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_thresh3 - INTEGER
30 Maximum number of neighbor entries allowed. Increase this
31 when using large numbers of interfaces and when communicating
32 with large numbers of directly-connected peers.
34 neigh/default/unres_qlen_bytes - INTEGER
35 The maximum number of bytes which may be used by packets
36 queued for each unresolved address by other network layers.
39 neigh/default/unres_qlen - INTEGER
40 The maximum number of packets which may be queued for each
41 unresolved address by other network layers.
42 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
45 Time, in seconds, that cached PMTU information is kept.
48 The advertised MSS depends on the first hop route MTU, but will
49 never be lower than this setting.
51 rt_cache_rebuild_count - INTEGER
52 The per net-namespace route cache emergency rebuild threshold.
53 Any net-namespace having its route cache rebuilt due to
54 a hash bucket chain being too long more than this many times
55 will have its route caching disabled
59 ipfrag_high_thresh - INTEGER
60 Maximum memory used to reassemble IP fragments. When
61 ipfrag_high_thresh bytes of memory is allocated for this purpose,
62 the fragment handler will toss packets until ipfrag_low_thresh
65 ipfrag_low_thresh - INTEGER
66 See ipfrag_high_thresh
69 Time in seconds to keep an IP fragment in memory.
71 ipfrag_secret_interval - INTEGER
72 Regeneration interval (in seconds) of the hash secret (or lifetime
73 for the hash secret) for IP fragments.
76 ipfrag_max_dist - INTEGER
77 ipfrag_max_dist is a non-negative integer value which defines the
78 maximum "disorder" which is allowed among fragments which share a
79 common IP source address. Note that reordering of packets is
80 not unusual, but if a large number of fragments arrive from a source
81 IP address while a particular fragment queue remains incomplete, it
82 probably indicates that one or more fragments belonging to that queue
83 have been lost. When ipfrag_max_dist is positive, an additional check
84 is done on fragments before they are added to a reassembly queue - if
85 ipfrag_max_dist (or more) fragments have arrived from a particular IP
86 address between additions to any IP fragment queue using that source
87 address, it's presumed that one or more fragments in the queue are
88 lost. The existing fragment queue will be dropped, and a new one
89 started. An ipfrag_max_dist value of zero disables this check.
91 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
92 result in unnecessarily dropping fragment queues when normal
93 reordering of packets occurs, which could lead to poor application
94 performance. Using a very large value, e.g. 50000, increases the
95 likelihood of incorrectly reassembling IP fragments that originate
96 from different IP datagrams, which could result in data corruption.
101 inet_peer_threshold - INTEGER
102 The approximate size of the storage. Starting from this threshold
103 entries will be thrown aggressively. This threshold also determines
104 entries' time-to-live and time intervals between garbage collection
105 passes. More entries, less time-to-live, less GC interval.
107 inet_peer_minttl - INTEGER
108 Minimum time-to-live of entries. Should be enough to cover fragment
109 time-to-live on the reassembling side. This minimum time-to-live is
110 guaranteed if the pool size is less than inet_peer_threshold.
113 inet_peer_maxttl - INTEGER
114 Maximum time-to-live of entries. Unused entries will expire after
115 this period of time if there is no memory pressure on the pool (i.e.
116 when the number of entries in the pool is very small).
122 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
123 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
127 Controls Appropriate Byte Count (ABC) defined in RFC3465.
128 ABC is a way of increasing congestion window (cwnd) more slowly
129 in response to partial acknowledgments.
131 0 increase cwnd once per acknowledgment (no ABC)
132 1 increase cwnd once per acknowledgment of full sized segment
133 2 allow increase cwnd by two if acknowledgment is
134 of two segments to compensate for delayed acknowledgments.
137 tcp_abort_on_overflow - BOOLEAN
138 If listening service is too slow to accept new connections,
139 reset them. Default state is FALSE. It means that if overflow
140 occurred due to a burst, connection will recover. Enable this
141 option _only_ if you are really sure that listening daemon
142 cannot be tuned to accept connections faster. Enabling this
143 option can harm clients of your server.
145 tcp_adv_win_scale - INTEGER
146 Count buffering overhead as bytes/2^tcp_adv_win_scale
147 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
149 Possible values are [-31, 31], inclusive.
152 tcp_allowed_congestion_control - STRING
153 Show/set the congestion control choices available to non-privileged
154 processes. The list is a subset of those listed in
155 tcp_available_congestion_control.
156 Default is "reno" and the default setting (tcp_congestion_control).
158 tcp_app_win - INTEGER
159 Reserve max(window/2^tcp_app_win, mss) of window for application
160 buffer. Value 0 is special, it means that nothing is reserved.
163 tcp_available_congestion_control - STRING
164 Shows the available congestion control choices that are registered.
165 More congestion control algorithms may be available as modules,
168 tcp_base_mss - INTEGER
169 The initial value of search_low to be used by the packetization layer
170 Path MTU discovery (MTU probing). If MTU probing is enabled,
171 this is the initial MSS used by the connection.
173 tcp_congestion_control - STRING
174 Set the congestion control algorithm to be used for new
175 connections. The algorithm "reno" is always available, but
176 additional choices may be available based on kernel configuration.
177 Default is set as part of kernel configuration.
178 For passive connections, the listener congestion control choice
180 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
182 tcp_cookie_size - INTEGER
183 Default size of TCP Cookie Transactions (TCPCT) option, that may be
184 overridden on a per socket basis by the TCPCT socket option.
185 Values greater than the maximum (16) are interpreted as the maximum.
186 Values greater than zero and less than the minimum (8) are interpreted
187 as the minimum. Odd values are interpreted as the next even value.
191 Allows TCP to send "duplicate" SACKs.
193 tcp_early_retrans - INTEGER
194 Enable Early Retransmit (ER), per RFC 5827. ER lowers the threshold
195 for triggering fast retransmit when the amount of outstanding data is
196 small and when no previously unsent data can be transmitted (such
197 that limited transmit could be used).
201 2 enables ER but delays fast recovery and fast retransmit
202 by a fourth of RTT. This mitigates connection falsely
203 recovers when network has a small degree of reordering
204 (less than 3 packets).
208 Enable Explicit Congestion Notification (ECN) in TCP. ECN is only
209 used when both ends of the TCP flow support it. It is useful to
210 avoid losses due to congestion (when the bottleneck router supports
215 2 Only server-side ECN enabled. If the other end does
216 not support ECN, behavior is like with ECN disabled.
