that limited transmit could be used).
208 2 enables ER but delays fast recovery and fast retransmit
209 by a fourth of RTT. This mitigates connection falsely
210 recovers when network has a small degree of reordering
211 (less than 3 packets).
215 Control use of Explicit Congestion Notification (ECN) by TCP.
216 ECN is used only when both ends of the TCP connection indicate
217 support for it. This feature is useful in avoiding losses due
218 to congestion by allowing supporting routers to signal
219 congestion before having to drop packets.
221 0 Disable ECN. Neither initiate nor accept ECN.
222 1 Enable ECN when requested by incoming connections and
223 also request ECN on outgoing connection attempts.
224 2 Enable ECN when requested by incoming connections
225 but do not request ECN on outgoing connections.
229 Enable FACK congestion avoidance and fast retransmission.
230 The value is not used, if tcp_sack is not enabled.
232 tcp_fin_timeout - INTEGER
233 The length of time an orphaned (no longer referenced by any
234 application) connection will remain in the FIN_WAIT_2 state
235 before it is aborted at the local end. While a perfectly
236 valid "receive only" state for an un-orphaned connection, an
237 orphaned connection in FIN_WAIT_2 state could otherwise wait
238 forever for the remote to close its end of the connection.
243 Enables Forward RTO-Recovery (F-RTO) defined in RFC4138.
244 F-RTO is an enhanced recovery algorithm for TCP retransmission
245 timeouts. It is particularly beneficial in wireless environments
246 where packet loss is typically due to random radio interference
247 rather than intermediate router congestion. F-RTO is sender-side
248 only modification. Therefore it does not require any support from
251 If set to 1, basic version is enabled. 2 enables SACK enhanced
252 F-RTO if flow uses SACK. The basic version can be used also when
253 SACK is in use though scenario(s) with it exists where F-RTO
254 interacts badly with the packet counting of the SACK enabled TCP
257 tcp_frto_response - INTEGER
258 When F-RTO has detected that a TCP retransmission timeout was
259 spurious (i.e, the timeout would have been avoided had TCP set a
260 longer retransmission timeout), TCP has several options what to do
261 next. Possible values are:
262 0 Rate halving based; a smooth and conservative response,
263 results in halved cwnd and ssthresh after one RTT
264 1 Very conservative response; not recommended because even
265 though being valid, it interacts poorly with the rest of
266 Linux TCP, halves cwnd and ssthresh immediately
267 2 Aggressive response; undoes congestion control measures
268 that are now known to be unnecessary (ignoring the
269 possibility of a lost retransmission that would require
270 TCP to be more cautious), cwnd and ssthresh are restored
271 to the values prior timeout
272 Default: 0 (rate halving based)
274 tcp_keepalive_time - INTEGER
275 How often TCP sends out keepalive messages when keepalive is enabled.
278 tcp_keepalive_probes - INTEGER
279 How many keepalive probes TCP sends out, until it decides that the
280 connection is broken. Default value: 9.
282 tcp_keepalive_intvl - INTEGER
283 How frequently the probes are send out. Multiplied by
284 tcp_keepalive_probes it is time to kill not responding connection,
285 after probes started. Default value: 75sec i.e. connection
286 will be aborted after ~11 minutes of retries.
288 tcp_low_latency - BOOLEAN
289 If set, the TCP stack makes decisions that prefer lower
290 latency as opposed to higher throughput. By default, this
291 option is not set meaning that higher throughput is preferred.
292 An example of an application where this default should be
293 changed would be a Beowulf compute cluster.
296 tcp_max_orphans - INTEGER
297 Maximal number of TCP sockets not attached to any user file handle,
298 held by system. If this number is exceeded orphaned connections are
299 reset immediately and warning is printed. This limit exists
300 only to prevent simple DoS attacks, you _must_ not rely on this
301 or lower the limit artificially, but rather increase it
302 (probably, after increasing installed memory),
303 if network conditions require more than default value,
304 and tune network services to linger and kill such states
305 more aggressively. Let me to remind again: each orphan eats
306 up to ~64K of unswappable memory.
308 tcp_max_ssthresh - INTEGER
309 Limited Slow-Start for TCP with large congestion windows (cwnd) defined in
310 RFC3742. Limited slow-start is a mechanism to limit growth of the cwnd
311 on the region where cwnd is larger than tcp_max_ssthresh. TCP increases cwnd
312 by at most tcp_max_ssthresh segments, and by at least tcp_max_ssthresh/2
313 segments per RTT when the cwnd is above tcp_max_ssthresh.
314 If TCP connection increased cwnd to thousands (or tens of thousands) segments,
315 and thousands of packets were being dropped during slow-start, you can set
316 tcp_max_ssthresh to improve performance for new TCP connection.
319 tcp_max_syn_backlog - INTEGER
320 Maximal number of remembered connection requests, which have not
321 received an acknowledgment from connecting client.
322 The minimal value is 128 for low memory machines, and it will
323 increase in proportion to the memory of machine.
324 If server suffers from overload, try increasing this number.
326 tcp_max_tw_buckets - INTEGER
327 Maximal number of timewait sockets held by system simultaneously.
328 If this number is exceeded time-wait socket is immediately destroyed
329 and warning is printed. This limit exists only to prevent
330 simple DoS attacks, you _must_ not lower the limit artificially,
331 but rather increase it (probably, after increasing installed memory),
332 if network conditions require more than default value.
334 tcp_mem - vector of 3 INTEGERs: min, pressure, max
335 min: below this number of pages TCP is not bothered about its
338 pressure: when amount of memory allocated by TCP exceeds this number
339 of pages, TCP moderates its memory consumption and enters memory
340 pressure mode, which is exited when memory consumption falls
343 max: number of pages allowed for queueing by all TCP sockets.
