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 fwmark_reflect - BOOLEAN
26 Controls the fwmark of kernel-generated IPv4 reply packets that are not
27 associated with a socket for example, TCP RSTs or ICMP echo replies).
28 If unset, these packets have a fwmark of zero. If set, they have the
29 fwmark of the packet they are replying to. Similarly affects the fwmark
30 used by internal routing lookups triggered by incoming packets, such as
31 the ones used for Path MTU Discovery.
34 route/max_size - INTEGER
35 Maximum number of routes allowed in the kernel. Increase
36 this when using large numbers of interfaces and/or routes.
38 neigh/default/gc_thresh1 - INTEGER
39 Minimum number of entries to keep. Garbage collector will not
40 purge entries if there are fewer than this number.
43 neigh/default/gc_thresh3 - INTEGER
44 Maximum number of neighbor entries allowed. Increase this
45 when using large numbers of interfaces and when communicating
46 with large numbers of directly-connected peers.
49 neigh/default/unres_qlen_bytes - INTEGER
50 The maximum number of bytes which may be used by packets
51 queued for each unresolved address by other network layers.
53 Setting negative value is meaningless and will return error.
54 Default: 65536 Bytes(64KB)
56 neigh/default/unres_qlen - INTEGER
57 The maximum number of packets which may be queued for each
58 unresolved address by other network layers.
59 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
60 Prior to linux 3.3, the default value is 3 which may cause
61 unexpected packet loss. The current default value is calculated
62 according to default value of unres_qlen_bytes and true size of
67 Time, in seconds, that cached PMTU information is kept.
70 The advertised MSS depends on the first hop route MTU, but will
71 never be lower than this setting.
75 ipfrag_high_thresh - INTEGER
76 Maximum memory used to reassemble IP fragments. When
77 ipfrag_high_thresh bytes of memory is allocated for this purpose,
78 the fragment handler will toss packets until ipfrag_low_thresh
81 ipfrag_low_thresh - INTEGER
82 See ipfrag_high_thresh
85 Time in seconds to keep an IP fragment in memory.
87 ipfrag_secret_interval - INTEGER
88 Regeneration interval (in seconds) of the hash secret (or lifetime
89 for the hash secret) for IP fragments.
92 ipfrag_max_dist - INTEGER
93 ipfrag_max_dist is a non-negative integer value which defines the
94 maximum "disorder" which is allowed among fragments which share a
95 common IP source address. Note that reordering of packets is
96 not unusual, but if a large number of fragments arrive from a source
97 IP address while a particular fragment queue remains incomplete, it
98 probably indicates that one or more fragments belonging to that queue
99 have been lost. When ipfrag_max_dist is positive, an additional check
100 is done on fragments before they are added to a reassembly queue - if
101 ipfrag_max_dist (or more) fragments have arrived from a particular IP
102 address between additions to any IP fragment queue using that source
103 address, it's presumed that one or more fragments in the queue are
104 lost. The existing fragment queue will be dropped, and a new one
105 started. An ipfrag_max_dist value of zero disables this check.
107 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
108 result in unnecessarily dropping fragment queues when normal
109 reordering of packets occurs, which could lead to poor application
110 performance. Using a very large value, e.g. 50000, increases the
111 likelihood of incorrectly reassembling IP fragments that originate
112 from different IP datagrams, which could result in data corruption.
117 inet_peer_threshold - INTEGER
118 The approximate size of the storage. Starting from this threshold
119 entries will be thrown aggressively. This threshold also determines
120 entries' time-to-live and time intervals between garbage collection
121 passes. More entries, less time-to-live, less GC interval.
123 inet_peer_minttl - INTEGER
124 Minimum time-to-live of entries. Should be enough to cover fragment
125 time-to-live on the reassembling side. This minimum time-to-live is
126 guaranteed if the pool size is less than inet_peer_threshold.
129 inet_peer_maxttl - INTEGER
130 Maximum time-to-live of entries. Unused entries will expire after
131 this period of time if there is no memory pressure on the pool (i.e.
132 when the number of entries in the pool is very small).
138 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
139 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
142 tcp_abort_on_overflow - BOOLEAN
143 If listening service is too slow to accept new connections,
144 reset them. Default state is FALSE. It means that if overflow
145 occurred due to a burst, connection will recover. Enable this
146 option _only_ if you are really sure that listening daemon
147 cannot be tuned to accept connections faster. Enabling this
148 option can harm clients of your server.
150 tcp_adv_win_scale - INTEGER
151 Count buffering overhead as bytes/2^tcp_adv_win_scale
152 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
154 Possible values are [-31, 31], inclusive.
157 tcp_allowed_congestion_control - STRING
158 Show/set the congestion control choices available to non-privileged
159 processes. The list is a subset of those listed in
160 tcp_available_congestion_control.