220 Enable FACK congestion avoidance and fast retransmission.
221 The value is not used, if tcp_sack is not enabled.
223 tcp_fin_timeout - INTEGER
224 Time to hold socket in state FIN-WAIT-2, if it was closed
225 by our side. Peer can be broken and never close its side,
226 or even died unexpectedly. Default value is 60sec.
227 Usual value used in 2.2 was 180 seconds, you may restore
228 it, but remember that if your machine is even underloaded WEB server,
229 you risk to overflow memory with kilotons of dead sockets,
230 FIN-WAIT-2 sockets are less dangerous than FIN-WAIT-1,
231 because they eat maximum 1.5K of memory, but they tend
232 to live longer. Cf. tcp_max_orphans.
235 Enables Forward RTO-Recovery (F-RTO) defined in RFC4138.
236 F-RTO is an enhanced recovery algorithm for TCP retransmission
237 timeouts. It is particularly beneficial in wireless environments
238 where packet loss is typically due to random radio interference
239 rather than intermediate router congestion. F-RTO is sender-side
240 only modification. Therefore it does not require any support from
243 If set to 1, basic version is enabled. 2 enables SACK enhanced
244 F-RTO if flow uses SACK. The basic version can be used also when
245 SACK is in use though scenario(s) with it exists where F-RTO
246 interacts badly with the packet counting of the SACK enabled TCP
249 tcp_frto_response - INTEGER
250 When F-RTO has detected that a TCP retransmission timeout was
251 spurious (i.e, the timeout would have been avoided had TCP set a
252 longer retransmission timeout), TCP has several options what to do
253 next. Possible values are:
254 0 Rate halving based; a smooth and conservative response,
255 results in halved cwnd and ssthresh after one RTT
256 1 Very conservative response; not recommended because even
257 though being valid, it interacts poorly with the rest of
258 Linux TCP, halves cwnd and ssthresh immediately
259 2 Aggressive response; undoes congestion control measures
260 that are now known to be unnecessary (ignoring the
261 possibility of a lost retransmission that would require
262 TCP to be more cautious), cwnd and ssthresh are restored
263 to the values prior timeout
264 Default: 0 (rate halving based)
266 tcp_keepalive_time - INTEGER
267 How often TCP sends out keepalive messages when keepalive is enabled.
270 tcp_keepalive_probes - INTEGER
271 How many keepalive probes TCP sends out, until it decides that the
272 connection is broken. Default value: 9.
274 tcp_keepalive_intvl - INTEGER
275 How frequently the probes are send out. Multiplied by
276 tcp_keepalive_probes it is time to kill not responding connection,
277 after probes started. Default value: 75sec i.e. connection
278 will be aborted after ~11 minutes of retries.
280 tcp_low_latency - BOOLEAN
281 If set, the TCP stack makes decisions that prefer lower
282 latency as opposed to higher throughput. By default, this
283 option is not set meaning that higher throughput is preferred.
284 An example of an application where this default should be
285 changed would be a Beowulf compute cluster.
288 tcp_max_orphans - INTEGER
289 Maximal number of TCP sockets not attached to any user file handle,
290 held by system. If this number is exceeded orphaned connections are
291 reset immediately and warning is printed. This limit exists
292 only to prevent simple DoS attacks, you _must_ not rely on this
293 or lower the limit artificially, but rather increase it
294 (probably, after increasing installed memory),
295 if network conditions require more than default value,
296 and tune network services to linger and kill such states
297 more aggressively. Let me to remind again: each orphan eats
298 up to ~64K of unswappable memory.
300 tcp_max_ssthresh - INTEGER
301 Limited Slow-Start for TCP with large congestion windows (cwnd) defined in
302 RFC3742. Limited slow-start is a mechanism to limit growth of the cwnd
303 on the region where cwnd is larger than tcp_max_ssthresh. TCP increases cwnd
304 by at most tcp_max_ssthresh segments, and by at least tcp_max_ssthresh/2
305 segments per RTT when the cwnd is above tcp_max_ssthresh.
306 If TCP connection increased cwnd to thousands (or tens of thousands) segments,
307 and thousands of packets were being dropped during slow-start, you can set
308 tcp_max_ssthresh to improve performance for new TCP connection.
311 tcp_max_syn_backlog - INTEGER
312 Maximal number of remembered connection requests, which have not
313 received an acknowledgment from connecting client.
314 The minimal value is 128 for low memory machines, and it will
315 increase in proportion to the memory of machine.
316 If server suffers from overload, try increasing this number.
318 tcp_max_tw_buckets - INTEGER
319 Maximal number of timewait sockets held by system simultaneously.
320 If this number is exceeded time-wait socket is immediately destroyed
321 and warning is printed. This limit exists only to prevent
322 simple DoS attacks, you _must_ not lower the limit artificially,
323 but rather increase it (probably, after increasing installed memory),
324 if network conditions require more than default value.
326 tcp_mem - vector of 3 INTEGERs: min, pressure, max
327 min: below this number of pages TCP is not bothered about its
330 pressure: when amount of memory allocated by TCP exceeds this number
331 of pages, TCP moderates its memory consumption and enters memory
332 pressure mode, which is exited when memory consumption falls
335 max: number of pages allowed for queueing by all TCP sockets.
337 Defaults are calculated at boot time from amount of available
340 tcp_moderate_rcvbuf - BOOLEAN
341 If set, TCP performs receive buffer auto-tuning, attempting to
342 automatically size the buffer (no greater than tcp_rmem[2]) to
343 match the size required by the path for full throughput. Enabled by
346 tcp_mtu_probing - INTEGER
347 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
350 1 - Disabled by default, enabled when an ICMP black hole detected
351 2 - Always enabled, use initial MSS of tcp_base_mss.
353 tcp_no_metrics_save - BOOLEAN
354 By default, TCP saves various connection metrics in the route cache
355 when the connection closes, so that connections established in the
356 near future can use these to set initial conditions. Usually, this
357 increases overall performance, but may sometimes cause performance
358 degradation. If set, TCP will not cache metrics on closing
361 tcp_orphan_retries - INTEGER
362 This value influences the timeout of a locally closed TCP connection,
363 when RTO retransmissions remain unacknowledged.
364 See tcp_retries2 for more details.
366 The default value is 8.
367 If your machine is a loaded WEB server,
368 you should think about lowering this value, such sockets
369 may consume significant resources. Cf. tcp_max_orphans.