345 Defaults are calculated at boot time from amount of available
348 tcp_moderate_rcvbuf - BOOLEAN
349 If set, TCP performs receive buffer auto-tuning, attempting to
350 automatically size the buffer (no greater than tcp_rmem[2]) to
351 match the size required by the path for full throughput. Enabled by
354 tcp_mtu_probing - INTEGER
355 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
358 1 - Disabled by default, enabled when an ICMP black hole detected
359 2 - Always enabled, use initial MSS of tcp_base_mss.
361 tcp_no_metrics_save - BOOLEAN
362 By default, TCP saves various connection metrics in the route cache
363 when the connection closes, so that connections established in the
364 near future can use these to set initial conditions. Usually, this
365 increases overall performance, but may sometimes cause performance
366 degradation. If set, TCP will not cache metrics on closing
369 tcp_orphan_retries - INTEGER
370 This value influences the timeout of a locally closed TCP connection,
371 when RTO retransmissions remain unacknowledged.
372 See tcp_retries2 for more details.
374 The default value is 8.
375 If your machine is a loaded WEB server,
376 you should think about lowering this value, such sockets
377 may consume significant resources. Cf. tcp_max_orphans.
379 tcp_reordering - INTEGER
380 Maximal reordering of packets in a TCP stream.
383 tcp_retrans_collapse - BOOLEAN
384 Bug-to-bug compatibility with some broken printers.
385 On retransmit try to send bigger packets to work around bugs in
388 tcp_retries1 - INTEGER
389 This value influences the time, after which TCP decides, that
390 something is wrong due to unacknowledged RTO retransmissions,
391 and reports this suspicion to the network layer.
392 See tcp_retries2 for more details.
394 RFC 1122 recommends at least 3 retransmissions, which is the
397 tcp_retries2 - INTEGER
398 This value influences the timeout of an alive TCP connection,
399 when RTO retransmissions remain unacknowledged.
400 Given a value of N, a hypothetical TCP connection following
401 exponential backoff with an initial RTO of TCP_RTO_MIN would
402 retransmit N times before killing the connection at the (N+1)th RTO.
404 The default value of 15 yields a hypothetical timeout of 924.6
405 seconds and is a lower bound for the effective timeout.
406 TCP will effectively time out at the first RTO which exceeds the
407 hypothetical timeout.
409 RFC 1122 recommends at least 100 seconds for the timeout,
410 which corresponds to a value of at least 8.
412 tcp_rfc1337 - BOOLEAN
413 If set, the TCP stack behaves conforming to RFC1337. If unset,
414 we are not conforming to RFC, but prevent TCP TIME_WAIT
418 tcp_rmem - vector of 3 INTEGERs: min, default, max
419 min: Minimal size of receive buffer used by TCP sockets.
420 It is guaranteed to each TCP socket, even under moderate memory
424 default: initial size of receive buffer used by TCP sockets.
425 This value overrides net.core.rmem_default used by other protocols.
426 Default: 87380 bytes. This value results in window of 65535 with
427 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
428 less for default tcp_app_win. See below about these variables.
430 max: maximal size of receive buffer allowed for automatically
431 selected receiver buffers for TCP socket. This value does not override
432 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
433 automatic tuning of that socket's receive buffer size, in which
434 case this value is ignored.
435 Default: between 87380B and 6MB, depending on RAM size.
438 Enable select acknowledgments (SACKS).
440 tcp_slow_start_after_idle - BOOLEAN
441 If set, provide RFC2861 behavior and time out the congestion
442 window after an idle period. An idle period is defined at
443 the current RTO. If unset, the congestion window will not
444 be timed out after an idle period.
448 Use the Host requirements interpretation of the TCP urgent pointer field.
449 Most hosts use the older BSD interpretation, so if you turn this on
450 Linux might not communicate correctly with them.
453 tcp_synack_retries - INTEGER
454 Number of times SYNACKs for a passive TCP connection attempt will
455 be retransmitted. Should not be higher than 255. Default value
456 is 5, which corresponds to 31seconds till the last retransmission
457 with the current initial RTO of 1second. With this the final timeout
458 for a passive TCP connection will happen after 63seconds.
460 tcp_syncookies - BOOLEAN
461 Only valid when the kernel was compiled with CONFIG_SYNCOOKIES
462 Send out syncookies when the syn backlog queue of a socket
463 overflows. This is to prevent against the common 'SYN flood attack'
466 Note, that syncookies is fallback facility.
467 It MUST NOT be used to help highly loaded servers to stand
468 against legal connection rate. If you see SYN flood warnings
469 in your logs, but investigation shows that they occur
470 because of overload with legal connections, you should tune
471 another parameters until this warning disappear.
472 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
474 syncookies seriously violate TCP protocol, do not allow
475 to use TCP extensions, can result in serious degradation
476 of some services (f.e. SMTP relaying), visible not by you,
477 but your clients and relays, contacting you. While you see
478 SYN flood warnings in logs not being really flooded, your server
479 is seriously misconfigured.
481 tcp_fastopen - INTEGER
482 Enable TCP Fast Open feature (draft-ietf-tcpm-fastopen) to send data
483 in the opening SYN packet. To use this feature, the client application
484 must use sendmsg() or sendto() with MSG_FASTOPEN flag rather than
485 connect() to perform a TCP handshake automatically.
487 The values (bitmap) are
488 1: Enables sending data in the opening SYN on the client.
489 2: Enables TCP Fast Open on the server side, i.e., allowing data in
490 a SYN packet to be accepted and passed to the application before
491 3-way hand shake finishes.
492 4: Send data in the opening SYN regardless of cookie availability and
493 without a cookie option.
494 0x100: Accept SYN data w/o validating the cookie.
495 0x200: Accept data-in-SYN w/o any cookie option present.