161 Default is "reno" and the default setting (tcp_congestion_control).
163 tcp_app_win - INTEGER
164 Reserve max(window/2^tcp_app_win, mss) of window for application
165 buffer. Value 0 is special, it means that nothing is reserved.
168 tcp_available_congestion_control - STRING
169 Shows the available congestion control choices that are registered.
170 More congestion control algorithms may be available as modules,
173 tcp_base_mss - INTEGER
174 The initial value of search_low to be used by the packetization layer
175 Path MTU discovery (MTU probing). If MTU probing is enabled,
176 this is the initial MSS used by the connection.
178 tcp_congestion_control - STRING
179 Set the congestion control algorithm to be used for new
180 connections. The algorithm "reno" is always available, but
181 additional choices may be available based on kernel configuration.
182 Default is set as part of kernel configuration.
183 For passive connections, the listener congestion control choice
185 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
188 Allows TCP to send "duplicate" SACKs.
190 tcp_early_retrans - INTEGER
191 Enable Early Retransmit (ER), per RFC 5827. ER lowers the threshold
192 for triggering fast retransmit when the amount of outstanding data is
193 small and when no previously unsent data can be transmitted (such
194 that limited transmit could be used). Also controls the use of
195 Tail loss probe (TLP) that converts RTOs occuring due to tail
196 losses into fast recovery (draft-dukkipati-tcpm-tcp-loss-probe-01).
200 2 enables ER but delays fast recovery and fast retransmit
201 by a fourth of RTT. This mitigates connection falsely
202 recovers when network has a small degree of reordering
203 (less than 3 packets).
204 3 enables delayed ER and TLP.
209 Control use of Explicit Congestion Notification (ECN) by TCP.
210 ECN is used only when both ends of the TCP connection indicate
211 support for it. This feature is useful in avoiding losses due
212 to congestion by allowing supporting routers to signal
213 congestion before having to drop packets.
215 0 Disable ECN. Neither initiate nor accept ECN.
216 1 Enable ECN when requested by incoming connections and
217 also request ECN on outgoing connection attempts.
218 2 Enable ECN when requested by incoming connections
219 but do not request ECN on outgoing connections.
223 Enable FACK congestion avoidance and fast retransmission.
224 The value is not used, if tcp_sack is not enabled.
226 tcp_fin_timeout - INTEGER
227 The length of time an orphaned (no longer referenced by any
228 application) connection will remain in the FIN_WAIT_2 state
229 before it is aborted at the local end. While a perfectly
230 valid "receive only" state for an un-orphaned connection, an
231 orphaned connection in FIN_WAIT_2 state could otherwise wait
232 forever for the remote to close its end of the connection.
237 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
238 F-RTO is an enhanced recovery algorithm for TCP retransmission
239 timeouts. It is particularly beneficial in networks where the
240 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
241 modification. It does not require any support from the peer.
243 By default it's enabled with a non-zero value. 0 disables F-RTO.
245 tcp_keepalive_time - INTEGER
246 How often TCP sends out keepalive messages when keepalive is enabled.
249 tcp_keepalive_probes - INTEGER
250 How many keepalive probes TCP sends out, until it decides that the
251 connection is broken. Default value: 9.
253 tcp_keepalive_intvl - INTEGER
254 How frequently the probes are send out. Multiplied by
255 tcp_keepalive_probes it is time to kill not responding connection,
256 after probes started. Default value: 75sec i.e. connection
257 will be aborted after ~11 minutes of retries.
259 tcp_low_latency - BOOLEAN
260 If set, the TCP stack makes decisions that prefer lower
261 latency as opposed to higher throughput. By default, this
262 option is not set meaning that higher throughput is preferred.
263 An example of an application where this default should be
264 changed would be a Beowulf compute cluster.
267 tcp_max_orphans - INTEGER
268 Maximal number of TCP sockets not attached to any user file handle,
269 held by system. If this number is exceeded orphaned connections are
270 reset immediately and warning is printed. This limit exists
271 only to prevent simple DoS attacks, you _must_ not rely on this
272 or lower the limit artificially, but rather increase it
273 (probably, after increasing installed memory),
274 if network conditions require more than default value,
275 and tune network services to linger and kill such states
276 more aggressively. Let me to remind again: each orphan eats
277 up to ~64K of unswappable memory.
279 tcp_max_ssthresh - INTEGER
280 Limited Slow-Start for TCP with large congestion windows (cwnd) defined in
281 RFC3742. Limited slow-start is a mechanism to limit growth of the cwnd
282 on the region where cwnd is larger than tcp_max_ssthresh. TCP increases cwnd
283 by at most tcp_max_ssthresh segments, and by at least tcp_max_ssthresh/2
284 segments per RTT when the cwnd is above tcp_max_ssthresh.