371 tcp_reordering - INTEGER
372 Maximal reordering of packets in a TCP stream.
375 tcp_retrans_collapse - BOOLEAN
376 Bug-to-bug compatibility with some broken printers.
377 On retransmit try to send bigger packets to work around bugs in
380 tcp_retries1 - INTEGER
381 This value influences the time, after which TCP decides, that
382 something is wrong due to unacknowledged RTO retransmissions,
383 and reports this suspicion to the network layer.
384 See tcp_retries2 for more details.
386 RFC 1122 recommends at least 3 retransmissions, which is the
389 tcp_retries2 - INTEGER
390 This value influences the timeout of an alive TCP connection,
391 when RTO retransmissions remain unacknowledged.
392 Given a value of N, a hypothetical TCP connection following
393 exponential backoff with an initial RTO of TCP_RTO_MIN would
394 retransmit N times before killing the connection at the (N+1)th RTO.
396 The default value of 15 yields a hypothetical timeout of 924.6
397 seconds and is a lower bound for the effective timeout.
398 TCP will effectively time out at the first RTO which exceeds the
399 hypothetical timeout.
401 RFC 1122 recommends at least 100 seconds for the timeout,
402 which corresponds to a value of at least 8.
404 tcp_rfc1337 - BOOLEAN
405 If set, the TCP stack behaves conforming to RFC1337. If unset,
406 we are not conforming to RFC, but prevent TCP TIME_WAIT
410 tcp_rmem - vector of 3 INTEGERs: min, default, max
411 min: Minimal size of receive buffer used by TCP sockets.
412 It is guaranteed to each TCP socket, even under moderate memory
416 default: initial size of receive buffer used by TCP sockets.
417 This value overrides net.core.rmem_default used by other protocols.
418 Default: 87380 bytes. This value results in window of 65535 with
419 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
420 less for default tcp_app_win. See below about these variables.
422 max: maximal size of receive buffer allowed for automatically
423 selected receiver buffers for TCP socket. This value does not override
424 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
425 automatic tuning of that socket's receive buffer size, in which
426 case this value is ignored.
427 Default: between 87380B and 6MB, depending on RAM size.
430 Enable select acknowledgments (SACKS).
432 tcp_slow_start_after_idle - BOOLEAN
433 If set, provide RFC2861 behavior and time out the congestion
434 window after an idle period. An idle period is defined at
435 the current RTO. If unset, the congestion window will not
436 be timed out after an idle period.
440 Use the Host requirements interpretation of the TCP urgent pointer field.
441 Most hosts use the older BSD interpretation, so if you turn this on
442 Linux might not communicate correctly with them.
445 tcp_synack_retries - INTEGER
446 Number of times SYNACKs for a passive TCP connection attempt will
447 be retransmitted. Should not be higher than 255. Default value
448 is 5, which corresponds to ~180seconds.
450 tcp_syncookies - BOOLEAN
451 Only valid when the kernel was compiled with CONFIG_SYNCOOKIES
452 Send out syncookies when the syn backlog queue of a socket
453 overflows. This is to prevent against the common 'SYN flood attack'
456 Note, that syncookies is fallback facility.
457 It MUST NOT be used to help highly loaded servers to stand
458 against legal connection rate. If you see SYN flood warnings
459 in your logs, but investigation shows that they occur
460 because of overload with legal connections, you should tune
461 another parameters until this warning disappear.
462 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
464 syncookies seriously violate TCP protocol, do not allow
465 to use TCP extensions, can result in serious degradation
466 of some services (f.e. SMTP relaying), visible not by you,
467 but your clients and relays, contacting you. While you see
468 SYN flood warnings in logs not being really flooded, your server
469 is seriously misconfigured.
471 tcp_fastopen - INTEGER
472 Enable TCP Fast Open feature (draft-ietf-tcpm-fastopen) to send data
473 in the opening SYN packet. To use this feature, the client application
474 must not use connect(). Instead, it should use sendmsg() or sendto()
475 with MSG_FASTOPEN flag which performs a TCP handshake automatically.
477 The values (bitmap) are:
478 1: Enables sending data in the opening SYN on the client
479 5: Enables sending data in the opening SYN on the client regardless
480 of cookie availability.
484 tcp_syn_retries - INTEGER
485 Number of times initial SYNs for an active TCP connection attempt
486 will be retransmitted. Should not be higher than 255. Default value
487 is 5, which corresponds to ~180seconds.
489 tcp_timestamps - BOOLEAN
490 Enable timestamps as defined in RFC1323.
492 tcp_tso_win_divisor - INTEGER
493 This allows control over what percentage of the congestion window
494 can be consumed by a single TSO frame.
495 The setting of this parameter is a choice between burstiness and
496 building larger TSO frames.
499 tcp_tw_recycle - BOOLEAN
500 Enable fast recycling TIME-WAIT sockets. Default value is 0.
501 It should not be changed without advice/request of technical
504 tcp_tw_reuse - BOOLEAN
505 Allow to reuse TIME-WAIT sockets for new connections when it is
506 safe from protocol viewpoint. Default value is 0.
507 It should not be changed without advice/request of technical
510 tcp_window_scaling - BOOLEAN
511 Enable window scaling as defined in RFC1323.
513 tcp_wmem - vector of 3 INTEGERs: min, default, max
514 min: Amount of memory reserved for send buffers for TCP sockets.
515 Each TCP socket has rights to use it due to fact of its birth.
518 default: initial size of send buffer used by TCP sockets. This
519 value overrides net.core.wmem_default used by other protocols.
520 It is usually lower than net.core.wmem_default.
523 max: Maximal amount of memory allowed for automatically tuned
524 send buffers for TCP sockets. This value does not override
525 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
526 automatic tuning of that socket's send buffer size, in which case
527 this value is ignored.
528 Default: between 64K and 4MB, depending on RAM size.
530 tcp_workaround_signed_windows - BOOLEAN
531 If set, assume no receipt of a window scaling option means the
532 remote TCP is broken and treats the window as a signed quantity.
533 If unset, assume the remote TCP is not broken even if we do
534 not receive a window scaling option from them.
537 tcp_dma_copybreak - INTEGER
538 Lower limit, in bytes, of the size of socket reads that will be
539 offloaded to a DMA copy engine, if one is present in the system
540 and CONFIG_NET_DMA is enabled.
543 tcp_thin_linear_timeouts - BOOLEAN
544 Enable dynamic triggering of linear timeouts for thin streams.
545 If set, a check is performed upon retransmission by timeout to
546 determine if the stream is thin (less than 4 packets in flight).