496 0x400/0x800: Enable Fast Open on all listeners regardless of the
497 TCP_FASTOPEN socket option. The two different flags designate two
498 different ways of setting max_qlen without the TCP_FASTOPEN socket
503 Note that the client & server side Fast Open flags (1 and 2
504 respectively) must be also enabled before the rest of flags can take
507 See include/net/tcp.h and the code for more details.
509 tcp_syn_retries - INTEGER
510 Number of times initial SYNs for an active TCP connection attempt
511 will be retransmitted. Should not be higher than 255. Default value
512 is 6, which corresponds to 63seconds till the last retransmission
513 with the current initial RTO of 1second. With this the final timeout
514 for an active TCP connection attempt will happen after 127seconds.
516 tcp_timestamps - BOOLEAN
517 Enable timestamps as defined in RFC1323.
519 tcp_tso_win_divisor - INTEGER
520 This allows control over what percentage of the congestion window
521 can be consumed by a single TSO frame.
522 The setting of this parameter is a choice between burstiness and
523 building larger TSO frames.
526 tcp_tw_recycle - BOOLEAN
527 Enable fast recycling TIME-WAIT sockets. Default value is 0.
528 It should not be changed without advice/request of technical
531 tcp_tw_reuse - BOOLEAN
532 Allow to reuse TIME-WAIT sockets for new connections when it is
533 safe from protocol viewpoint. Default value is 0.
534 It should not be changed without advice/request of technical
537 tcp_window_scaling - BOOLEAN
538 Enable window scaling as defined in RFC1323.
540 tcp_wmem - vector of 3 INTEGERs: min, default, max
541 min: Amount of memory reserved for send buffers for TCP sockets.
542 Each TCP socket has rights to use it due to fact of its birth.
545 default: initial size of send buffer used by TCP sockets. This
546 value overrides net.core.wmem_default used by other protocols.
547 It is usually lower than net.core.wmem_default.
550 max: Maximal amount of memory allowed for automatically tuned
551 send buffers for TCP sockets. This value does not override
552 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
553 automatic tuning of that socket's send buffer size, in which case
554 this value is ignored.
555 Default: between 64K and 4MB, depending on RAM size.
557 tcp_workaround_signed_windows - BOOLEAN
558 If set, assume no receipt of a window scaling option means the
559 remote TCP is broken and treats the window as a signed quantity.
560 If unset, assume the remote TCP is not broken even if we do
561 not receive a window scaling option from them.
564 tcp_dma_copybreak - INTEGER
565 Lower limit, in bytes, of the size of socket reads that will be
566 offloaded to a DMA copy engine, if one is present in the system
567 and CONFIG_NET_DMA is enabled.
570 tcp_thin_linear_timeouts - BOOLEAN
571 Enable dynamic triggering of linear timeouts for thin streams.
572 If set, a check is performed upon retransmission by timeout to
573 determine if the stream is thin (less than 4 packets in flight).
574 As long as the stream is found to be thin, up to 6 linear
575 timeouts may be performed before exponential backoff mode is
576 initiated. This improves retransmission latency for
577 non-aggressive thin streams, often found to be time-dependent.
578 For more information on thin streams, see
579 Documentation/networking/tcp-thin.txt
582 tcp_thin_dupack - BOOLEAN
583 Enable dynamic triggering of retransmissions after one dupACK
584 for thin streams. If set, a check is performed upon reception
585 of a dupACK to determine if the stream is thin (less than 4
586 packets in flight). As long as the stream is found to be thin,
587 data is retransmitted on the first received dupACK. This
588 improves retransmission latency for non-aggressive thin
589 streams, often found to be time-dependent.
590 For more information on thin streams, see
591 Documentation/networking/tcp-thin.txt
594 tcp_limit_output_bytes - INTEGER
595 Controls TCP Small Queue limit per tcp socket.
596 TCP bulk sender tends to increase packets in flight until it
597 gets losses notifications. With SNDBUF autotuning, this can
598 result in a large amount of packets queued in qdisc/device
599 on the local machine, hurting latency of other flows, for
600 typical pfifo_fast qdiscs.
601 tcp_limit_output_bytes limits the number of bytes on qdisc
602 or device to reduce artificial RTT/cwnd and reduce bufferbloat.
603 Note: For GSO/TSO enabled flows, we try to have at least two
604 packets in flight. Reducing tcp_limit_output_bytes might also
605 reduce the size of individual GSO packet (64KB being the max)
608 tcp_challenge_ack_limit - INTEGER
609 Limits number of Challenge ACK sent per second, as recommended
610 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
615 udp_mem - vector of 3 INTEGERs: min, pressure, max
616 Number of pages allowed for queueing by all UDP sockets.
618 min: Below this number of pages UDP is not bothered about its
619 memory appetite. When amount of memory allocated by UDP exceeds
620 this number, UDP starts to moderate memory usage.
622 pressure: This value was introduced to follow format of tcp_mem.
624 max: Number of pages allowed for queueing by all UDP sockets.
626 Default is calculated at boot time from amount of available memory.
628 udp_rmem_min - INTEGER
629 Minimal size of receive buffer used by UDP sockets in moderation.
630 Each UDP socket is able to use the size for receiving data, even if
631 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
634 udp_wmem_min - INTEGER
635 Minimal size of send buffer used by UDP sockets in moderation.
636 Each UDP socket is able to use the size for sending data, even if
637 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
642 cipso_cache_enable - BOOLEAN
643 If set, enable additions to and lookups from the CIPSO label mapping
644 cache. If unset, additions are ignored and lookups always result in a
645 miss. However, regardless of the setting the cache is still
646 invalidated when required when means you can safely toggle this on and
647 off and the cache will always be "safe".
650 cipso_cache_bucket_size - INTEGER
651 The CIPSO label cache consists of a fixed size hash table with each
652 hash bucket containing a number of cache entries. This variable limits
653 the number of entries in each hash bucket; the larger the value the
654 more CIPSO label mappings that can be cached. When the number of
655 entries in a given hash bucket reaches this limit adding new entries
656 causes the oldest entry in the bucket to be removed to make room.
659 cipso_rbm_optfmt - BOOLEAN
660 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
661 the CIPSO draft specification (see Documentation/netlabel for details).
662 This means that when set the CIPSO tag will be padded with empty
663 categories in order to make the packet data 32-bit aligned.
666 cipso_rbm_structvalid - BOOLEAN
667 If set, do a very strict check of the CIPSO option when
668 ip_options_compile() is called. If unset, relax the checks done during
669 ip_options_compile(). Either way is "safe" as errors are caught else
670 where in the CIPSO processing code but setting this to 0 (False) should
671 result in less work (i.e. it should be faster) but could cause problems
672 with other implementations that require strict checking.
677 ip_local_port_range - 2 INTEGERS
678 Defines the local port range that is used by TCP and UDP to
679 choose the local port. The first number is the first, the
680 second the last local port number. The default values are
681 32768 and 61000 respectively.
683 ip_local_reserved_ports - list of comma separated ranges
684 Specify the ports which are reserved for known third-party
685 applications. These ports will not be used by automatic port
686 assignments (e.g. when calling connect() or bind() with port
687 number 0). Explicit port allocation behavior is unchanged.
689 The format used for both input and output is a comma separated
690 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
691 10). Writing to the file will clear all previously reserved
692 ports and update the current list with the one given in the
695 Note that ip_local_port_range and ip_local_reserved_ports
696 settings are independent and both are considered by the kernel
697 when determining which ports are available for automatic port
700 You can reserve ports which are not in the current
701 ip_local_port_range, e.g.:
703 $ cat /proc/sys/net/ipv4/ip_local_port_range
705 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
708 although this is redundant. However such a setting is useful
709 if later the port range is changed to a value that will
710 include the reserved ports.
714 ip_nonlocal_bind - BOOLEAN
715 If set, allows processes to bind() to non-local IP addresses,
716 which can be quite useful - but may break some applications.
720 If set non-zero, enables support for dynamic addresses.
721 If set to a non-zero value larger than 1, a kernel log
722 message will be printed when dynamic address rewriting
726 icmp_echo_ignore_all - BOOLEAN
727 If set non-zero, then the kernel will ignore all ICMP ECHO
731 icmp_echo_ignore_broadcasts - BOOLEAN
732 If set non-zero, then the kernel will ignore all ICMP ECHO and
733 TIMESTAMP requests sent to it via broadcast/multicast.
736 icmp_ratelimit - INTEGER
737 Limit the maximal rates for sending ICMP packets whose type matches
738 icmp_ratemask (see below) to specific targets.
739 0 to disable any limiting,
740 otherwise the minimal space between responses in milliseconds.
743 icmp_ratemask - INTEGER
744 Mask made of ICMP types for which rates are being limited.
745 Significant bits: IHGFEDCBA9876543210
746 Default mask: 0000001100000011000 (6168)
748 Bit definitions (see include/linux/icmp.h):
750 3 Destination Unreachable *
755 C Parameter Problem *
760 H Address Mask Request
763 * These are rate limited by default (see default mask above)
765 icmp_ignore_bogus_error_responses - BOOLEAN
766 Some routers violate RFC1122 by sending bogus responses to broadcast
767 frames. Such violations are normally logged via a kernel warning.
768 If this is set to TRUE, the kernel will not give such warnings, which
769 will avoid log file clutter.
772 icmp_errors_use_inbound_ifaddr - BOOLEAN
774 If zero, icmp error messages are sent with the primary address of
775 the exiting interface.
777 If non-zero, the message will be sent with the primary address of
778 the interface that received the packet that caused the icmp error.
779 This is the behaviour network many administrators will expect from
780 a router. And it can make debugging complicated network layouts
783 Note that if no primary address exists for the interface selected,
784 then the primary address of the first non-loopback interface that
785 has one will be used regardless of this setting.
789 igmp_max_memberships - INTEGER
790 Change the maximum number of multicast groups we can subscribe to.
793 Theoretical maximum value is bounded by having to send a membership
794 report in a single datagram (i.e. the report can't span multiple
795 datagrams, or risk confusing the switch and leaving groups you don't
798 The number of supported groups 'M' is bounded by the number of group
799 report entries you can fit into a single datagram of 65535 bytes.
801 M = 65536-sizeof (ip header)/(sizeof(Group record))
803 Group records are variable length, with a minimum of 12 bytes.
804 So net.ipv4.igmp_max_memberships should not be set higher than:
806 (65536-24) / 12 = 5459
808 The value 5459 assumes no IP header options, so in practice
809 this number may be lower.
811 conf/interface/* changes special settings per interface (where
812 "interface" is the name of your network interface)
814 conf/all/* is special, changes the settings for all interfaces
816 log_martians - BOOLEAN
817 Log packets with impossible addresses to kernel log.
818 log_martians for the interface will be enabled if at least one of
819 conf/{all,interface}/log_martians is set to TRUE,
820 it will be disabled otherwise
822 accept_redirects - BOOLEAN
823 Accept ICMP redirect messages.
824 accept_redirects for the interface will be enabled if:
825 - both conf/{all,interface}/accept_redirects are TRUE in the case
826 forwarding for the interface is enabled
828 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
829 case forwarding for the interface is disabled
830 accept_redirects for the interface will be disabled otherwise
835 Enable IP forwarding on this interface.
837 mc_forwarding - BOOLEAN
838 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
839 and a multicast routing daemon is required.
840 conf/all/mc_forwarding must also be set to TRUE to enable multicast
841 routing for the interface
844 Integer value used to differentiate the devices by the medium they
845 are attached to. Two devices can have different id values when
846 the broadcast packets are received only on one of them.
847 The default value 0 means that the device is the only interface
848 to its medium, value of -1 means that medium is not known.
850 Currently, it is used to change the proxy_arp behavior:
851 the proxy_arp feature is enabled for packets forwarded between
852 two devices attached to different media.