547 As long as the stream is found to be thin, up to 6 linear
548 timeouts may be performed before exponential backoff mode is
549 initiated. This improves retransmission latency for
550 non-aggressive thin streams, often found to be time-dependent.
551 For more information on thin streams, see
552 Documentation/networking/tcp-thin.txt
555 tcp_thin_dupack - BOOLEAN
556 Enable dynamic triggering of retransmissions after one dupACK
557 for thin streams. If set, a check is performed upon reception
558 of a dupACK to determine if the stream is thin (less than 4
559 packets in flight). As long as the stream is found to be thin,
560 data is retransmitted on the first received dupACK. This
561 improves retransmission latency for non-aggressive thin
562 streams, often found to be time-dependent.
563 For more information on thin streams, see
564 Documentation/networking/tcp-thin.txt
567 tcp_limit_output_bytes - INTEGER
568 Controls TCP Small Queue limit per tcp socket.
569 TCP bulk sender tends to increase packets in flight until it
570 gets losses notifications. With SNDBUF autotuning, this can
571 result in a large amount of packets queued in qdisc/device
572 on the local machine, hurting latency of other flows, for
573 typical pfifo_fast qdiscs.
574 tcp_limit_output_bytes limits the number of bytes on qdisc
575 or device to reduce artificial RTT/cwnd and reduce bufferbloat.
576 Note: For GSO/TSO enabled flows, we try to have at least two
577 packets in flight. Reducing tcp_limit_output_bytes might also
578 reduce the size of individual GSO packet (64KB being the max)
581 tcp_challenge_ack_limit - INTEGER
582 Limits number of Challenge ACK sent per second, as recommended
583 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
588 udp_mem - vector of 3 INTEGERs: min, pressure, max
589 Number of pages allowed for queueing by all UDP sockets.
591 min: Below this number of pages UDP is not bothered about its
592 memory appetite. When amount of memory allocated by UDP exceeds
593 this number, UDP starts to moderate memory usage.
595 pressure: This value was introduced to follow format of tcp_mem.
597 max: Number of pages allowed for queueing by all UDP sockets.
599 Default is calculated at boot time from amount of available memory.
601 udp_rmem_min - INTEGER
602 Minimal size of receive buffer used by UDP sockets in moderation.
603 Each UDP socket is able to use the size for receiving data, even if
604 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
607 udp_wmem_min - INTEGER
608 Minimal size of send buffer used by UDP sockets in moderation.
609 Each UDP socket is able to use the size for sending data, even if
610 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
615 cipso_cache_enable - BOOLEAN
616 If set, enable additions to and lookups from the CIPSO label mapping
617 cache. If unset, additions are ignored and lookups always result in a
618 miss. However, regardless of the setting the cache is still
619 invalidated when required when means you can safely toggle this on and
620 off and the cache will always be "safe".
623 cipso_cache_bucket_size - INTEGER
624 The CIPSO label cache consists of a fixed size hash table with each
625 hash bucket containing a number of cache entries. This variable limits
626 the number of entries in each hash bucket; the larger the value the
627 more CIPSO label mappings that can be cached. When the number of
628 entries in a given hash bucket reaches this limit adding new entries
629 causes the oldest entry in the bucket to be removed to make room.
632 cipso_rbm_optfmt - BOOLEAN
633 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
634 the CIPSO draft specification (see Documentation/netlabel for details).
635 This means that when set the CIPSO tag will be padded with empty
636 categories in order to make the packet data 32-bit aligned.
639 cipso_rbm_structvalid - BOOLEAN
640 If set, do a very strict check of the CIPSO option when
641 ip_options_compile() is called. If unset, relax the checks done during
642 ip_options_compile(). Either way is "safe" as errors are caught else
643 where in the CIPSO processing code but setting this to 0 (False) should
644 result in less work (i.e. it should be faster) but could cause problems
645 with other implementations that require strict checking.
650 ip_local_port_range - 2 INTEGERS
651 Defines the local port range that is used by TCP and UDP to
652 choose the local port. The first number is the first, the
653 second the last local port number. The default values are
654 32768 and 61000 respectively.
656 ip_local_reserved_ports - list of comma separated ranges
657 Specify the ports which are reserved for known third-party
658 applications. These ports will not be used by automatic port
659 assignments (e.g. when calling connect() or bind() with port
660 number 0). Explicit port allocation behavior is unchanged.
662 The format used for both input and output is a comma separated
663 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
664 10). Writing to the file will clear all previously reserved
665 ports and update the current list with the one given in the
668 Note that ip_local_port_range and ip_local_reserved_ports
669 settings are independent and both are considered by the kernel
670 when determining which ports are available for automatic port
673 You can reserve ports which are not in the current
674 ip_local_port_range, e.g.:
676 $ cat /proc/sys/net/ipv4/ip_local_port_range
678 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
681 although this is redundant. However such a setting is useful
682 if later the port range is changed to a value that will
683 include the reserved ports.
687 ip_nonlocal_bind - BOOLEAN
688 If set, allows processes to bind() to non-local IP addresses,
689 which can be quite useful - but may break some applications.
693 If set non-zero, enables support for dynamic addresses.
694 If set to a non-zero value larger than 1, a kernel log
695 message will be printed when dynamic address rewriting
699 icmp_echo_ignore_all - BOOLEAN
700 If set non-zero, then the kernel will ignore all ICMP ECHO
704 icmp_echo_ignore_broadcasts - BOOLEAN
705 If set non-zero, then the kernel will ignore all ICMP ECHO and
706 TIMESTAMP requests sent to it via broadcast/multicast.
709 icmp_ratelimit - INTEGER
710 Limit the maximal rates for sending ICMP packets whose type matches
711 icmp_ratemask (see below) to specific targets.
712 0 to disable any limiting,
713 otherwise the minimal space between responses in milliseconds.
716 icmp_ratemask - INTEGER
717 Mask made of ICMP types for which rates are being limited.
718 Significant bits: IHGFEDCBA9876543210
719 Default mask: 0000001100000011000 (6168)
721 Bit definitions (see include/linux/icmp.h):
723 3 Destination Unreachable *
728 C Parameter Problem *
733 H Address Mask Request
736 * These are rate limited by default (see default mask above)
738 icmp_ignore_bogus_error_responses - BOOLEAN
739 Some routers violate RFC1122 by sending bogus responses to broadcast
740 frames. Such violations are normally logged via a kernel warning.