856 proxy_arp for the interface will be enabled if at least one of
857 conf/{all,interface}/proxy_arp is set to TRUE,
858 it will be disabled otherwise
860 proxy_arp_pvlan - BOOLEAN
861 Private VLAN proxy arp.
862 Basically allow proxy arp replies back to the same interface
863 (from which the ARP request/solicitation was received).
865 This is done to support (ethernet) switch features, like RFC
866 3069, where the individual ports are NOT allowed to
867 communicate with each other, but they are allowed to talk to
868 the upstream router. As described in RFC 3069, it is possible
869 to allow these hosts to communicate through the upstream
870 router by proxy_arp'ing. Don't need to be used together with
873 This technology is known by different names:
874 In RFC 3069 it is called VLAN Aggregation.
875 Cisco and Allied Telesyn call it Private VLAN.
876 Hewlett-Packard call it Source-Port filtering or port-isolation.
877 Ericsson call it MAC-Forced Forwarding (RFC Draft).
879 shared_media - BOOLEAN
880 Send(router) or accept(host) RFC1620 shared media redirects.
881 Overrides ip_secure_redirects.
882 shared_media for the interface will be enabled if at least one of
883 conf/{all,interface}/shared_media is set to TRUE,
884 it will be disabled otherwise
887 secure_redirects - BOOLEAN
888 Accept ICMP redirect messages only for gateways,
889 listed in default gateway list.
890 secure_redirects for the interface will be enabled if at least one of
891 conf/{all,interface}/secure_redirects is set to TRUE,
892 it will be disabled otherwise
895 send_redirects - BOOLEAN
896 Send redirects, if router.
897 send_redirects for the interface will be enabled if at least one of
898 conf/{all,interface}/send_redirects is set to TRUE,
899 it will be disabled otherwise
902 bootp_relay - BOOLEAN
903 Accept packets with source address 0.b.c.d destined
904 not to this host as local ones. It is supposed, that
905 BOOTP relay daemon will catch and forward such packets.
906 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
911 accept_source_route - BOOLEAN
912 Accept packets with SRR option.
913 conf/all/accept_source_route must also be set to TRUE to accept packets
914 with SRR option on the interface
915 default TRUE (router)
918 accept_local - BOOLEAN
919 Accept packets with local source addresses. In combination
920 with suitable routing, this can be used to direct packets
921 between two local interfaces over the wire and have them
924 rp_filter must be set to a non-zero value in order for
925 accept_local to have an effect.
929 route_localnet - BOOLEAN
930 Do not consider loopback addresses as martian source or destination
931 while routing. This enables the use of 127/8 for local routing purposes.
935 0 - No source validation.
936 1 - Strict mode as defined in RFC3704 Strict Reverse Path
937 Each incoming packet is tested against the FIB and if the interface
938 is not the best reverse path the packet check will fail.
939 By default failed packets are discarded.
940 2 - Loose mode as defined in RFC3704 Loose Reverse Path
941 Each incoming packet's source address is also tested against the FIB
942 and if the source address is not reachable via any interface
943 the packet check will fail.
945 Current recommended practice in RFC3704 is to enable strict mode
946 to prevent IP spoofing from DDos attacks. If using asymmetric routing
947 or other complicated routing, then loose mode is recommended.
949 The max value from conf/{all,interface}/rp_filter is used
950 when doing source validation on the {interface}.
952 Default value is 0. Note that some distributions enable it
956 1 - Allows you to have multiple network interfaces on the same
957 subnet, and have the ARPs for each interface be answered
958 based on whether or not the kernel would route a packet from
959 the ARP'd IP out that interface (therefore you must use source
960 based routing for this to work). In other words it allows control
961 of which cards (usually 1) will respond to an arp request.
963 0 - (default) The kernel can respond to arp requests with addresses
964 from other interfaces. This may seem wrong but it usually makes
965 sense, because it increases the chance of successful communication.
966 IP addresses are owned by the complete host on Linux, not by
967 particular interfaces. Only for more complex setups like load-
968 balancing, does this behaviour cause problems.
970 arp_filter for the interface will be enabled if at least one of
971 conf/{all,interface}/arp_filter is set to TRUE,
972 it will be disabled otherwise
974 arp_announce - INTEGER
975 Define different restriction levels for announcing the local
976 source IP address from IP packets in ARP requests sent on
978 0 - (default) Use any local address, configured on any interface
979 1 - Try to avoid local addresses that are not in the target's
980 subnet for this interface. This mode is useful when target
981 hosts reachable via this interface require the source IP
982 address in ARP requests to be part of their logical network
983 configured on the receiving interface. When we generate the
984 request we will check all our subnets that include the
985 target IP and will preserve the source address if it is from
986 such subnet. If there is no such subnet we select source
987 address according to the rules for level 2.
988 2 - Always use the best local address for this target.
989 In this mode we ignore the source address in the IP packet
990 and try to select local address that we prefer for talks with
991 the target host. Such local address is selected by looking
992 for primary IP addresses on all our subnets on the outgoing
993 interface that include the target IP address. If no suitable
994 local address is found we select the first local address
995 we have on the outgoing interface or on all other interfaces,
996 with the hope we will receive reply for our request and
997 even sometimes no matter the source IP address we announce.
999 The max value from conf/{all,interface}/arp_announce is used.
1001 Increasing the restriction level gives more chance for
1002 receiving answer from the resolved target while decreasing
1003 the level announces more valid sender's information.
1005 arp_ignore - INTEGER
1006 Define different modes for sending replies in response to
1007 received ARP requests that resolve local target IP addresses:
1008 0 - (default): reply for any local target IP address, configured
1010 1 - reply only if the target IP address is local address
1011 configured on the incoming interface
1012 2 - reply only if the target IP address is local address
1013 configured on the incoming interface and both with the
1014 sender's IP address are part from same subnet on this interface
1015 3 - do not reply for local addresses configured with scope host,
1016 only resolutions for global and link addresses are replied
1018 8 - do not reply for all local addresses
1020 The max value from conf/{all,interface}/arp_ignore is used
1021 when ARP request is received on the {interface}
1023 arp_notify - BOOLEAN
1024 Define mode for notification of address and device changes.
1025 0 - (default): do nothing
1026 1 - Generate gratuitous arp requests when device is brought up
1027 or hardware address changes.
1029 arp_accept - BOOLEAN
1030 Define behavior for gratuitous ARP frames who's IP is not
1031 already present in the ARP table:
1032 0 - don't create new entries in the ARP table
1033 1 - create new entries in the ARP table
1035 Both replies and requests type gratuitous arp will trigger the
1036 ARP table to be updated, if this setting is on.
1038 If the ARP table already contains the IP address of the
1039 gratuitous arp frame, the arp table will be updated regardless
1040 if this setting is on or off.
1043 app_solicit - INTEGER
1044 The maximum number of probes to send to the user space ARP daemon
1045 via netlink before dropping back to multicast probes (see
1046 mcast_solicit). Defaults to 0.
1048 disable_policy - BOOLEAN
1049 Disable IPSEC policy (SPD) for this interface
1051 disable_xfrm - BOOLEAN
1052 Disable IPSEC encryption on this interface, whatever the policy
1057 Allows you to write a number, which can be used as required.
1061 kuznet@ms2.inr.ac.ru
1067 delon.nicolas@wanadoo.fr
1072 /proc/sys/net/ipv6/* Variables:
1074 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1075 apply to IPv6 [XXX?].
1077 bindv6only - BOOLEAN
1078 Default value for IPV6_V6ONLY socket option,
1079 which restricts use of the IPv6 socket to IPv6 communication
1081 TRUE: disable IPv4-mapped address feature
1082 FALSE: enable IPv4-mapped address feature
1084 Default: FALSE (as specified in RFC3493)
1088 ip6frag_high_thresh - INTEGER
1089 Maximum memory used to reassemble IPv6 fragments. When
1090 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1091 the fragment handler will toss packets until ip6frag_low_thresh
1094 ip6frag_low_thresh - INTEGER
1095 See ip6frag_high_thresh
1097 ip6frag_time - INTEGER
1098 Time in seconds to keep an IPv6 fragment in memory.
1100 ip6frag_secret_interval - INTEGER
1101 Regeneration interval (in seconds) of the hash secret (or lifetime
1102 for the hash secret) for IPv6 fragments.
1106 Change the interface-specific default settings.
1110 Change all the interface-specific settings.
1112 [XXX: Other special features than forwarding?]
1114 conf/all/forwarding - BOOLEAN
1115 Enable global IPv6 forwarding between all interfaces.
1117 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1118 to control which interfaces may forward packets and which not.
1120 This also sets all interfaces' Host/Router setting
1121 'forwarding' to the specified value. See below for details.
1123 This referred to as global forwarding.
1129 Change special settings per interface.
1131 The functional behaviour for certain settings is different
1132 depending on whether local forwarding is enabled or not.
1135 Accept Router Advertisements; autoconfigure using them.
1137 It also determines whether or not to transmit Router
1138 Solicitations. If and only if the functional setting is to
1139 accept Router Advertisements, Router Solicitations will be
1142 Possible values are:
1143 0 Do not accept Router Advertisements.
1144 1 Accept Router Advertisements if forwarding is disabled.
1145 2 Overrule forwarding behaviour. Accept Router Advertisements
1146 even if forwarding is enabled.
1148 Functional default: enabled if local forwarding is disabled.
1149 disabled if local forwarding is enabled.
1151 accept_ra_defrtr - BOOLEAN
1152 Learn default router in Router Advertisement.
1154 Functional default: enabled if accept_ra is enabled.
1155 disabled if accept_ra is disabled.
1157 accept_ra_pinfo - BOOLEAN
1158 Learn Prefix Information in Router Advertisement.
1160 Functional default: enabled if accept_ra is enabled.
1161 disabled if accept_ra is disabled.
1163 accept_ra_rt_info_max_plen - INTEGER
1164 Maximum prefix length of Route Information in RA.
1166 Route Information w/ prefix larger than or equal to this
1167 variable shall be ignored.
1169 Functional default: 0 if accept_ra_rtr_pref is enabled.
1170 -1 if accept_ra_rtr_pref is disabled.
1172 accept_ra_rtr_pref - BOOLEAN
1173 Accept Router Preference in RA.
1175 Functional default: enabled if accept_ra is enabled.
1176 disabled if accept_ra is disabled.
1178 accept_redirects - BOOLEAN
1181 Functional default: enabled if local forwarding is disabled.
1182 disabled if local forwarding is enabled.
1184 accept_source_route - INTEGER
1185 Accept source routing (routing extension header).
1187 >= 0: Accept only routing header type 2.
1188 < 0: Do not accept routing header.
1193 Autoconfigure addresses using Prefix Information in Router
1196 Functional default: enabled if accept_ra_pinfo is enabled.
1197 disabled if accept_ra_pinfo is disabled.
1199 dad_transmits - INTEGER
1200 The amount of Duplicate Address Detection probes to send.
1203 forwarding - INTEGER
1204 Configure interface-specific Host/Router behaviour.
1206 Note: It is recommended to have the same setting on all
1207 interfaces; mixed router/host scenarios are rather uncommon.
1209 Possible values are:
1210 0 Forwarding disabled
1211 1 Forwarding enabled
1215 By default, Host behaviour is assumed. This means:
1217 1. IsRouter flag is not set in Neighbour Advertisements.
1218 2. If accept_ra is TRUE (default), transmit Router
1220 3. If accept_ra is TRUE (default), accept Router
1221 Advertisements (and do autoconfiguration).
1222 4. If accept_redirects is TRUE (default), accept Redirects.
1226 If local forwarding is enabled, Router behaviour is assumed.
1227 This means exactly the reverse from the above:
1229 1. IsRouter flag is set in Neighbour Advertisements.
1230 2. Router Solicitations are not sent unless accept_ra is 2.