741 If this is set to TRUE, the kernel will not give such warnings, which
742 will avoid log file clutter.
745 icmp_errors_use_inbound_ifaddr - BOOLEAN
747 If zero, icmp error messages are sent with the primary address of
748 the exiting interface.
750 If non-zero, the message will be sent with the primary address of
751 the interface that received the packet that caused the icmp error.
752 This is the behaviour network many administrators will expect from
753 a router. And it can make debugging complicated network layouts
756 Note that if no primary address exists for the interface selected,
757 then the primary address of the first non-loopback interface that
758 has one will be used regardless of this setting.
762 igmp_max_memberships - INTEGER
763 Change the maximum number of multicast groups we can subscribe to.
766 Theoretical maximum value is bounded by having to send a membership
767 report in a single datagram (i.e. the report can't span multiple
768 datagrams, or risk confusing the switch and leaving groups you don't
771 The number of supported groups 'M' is bounded by the number of group
772 report entries you can fit into a single datagram of 65535 bytes.
774 M = 65536-sizeof (ip header)/(sizeof(Group record))
776 Group records are variable length, with a minimum of 12 bytes.
777 So net.ipv4.igmp_max_memberships should not be set higher than:
779 (65536-24) / 12 = 5459
781 The value 5459 assumes no IP header options, so in practice
782 this number may be lower.
784 conf/interface/* changes special settings per interface (where
785 "interface" is the name of your network interface)
787 conf/all/* is special, changes the settings for all interfaces
789 log_martians - BOOLEAN
790 Log packets with impossible addresses to kernel log.
791 log_martians for the interface will be enabled if at least one of
792 conf/{all,interface}/log_martians is set to TRUE,
793 it will be disabled otherwise
795 accept_redirects - BOOLEAN
796 Accept ICMP redirect messages.
797 accept_redirects for the interface will be enabled if:
798 - both conf/{all,interface}/accept_redirects are TRUE in the case
799 forwarding for the interface is enabled
801 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
802 case forwarding for the interface is disabled
803 accept_redirects for the interface will be disabled otherwise
808 Enable IP forwarding on this interface.
810 mc_forwarding - BOOLEAN
811 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
812 and a multicast routing daemon is required.
813 conf/all/mc_forwarding must also be set to TRUE to enable multicast
814 routing for the interface
817 Integer value used to differentiate the devices by the medium they
818 are attached to. Two devices can have different id values when
819 the broadcast packets are received only on one of them.
820 The default value 0 means that the device is the only interface
821 to its medium, value of -1 means that medium is not known.
823 Currently, it is used to change the proxy_arp behavior:
824 the proxy_arp feature is enabled for packets forwarded between
825 two devices attached to different media.
829 proxy_arp for the interface will be enabled if at least one of
830 conf/{all,interface}/proxy_arp is set to TRUE,
831 it will be disabled otherwise
833 proxy_arp_pvlan - BOOLEAN
834 Private VLAN proxy arp.
835 Basically allow proxy arp replies back to the same interface
836 (from which the ARP request/solicitation was received).
838 This is done to support (ethernet) switch features, like RFC
839 3069, where the individual ports are NOT allowed to
840 communicate with each other, but they are allowed to talk to
841 the upstream router. As described in RFC 3069, it is possible
842 to allow these hosts to communicate through the upstream
843 router by proxy_arp'ing. Don't need to be used together with
846 This technology is known by different names:
847 In RFC 3069 it is called VLAN Aggregation.
848 Cisco and Allied Telesyn call it Private VLAN.
849 Hewlett-Packard call it Source-Port filtering or port-isolation.
850 Ericsson call it MAC-Forced Forwarding (RFC Draft).
852 shared_media - BOOLEAN
853 Send(router) or accept(host) RFC1620 shared media redirects.
854 Overrides ip_secure_redirects.
855 shared_media for the interface will be enabled if at least one of
856 conf/{all,interface}/shared_media is set to TRUE,
857 it will be disabled otherwise
860 secure_redirects - BOOLEAN
861 Accept ICMP redirect messages only for gateways,
862 listed in default gateway list.
863 secure_redirects for the interface will be enabled if at least one of
864 conf/{all,interface}/secure_redirects is set to TRUE,
865 it will be disabled otherwise
868 send_redirects - BOOLEAN
869 Send redirects, if router.
870 send_redirects for the interface will be enabled if at least one of
871 conf/{all,interface}/send_redirects is set to TRUE,
872 it will be disabled otherwise
875 bootp_relay - BOOLEAN
876 Accept packets with source address 0.b.c.d destined
877 not to this host as local ones. It is supposed, that
878 BOOTP relay daemon will catch and forward such packets.
879 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
884 accept_source_route - BOOLEAN
885 Accept packets with SRR option.
886 conf/all/accept_source_route must also be set to TRUE to accept packets
887 with SRR option on the interface
888 default TRUE (router)
891 accept_local - BOOLEAN
892 Accept packets with local source addresses. In combination
893 with suitable routing, this can be used to direct packets
894 between two local interfaces over the wire and have them
897 rp_filter must be set to a non-zero value in order for
898 accept_local to have an effect.
902 route_localnet - BOOLEAN
903 Do not consider loopback addresses as martian source or destination
904 while routing. This enables the use of 127/8 for local routing purposes.
908 0 - No source validation.
909 1 - Strict mode as defined in RFC3704 Strict Reverse Path
910 Each incoming packet is tested against the FIB and if the interface
911 is not the best reverse path the packet check will fail.
912 By default failed packets are discarded.
913 2 - Loose mode as defined in RFC3704 Loose Reverse Path
914 Each incoming packet's source address is also tested against the FIB
915 and if the source address is not reachable via any interface
916 the packet check will fail.
918 Current recommended practice in RFC3704 is to enable strict mode
919 to prevent IP spoofing from DDos attacks. If using asymmetric routing
920 or other complicated routing, then loose mode is recommended.
922 The max value from conf/{all,interface}/rp_filter is used
923 when doing source validation on the {interface}.
925 Default value is 0. Note that some distributions enable it
929 1 - Allows you to have multiple network interfaces on the same
930 subnet, and have the ARPs for each interface be answered
931 based on whether or not the kernel would route a packet from
932 the ARP'd IP out that interface (therefore you must use source
933 based routing for this to work). In other words it allows control
934 of which cards (usually 1) will respond to an arp request.