1231 3. Router Advertisements are ignored unless accept_ra is 2.
1232 4. Redirects are ignored.
1234 Default: 0 (disabled) if global forwarding is disabled (default),
1235 otherwise 1 (enabled).
1238 Default Hop Limit to set.
1242 Default Maximum Transfer Unit
1243 Default: 1280 (IPv6 required minimum)
1245 router_probe_interval - INTEGER
1246 Minimum interval (in seconds) between Router Probing described
1251 router_solicitation_delay - INTEGER
1252 Number of seconds to wait after interface is brought up
1253 before sending Router Solicitations.
1256 router_solicitation_interval - INTEGER
1257 Number of seconds to wait between Router Solicitations.
1260 router_solicitations - INTEGER
1261 Number of Router Solicitations to send until assuming no
1262 routers are present.
1265 use_tempaddr - INTEGER
1266 Preference for Privacy Extensions (RFC3041).
1267 <= 0 : disable Privacy Extensions
1268 == 1 : enable Privacy Extensions, but prefer public
1269 addresses over temporary addresses.
1270 > 1 : enable Privacy Extensions and prefer temporary
1271 addresses over public addresses.
1272 Default: 0 (for most devices)
1273 -1 (for point-to-point devices and loopback devices)
1275 temp_valid_lft - INTEGER
1276 valid lifetime (in seconds) for temporary addresses.
1277 Default: 604800 (7 days)
1279 temp_prefered_lft - INTEGER
1280 Preferred lifetime (in seconds) for temporary addresses.
1281 Default: 86400 (1 day)
1283 max_desync_factor - INTEGER
1284 Maximum value for DESYNC_FACTOR, which is a random value
1285 that ensures that clients don't synchronize with each
1286 other and generate new addresses at exactly the same time.
1287 value is in seconds.
1290 regen_max_retry - INTEGER
1291 Number of attempts before give up attempting to generate
1292 valid temporary addresses.
1295 max_addresses - INTEGER
1296 Maximum number of autoconfigured addresses per interface. Setting
1297 to zero disables the limitation. It is not recommended to set this
1298 value too large (or to zero) because it would be an easy way to
1299 crash the kernel by allowing too many addresses to be created.
1302 disable_ipv6 - BOOLEAN
1303 Disable IPv6 operation. If accept_dad is set to 2, this value
1304 will be dynamically set to TRUE if DAD fails for the link-local
1306 Default: FALSE (enable IPv6 operation)
1308 When this value is changed from 1 to 0 (IPv6 is being enabled),
1309 it will dynamically create a link-local address on the given
1310 interface and start Duplicate Address Detection, if necessary.
1312 When this value is changed from 0 to 1 (IPv6 is being disabled),
1313 it will dynamically delete all address on the given interface.
1315 accept_dad - INTEGER
1316 Whether to accept DAD (Duplicate Address Detection).
1318 1: Enable DAD (default)
1319 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1320 link-local address has been found.
1322 force_tllao - BOOLEAN
1323 Enable sending the target link-layer address option even when
1324 responding to a unicast neighbor solicitation.
1327 Quoting from RFC 2461, section 4.4, Target link-layer address:
1329 "The option MUST be included for multicast solicitations in order to
1330 avoid infinite Neighbor Solicitation "recursion" when the peer node
1331 does not have a cache entry to return a Neighbor Advertisements
1332 message. When responding to unicast solicitations, the option can be
1333 omitted since the sender of the solicitation has the correct link-
1334 layer address; otherwise it would not have be able to send the unicast
1335 solicitation in the first place. However, including the link-layer
1336 address in this case adds little overhead and eliminates a potential
1337 race condition where the sender deletes the cached link-layer address
1338 prior to receiving a response to a previous solicitation."
1340 ndisc_notify - BOOLEAN
1341 Define mode for notification of address and device changes.
1342 0 - (default): do nothing
1343 1 - Generate unsolicited neighbour advertisements when device is brought
1344 up or hardware address changes.
1348 Limit the maximal rates for sending ICMPv6 packets.
1349 0 to disable any limiting,
1350 otherwise the minimal space between responses in milliseconds.
1355 Pekka Savola <pekkas@netcore.fi>
1356 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1359 /proc/sys/net/bridge/* Variables:
1361 bridge-nf-call-arptables - BOOLEAN
1362 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1366 bridge-nf-call-iptables - BOOLEAN
1367 1 : pass bridged IPv4 traffic to iptables' chains.
1371 bridge-nf-call-ip6tables - BOOLEAN
1372 1 : pass bridged IPv6 traffic to ip6tables' chains.
1376 bridge-nf-filter-vlan-tagged - BOOLEAN
1377 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1381 bridge-nf-filter-pppoe-tagged - BOOLEAN
1382 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1386 bridge-nf-pass-vlan-input-dev - BOOLEAN
1387 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
1388 interface on the bridge and set the netfilter input device to the vlan.
1389 This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
1390 target work with vlan-on-top-of-bridge interfaces. When no matching
1391 vlan interface is found, or this switch is off, the input device is
1392 set to the bridge interface.
1393 0: disable bridge netfilter vlan interface lookup.
1396 proc/sys/net/sctp/* Variables:
1398 addip_enable - BOOLEAN
1399 Enable or disable extension of Dynamic Address Reconfiguration
1400 (ADD-IP) functionality specified in RFC5061. This extension provides
1401 the ability to dynamically add and remove new addresses for the SCTP
1404 1: Enable extension.
1406 0: Disable extension.
1410 addip_noauth_enable - BOOLEAN
1411 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1412 authentication to protect the operations of adding or removing new
1413 addresses. This requirement is mandated so that unauthorized hosts
1414 would not be able to hijack associations. However, older
1415 implementations may not have implemented this requirement while
1416 allowing the ADD-IP extension. For reasons of interoperability,
1417 we provide this variable to control the enforcement of the
1418 authentication requirement.