936 0 - (default) The kernel can respond to arp requests with addresses
937 from other interfaces. This may seem wrong but it usually makes
938 sense, because it increases the chance of successful communication.
939 IP addresses are owned by the complete host on Linux, not by
940 particular interfaces. Only for more complex setups like load-
941 balancing, does this behaviour cause problems.
943 arp_filter for the interface will be enabled if at least one of
944 conf/{all,interface}/arp_filter is set to TRUE,
945 it will be disabled otherwise
947 arp_announce - INTEGER
948 Define different restriction levels for announcing the local
949 source IP address from IP packets in ARP requests sent on
951 0 - (default) Use any local address, configured on any interface
952 1 - Try to avoid local addresses that are not in the target's
953 subnet for this interface. This mode is useful when target
954 hosts reachable via this interface require the source IP
955 address in ARP requests to be part of their logical network
956 configured on the receiving interface. When we generate the
957 request we will check all our subnets that include the
958 target IP and will preserve the source address if it is from
959 such subnet. If there is no such subnet we select source
960 address according to the rules for level 2.
961 2 - Always use the best local address for this target.
962 In this mode we ignore the source address in the IP packet
963 and try to select local address that we prefer for talks with
964 the target host. Such local address is selected by looking
965 for primary IP addresses on all our subnets on the outgoing
966 interface that include the target IP address. If no suitable
967 local address is found we select the first local address
968 we have on the outgoing interface or on all other interfaces,
969 with the hope we will receive reply for our request and
970 even sometimes no matter the source IP address we announce.
972 The max value from conf/{all,interface}/arp_announce is used.
974 Increasing the restriction level gives more chance for
975 receiving answer from the resolved target while decreasing
976 the level announces more valid sender's information.
979 Define different modes for sending replies in response to
980 received ARP requests that resolve local target IP addresses:
981 0 - (default): reply for any local target IP address, configured
983 1 - reply only if the target IP address is local address
984 configured on the incoming interface
985 2 - reply only if the target IP address is local address
986 configured on the incoming interface and both with the
987 sender's IP address are part from same subnet on this interface
988 3 - do not reply for local addresses configured with scope host,
989 only resolutions for global and link addresses are replied
991 8 - do not reply for all local addresses
993 The max value from conf/{all,interface}/arp_ignore is used
994 when ARP request is received on the {interface}
997 Define mode for notification of address and device changes.
998 0 - (default): do nothing
999 1 - Generate gratuitous arp requests when device is brought up
1000 or hardware address changes.
1002 arp_accept - BOOLEAN
1003 Define behavior for gratuitous ARP frames who's IP is not
1004 already present in the ARP table:
1005 0 - don't create new entries in the ARP table
1006 1 - create new entries in the ARP table
1008 Both replies and requests type gratuitous arp will trigger the
1009 ARP table to be updated, if this setting is on.
1011 If the ARP table already contains the IP address of the
1012 gratuitous arp frame, the arp table will be updated regardless
1013 if this setting is on or off.
1016 app_solicit - INTEGER
1017 The maximum number of probes to send to the user space ARP daemon
1018 via netlink before dropping back to multicast probes (see
1019 mcast_solicit). Defaults to 0.
1021 disable_policy - BOOLEAN
1022 Disable IPSEC policy (SPD) for this interface
1024 disable_xfrm - BOOLEAN
1025 Disable IPSEC encryption on this interface, whatever the policy
1030 Allows you to write a number, which can be used as required.
1034 kuznet@ms2.inr.ac.ru
1040 delon.nicolas@wanadoo.fr
1045 /proc/sys/net/ipv6/* Variables:
1047 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1048 apply to IPv6 [XXX?].
1050 bindv6only - BOOLEAN
1051 Default value for IPV6_V6ONLY socket option,
1052 which restricts use of the IPv6 socket to IPv6 communication
1054 TRUE: disable IPv4-mapped address feature
1055 FALSE: enable IPv4-mapped address feature
1057 Default: FALSE (as specified in RFC3493)
1061 ip6frag_high_thresh - INTEGER
1062 Maximum memory used to reassemble IPv6 fragments. When
1063 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1064 the fragment handler will toss packets until ip6frag_low_thresh
1067 ip6frag_low_thresh - INTEGER
1068 See ip6frag_high_thresh
1070 ip6frag_time - INTEGER
1071 Time in seconds to keep an IPv6 fragment in memory.
1073 ip6frag_secret_interval - INTEGER
1074 Regeneration interval (in seconds) of the hash secret (or lifetime
1075 for the hash secret) for IPv6 fragments.
1079 Change the interface-specific default settings.
1083 Change all the interface-specific settings.
1085 [XXX: Other special features than forwarding?]
1087 conf/all/forwarding - BOOLEAN
1088 Enable global IPv6 forwarding between all interfaces.
1090 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1091 to control which interfaces may forward packets and which not.
1093 This also sets all interfaces' Host/Router setting
1094 'forwarding' to the specified value. See below for details.
1096 This referred to as global forwarding.
1102 Change special settings per interface.
1104 The functional behaviour for certain settings is different
1105 depending on whether local forwarding is enabled or not.
1108 Accept Router Advertisements; autoconfigure using them.
1110 It also determines whether or not to transmit Router
1111 Solicitations. If and only if the functional setting is to
1112 accept Router Advertisements, Router Solicitations will be
1115 Possible values are:
1116 0 Do not accept Router Advertisements.
1117 1 Accept Router Advertisements if forwarding is disabled.
1118 2 Overrule forwarding behaviour. Accept Router Advertisements
1119 even if forwarding is enabled.
1121 Functional default: enabled if local forwarding is disabled.
1122 disabled if local forwarding is enabled.
1124 accept_ra_defrtr - BOOLEAN
1125 Learn default router in Router Advertisement.
1127 Functional default: enabled if accept_ra is enabled.
1128 disabled if accept_ra is disabled.
1130 accept_ra_pinfo - BOOLEAN
1131 Learn Prefix Information in Router Advertisement.
1133 Functional default: enabled if accept_ra is enabled.
1134 disabled if accept_ra is disabled.
1136 accept_ra_rt_info_max_plen - INTEGER
1137 Maximum prefix length of Route Information in RA.
1139 Route Information w/ prefix larger than or equal to this
1140 variable shall be ignored.
1142 Functional default: 0 if accept_ra_rtr_pref is enabled.
1143 -1 if accept_ra_rtr_pref is disabled.