1420 1: Allow ADD-IP extension to be used without authentication. This
1421 should only be set in a closed environment for interoperability
1422 with older implementations.
1424 0: Enforce the authentication requirement
1428 auth_enable - BOOLEAN
1429 Enable or disable Authenticated Chunks extension. This extension
1430 provides the ability to send and receive authenticated chunks and is
1431 required for secure operation of Dynamic Address Reconfiguration
1434 1: Enable this extension.
1435 0: Disable this extension.
1439 prsctp_enable - BOOLEAN
1440 Enable or disable the Partial Reliability extension (RFC3758) which
1441 is used to notify peers that a given DATA should no longer be expected.
1449 The limit of the number of new packets that can be initially sent. It
1450 controls how bursty the generated traffic can be.
1454 association_max_retrans - INTEGER
1455 Set the maximum number for retransmissions that an association can
1456 attempt deciding that the remote end is unreachable. If this value
1457 is exceeded, the association is terminated.
1461 max_init_retransmits - INTEGER
1462 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1463 that an association will attempt before declaring the destination
1464 unreachable and terminating.
1468 path_max_retrans - INTEGER
1469 The maximum number of retransmissions that will be attempted on a given
1470 path. Once this threshold is exceeded, the path is considered
1471 unreachable, and new traffic will use a different path when the
1472 association is multihomed.
1476 pf_retrans - INTEGER
1477 The number of retransmissions that will be attempted on a given path
1478 before traffic is redirected to an alternate transport (should one
1479 exist). Note this is distinct from path_max_retrans, as a path that
1480 passes the pf_retrans threshold can still be used. Its only
1481 deprioritized when a transmission path is selected by the stack. This
1482 setting is primarily used to enable fast failover mechanisms without
1483 having to reduce path_max_retrans to a very low value. See:
1484 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
1485 for details. Note also that a value of pf_retrans > path_max_retrans
1486 disables this feature
1490 rto_initial - INTEGER
1491 The initial round trip timeout value in milliseconds that will be used
1492 in calculating round trip times. This is the initial time interval
1493 for retransmissions.
1498 The maximum value (in milliseconds) of the round trip timeout. This
1499 is the largest time interval that can elapse between retransmissions.
1504 The minimum value (in milliseconds) of the round trip timeout. This
1505 is the smallest time interval the can elapse between retransmissions.
1509 hb_interval - INTEGER
1510 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
1511 are sent at the specified interval on idle paths to probe the state of
1512 a given path between 2 associations.
1516 sack_timeout - INTEGER
1517 The amount of time (in milliseconds) that the implementation will wait
1522 valid_cookie_life - INTEGER
1523 The default lifetime of the SCTP cookie (in milliseconds). The cookie
1524 is used during association establishment.
1528 cookie_preserve_enable - BOOLEAN
1529 Enable or disable the ability to extend the lifetime of the SCTP cookie
1530 that is used during the establishment phase of SCTP association
1532 1: Enable cookie lifetime extension.
1537 cookie_hmac_alg - STRING
1538 Select the hmac algorithm used when generating the cookie value sent by
1539 a listening sctp socket to a connecting client in the INIT-ACK chunk.
1544 Ability to assign md5 or sha1 as the selected alg is predicated on the
1545 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
1546 CONFIG_CRYPTO_SHA1).
1548 Default: Dependent on configuration. MD5 if available, else SHA1 if
1549 available, else none.
1551 rcvbuf_policy - INTEGER
1552 Determines if the receive buffer is attributed to the socket or to
1553 association. SCTP supports the capability to create multiple
1554 associations on a single socket. When using this capability, it is
1555 possible that a single stalled association that's buffering a lot
1556 of data may block other associations from delivering their data by
1557 consuming all of the receive buffer space. To work around this,
1558 the rcvbuf_policy could be set to attribute the receiver buffer space
1559 to each association instead of the socket. This prevents the described
1562 1: rcvbuf space is per association
1563 0: rcvbuf space is per socket
1567 sndbuf_policy - INTEGER
1568 Similar to rcvbuf_policy above, this applies to send buffer space.
1570 1: Send buffer is tracked per association
1571 0: Send buffer is tracked per socket.
1575 sctp_mem - vector of 3 INTEGERs: min, pressure, max
1576 Number of pages allowed for queueing by all SCTP sockets.
1578 min: Below this number of pages SCTP is not bothered about its
1579 memory appetite. When amount of memory allocated by SCTP exceeds
1580 this number, SCTP starts to moderate memory usage.
1582 pressure: This value was introduced to follow format of tcp_mem.
1584 max: Number of pages allowed for queueing by all SCTP sockets.
1586 Default is calculated at boot time from amount of available memory.
1588 sctp_rmem - vector of 3 INTEGERs: min, default, max
1589 Only the first value ("min") is used, "default" and "max" are
1592 min: Minimal size of receive buffer used by SCTP socket.
1593 It is guaranteed to each SCTP socket (but not association) even
1594 under moderate memory pressure.
1598 sctp_wmem - vector of 3 INTEGERs: min, default, max
1599 Currently this tunable has no effect.
1601 addr_scope_policy - INTEGER
1602 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
1604 0 - Disable IPv4 address scoping
1605 1 - Enable IPv4 address scoping
1606 2 - Follow draft but allow IPv4 private addresses
1607 3 - Follow draft but allow IPv4 link local addresses
1612 /proc/sys/net/core/*
1613 Please see: Documentation/sysctl/net.txt for descriptions of these entries.
1616 /proc/sys/net/unix/*
1617 max_dgram_qlen - INTEGER
1618 The maximum length of dgram socket receive queue
1625 /proc/sys/net/irda/*
1626 fast_poll_increase FIXME
1627 warn_noreply_time FIXME
1628 discovery_slots FIXME
1631 discovery_timeout FIXME
1632 lap_keepalive_time FIXME
1633 max_noreply_time FIXME
1634 max_tx_data_size FIXME
1636 min_tx_turn_time FIXME