1145 accept_ra_rtr_pref - BOOLEAN
1146 Accept Router Preference in RA.
1148 Functional default: enabled if accept_ra is enabled.
1149 disabled if accept_ra is disabled.
1151 accept_redirects - BOOLEAN
1154 Functional default: enabled if local forwarding is disabled.
1155 disabled if local forwarding is enabled.
1157 accept_source_route - INTEGER
1158 Accept source routing (routing extension header).
1160 >= 0: Accept only routing header type 2.
1161 < 0: Do not accept routing header.
1166 Autoconfigure addresses using Prefix Information in Router
1169 Functional default: enabled if accept_ra_pinfo is enabled.
1170 disabled if accept_ra_pinfo is disabled.
1172 dad_transmits - INTEGER
1173 The amount of Duplicate Address Detection probes to send.
1176 forwarding - INTEGER
1177 Configure interface-specific Host/Router behaviour.
1179 Note: It is recommended to have the same setting on all
1180 interfaces; mixed router/host scenarios are rather uncommon.
1182 Possible values are:
1183 0 Forwarding disabled
1184 1 Forwarding enabled
1188 By default, Host behaviour is assumed. This means:
1190 1. IsRouter flag is not set in Neighbour Advertisements.
1191 2. If accept_ra is TRUE (default), transmit Router
1193 3. If accept_ra is TRUE (default), accept Router
1194 Advertisements (and do autoconfiguration).
1195 4. If accept_redirects is TRUE (default), accept Redirects.
1199 If local forwarding is enabled, Router behaviour is assumed.
1200 This means exactly the reverse from the above:
1202 1. IsRouter flag is set in Neighbour Advertisements.
1203 2. Router Solicitations are not sent unless accept_ra is 2.
1204 3. Router Advertisements are ignored unless accept_ra is 2.
1205 4. Redirects are ignored.
1207 Default: 0 (disabled) if global forwarding is disabled (default),
1208 otherwise 1 (enabled).
1211 Default Hop Limit to set.
1215 Default Maximum Transfer Unit
1216 Default: 1280 (IPv6 required minimum)
1218 router_probe_interval - INTEGER
1219 Minimum interval (in seconds) between Router Probing described
1224 router_solicitation_delay - INTEGER
1225 Number of seconds to wait after interface is brought up
1226 before sending Router Solicitations.
1229 router_solicitation_interval - INTEGER
1230 Number of seconds to wait between Router Solicitations.
1233 router_solicitations - INTEGER
1234 Number of Router Solicitations to send until assuming no
1235 routers are present.
1238 use_tempaddr - INTEGER
1239 Preference for Privacy Extensions (RFC3041).
1240 <= 0 : disable Privacy Extensions
1241 == 1 : enable Privacy Extensions, but prefer public
1242 addresses over temporary addresses.
1243 > 1 : enable Privacy Extensions and prefer temporary
1244 addresses over public addresses.
1245 Default: 0 (for most devices)
1246 -1 (for point-to-point devices and loopback devices)
1248 temp_valid_lft - INTEGER
1249 valid lifetime (in seconds) for temporary addresses.
1250 Default: 604800 (7 days)
1252 temp_prefered_lft - INTEGER
1253 Preferred lifetime (in seconds) for temporary addresses.
1254 Default: 86400 (1 day)
1256 max_desync_factor - INTEGER
1257 Maximum value for DESYNC_FACTOR, which is a random value
1258 that ensures that clients don't synchronize with each
1259 other and generate new addresses at exactly the same time.
1260 value is in seconds.
1263 regen_max_retry - INTEGER
1264 Number of attempts before give up attempting to generate
1265 valid temporary addresses.
1268 max_addresses - INTEGER
1269 Maximum number of autoconfigured addresses per interface. Setting
1270 to zero disables the limitation. It is not recommended to set this
1271 value too large (or to zero) because it would be an easy way to
1272 crash the kernel by allowing too many addresses to be created.
1275 disable_ipv6 - BOOLEAN
1276 Disable IPv6 operation. If accept_dad is set to 2, this value
1277 will be dynamically set to TRUE if DAD fails for the link-local
1279 Default: FALSE (enable IPv6 operation)
1281 When this value is changed from 1 to 0 (IPv6 is being enabled),
1282 it will dynamically create a link-local address on the given
1283 interface and start Duplicate Address Detection, if necessary.
1285 When this value is changed from 0 to 1 (IPv6 is being disabled),
1286 it will dynamically delete all address on the given interface.
1288 accept_dad - INTEGER
1289 Whether to accept DAD (Duplicate Address Detection).
1291 1: Enable DAD (default)
1292 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1293 link-local address has been found.
1295 force_tllao - BOOLEAN
1296 Enable sending the target link-layer address option even when
1297 responding to a unicast neighbor solicitation.
1300 Quoting from RFC 2461, section 4.4, Target link-layer address:
1302 "The option MUST be included for multicast solicitations in order to
1303 avoid infinite Neighbor Solicitation "recursion" when the peer node
1304 does not have a cache entry to return a Neighbor Advertisements
1305 message. When responding to unicast solicitations, the option can be
1306 omitted since the sender of the solicitation has the correct link-
1307 layer address; otherwise it would not have be able to send the unicast
1308 solicitation in the first place. However, including the link-layer
1309 address in this case adds little overhead and eliminates a potential
1310 race condition where the sender deletes the cached link-layer address
1311 prior to receiving a response to a previous solicitation."
1315 Limit the maximal rates for sending ICMPv6 packets.
1316 0 to disable any limiting,
1317 otherwise the minimal space between responses in milliseconds.
1322 Pekka Savola <pekkas@netcore.fi>
1323 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1326 /proc/sys/net/bridge/* Variables:
1328 bridge-nf-call-arptables - BOOLEAN
1329 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1333 bridge-nf-call-iptables - BOOLEAN
1334 1 : pass bridged IPv4 traffic to iptables' chains.
1338 bridge-nf-call-ip6tables - BOOLEAN
1339 1 : pass bridged IPv6 traffic to ip6tables' chains.
1343 bridge-nf-filter-vlan-tagged - BOOLEAN
1344 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1348 bridge-nf-filter-pppoe-tagged - BOOLEAN
1349 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1353 bridge-nf-pass-vlan-input-dev - BOOLEAN
1354 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
1355 interface on the bridge and set the netfilter input device to the vlan.
1356 This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
1357 target work with vlan-on-top-of-bridge interfaces. When no matching
1358 vlan interface is found, or this switch is off, the input device is
1359 set to the bridge interface.
1360 0: disable bridge netfilter vlan interface lookup.
1363 proc/sys/net/sctp/* Variables:
1365 addip_enable - BOOLEAN
1366 Enable or disable extension of Dynamic Address Reconfiguration
1367 (ADD-IP) functionality specified in RFC5061. This extension provides
1368 the ability to dynamically add and remove new addresses for the SCTP
1371 1: Enable extension.
1373 0: Disable extension.
1377 addip_noauth_enable - BOOLEAN
1378 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1379 authentication to protect the operations of adding or removing new
1380 addresses. This requirement is mandated so that unauthorized hosts
1381 would not be able to hijack associations. However, older
1382 implementations may not have implemented this requirement while
1383 allowing the ADD-IP extension. For reasons of interoperability,
1384 we provide this variable to control the enforcement of the
1385 authentication requirement.
1387 1: Allow ADD-IP extension to be used without authentication. This
1388 should only be set in a closed environment for interoperability
1389 with older implementations.
1391 0: Enforce the authentication requirement
1395 auth_enable - BOOLEAN
1396 Enable or disable Authenticated Chunks extension. This extension
1397 provides the ability to send and receive authenticated chunks and is
1398 required for secure operation of Dynamic Address Reconfiguration
1401 1: Enable this extension.
1402 0: Disable this extension.
1406 prsctp_enable - BOOLEAN
1407 Enable or disable the Partial Reliability extension (RFC3758) which
1408 is used to notify peers that a given DATA should no longer be expected.
1416 The limit of the number of new packets that can be initially sent. It
1417 controls how bursty the generated traffic can be.
1421 association_max_retrans - INTEGER
1422 Set the maximum number for retransmissions that an association can
1423 attempt deciding that the remote end is unreachable. If this value
1424 is exceeded, the association is terminated.
1428 max_init_retransmits - INTEGER
1429 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1430 that an association will attempt before declaring the destination
1431 unreachable and terminating.
1435 path_max_retrans - INTEGER
1436 The maximum number of retransmissions that will be attempted on a given
1437 path. Once this threshold is exceeded, the path is considered
1438 unreachable, and new traffic will use a different path when the
1439 association is multihomed.
1443 pf_retrans - INTEGER
1444 The number of retransmissions that will be attempted on a given path
1445 before traffic is redirected to an alternate transport (should one
1446 exist). Note this is distinct from path_max_retrans, as a path that
1447 passes the pf_retrans threshold can still be used. Its only
1448 deprioritized when a transmission path is selected by the stack. This
1449 setting is primarily used to enable fast failover mechanisms without
1450 having to reduce path_max_retrans to a very low value. See:
1451 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
1452 for details. Note also that a value of pf_retrans > path_max_retrans
1453 disables this feature
1457 rto_initial - INTEGER
1458 The initial round trip timeout value in milliseconds that will be used
1459 in calculating round trip times. This is the initial time interval
1460 for retransmissions.
1465 The maximum value (in milliseconds) of the round trip timeout. This
1466 is the largest time interval that can elapse between retransmissions.
1471 The minimum value (in milliseconds) of the round trip timeout. This
1472 is the smallest time interval the can elapse between retransmissions.
1476 hb_interval - INTEGER
1477 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
1478 are sent at the specified interval on idle paths to probe the state of
1479 a given path between 2 associations.
1483 sack_timeout - INTEGER
1484 The amount of time (in milliseconds) that the implementation will wait
1489 valid_cookie_life - INTEGER
1490 The default lifetime of the SCTP cookie (in milliseconds). The cookie
1491 is used during association establishment.
1495 cookie_preserve_enable - BOOLEAN
1496 Enable or disable the ability to extend the lifetime of the SCTP cookie
1497 that is used during the establishment phase of SCTP association
1499 1: Enable cookie lifetime extension.
1504 rcvbuf_policy - INTEGER
1505 Determines if the receive buffer is attributed to the socket or to
1506 association. SCTP supports the capability to create multiple
1507 associations on a single socket. When using this capability, it is
1508 possible that a single stalled association that's buffering a lot
1509 of data may block other associations from delivering their data by
1510 consuming all of the receive buffer space. To work around this,
1511 the rcvbuf_policy could be set to attribute the receiver buffer space
1512 to each association instead of the socket. This prevents the described
1515 1: rcvbuf space is per association
1516 0: recbuf space is per socket
1520 sndbuf_policy - INTEGER
1521 Similar to rcvbuf_policy above, this applies to send buffer space.
1523 1: Send buffer is tracked per association
1524 0: Send buffer is tracked per socket.
1528 sctp_mem - vector of 3 INTEGERs: min, pressure, max
1529 Number of pages allowed for queueing by all SCTP sockets.
1531 min: Below this number of pages SCTP is not bothered about its
1532 memory appetite. When amount of memory allocated by SCTP exceeds
1533 this number, SCTP starts to moderate memory usage.
1535 pressure: This value was introduced to follow format of tcp_mem.
1537 max: Number of pages allowed for queueing by all SCTP sockets.
1539 Default is calculated at boot time from amount of available memory.
1541 sctp_rmem - vector of 3 INTEGERs: min, default, max
1542 Only the first value ("min") is used, "default" and "max" are
1545 min: Minimal size of receive buffer used by SCTP socket.
1546 It is guaranteed to each SCTP socket (but not association) even
1547 under moderate memory pressure.
1551 sctp_wmem - vector of 3 INTEGERs: min, default, max
1552 Currently this tunable has no effect.
1554 addr_scope_policy - INTEGER
1555 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
1557 0 - Disable IPv4 address scoping
1558 1 - Enable IPv4 address scoping
1559 2 - Follow draft but allow IPv4 private addresses
1560 3 - Follow draft but allow IPv4 link local addresses
1565 /proc/sys/net/core/*
1566 Please see: Documentation/sysctl/net.txt for descriptions of these entries.
1569 /proc/sys/net/unix/*
1570 max_dgram_qlen - INTEGER
1571 The maximum length of dgram socket receive queue
1578 /proc/sys/net/irda/*
1579 fast_poll_increase FIXME
1580 warn_noreply_time FIXME
1581 discovery_slots FIXME
1584 discovery_timeout FIXME
1585 lap_keepalive_time FIXME
1586 max_noreply_time FIXME
1587 max_tx_data_size FIXME
1589 min_tx_turn_time FIXME