285 If TCP connection increased cwnd to thousands (or tens of thousands) segments,
286 and thousands of packets were being dropped during slow-start, you can set
287 tcp_max_ssthresh to improve performance for new TCP connection.
290 tcp_max_syn_backlog - INTEGER
291 Maximal number of remembered connection requests, which have not
292 received an acknowledgment from connecting client.
293 The minimal value is 128 for low memory machines, and it will
294 increase in proportion to the memory of machine.
295 If server suffers from overload, try increasing this number.
297 tcp_max_tw_buckets - INTEGER
298 Maximal number of timewait sockets held by system simultaneously.
299 If this number is exceeded time-wait socket is immediately destroyed
300 and warning is printed. This limit exists only to prevent
301 simple DoS attacks, you _must_ not lower the limit artificially,
302 but rather increase it (probably, after increasing installed memory),
303 if network conditions require more than default value.
305 tcp_mem - vector of 3 INTEGERs: min, pressure, max
306 min: below this number of pages TCP is not bothered about its
309 pressure: when amount of memory allocated by TCP exceeds this number
310 of pages, TCP moderates its memory consumption and enters memory
311 pressure mode, which is exited when memory consumption falls
314 max: number of pages allowed for queueing by all TCP sockets.
316 Defaults are calculated at boot time from amount of available
319 tcp_moderate_rcvbuf - BOOLEAN
320 If set, TCP performs receive buffer auto-tuning, attempting to
321 automatically size the buffer (no greater than tcp_rmem[2]) to
322 match the size required by the path for full throughput. Enabled by
325 tcp_mtu_probing - INTEGER
326 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
329 1 - Disabled by default, enabled when an ICMP black hole detected
330 2 - Always enabled, use initial MSS of tcp_base_mss.
332 tcp_no_metrics_save - BOOLEAN
333 By default, TCP saves various connection metrics in the route cache
334 when the connection closes, so that connections established in the
335 near future can use these to set initial conditions. Usually, this
336 increases overall performance, but may sometimes cause performance
337 degradation. If set, TCP will not cache metrics on closing
340 tcp_orphan_retries - INTEGER
341 This value influences the timeout of a locally closed TCP connection,
342 when RTO retransmissions remain unacknowledged.
343 See tcp_retries2 for more details.
345 The default value is 8.
346 If your machine is a loaded WEB server,
347 you should think about lowering this value, such sockets
348 may consume significant resources. Cf. tcp_max_orphans.
350 tcp_reordering - INTEGER
351 Maximal reordering of packets in a TCP stream.
354 tcp_retrans_collapse - BOOLEAN
355 Bug-to-bug compatibility with some broken printers.
356 On retransmit try to send bigger packets to work around bugs in
359 tcp_retries1 - INTEGER
360 This value influences the time, after which TCP decides, that
361 something is wrong due to unacknowledged RTO retransmissions,
362 and reports this suspicion to the network layer.
363 See tcp_retries2 for more details.
365 RFC 1122 recommends at least 3 retransmissions, which is the
368 tcp_retries2 - INTEGER
369 This value influences the timeout of an alive TCP connection,
370 when RTO retransmissions remain unacknowledged.
371 Given a value of N, a hypothetical TCP connection following
372 exponential backoff with an initial RTO of TCP_RTO_MIN would
373 retransmit N times before killing the connection at the (N+1)th RTO.
375 The default value of 15 yields a hypothetical timeout of 924.6
376 seconds and is a lower bound for the effective timeout.
377 TCP will effectively time out at the first RTO which exceeds the
378 hypothetical timeout.
380 RFC 1122 recommends at least 100 seconds for the timeout,
381 which corresponds to a value of at least 8.
383 tcp_rfc1337 - BOOLEAN
384 If set, the TCP stack behaves conforming to RFC1337. If unset,
385 we are not conforming to RFC, but prevent TCP TIME_WAIT
389 tcp_rmem - vector of 3 INTEGERs: min, default, max
390 min: Minimal size of receive buffer used by TCP sockets.
391 It is guaranteed to each TCP socket, even under moderate memory
395 default: initial size of receive buffer used by TCP sockets.
396 This value overrides net.core.rmem_default used by other protocols.
397 Default: 87380 bytes. This value results in window of 65535 with
398 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
399 less for default tcp_app_win. See below about these variables.
401 max: maximal size of receive buffer allowed for automatically
402 selected receiver buffers for TCP socket. This value does not override
403 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
404 automatic tuning of that socket's receive buffer size, in which
405 case this value is ignored.
406 Default: between 87380B and 6MB, depending on RAM size.
409 Enable select acknowledgments (SACKS).
411 tcp_slow_start_after_idle - BOOLEAN
412 If set, provide RFC2861 behavior and time out the congestion
413 window after an idle period. An idle period is defined at
414 the current RTO. If unset, the congestion window will not
415 be timed out after an idle period.
419 Use the Host requirements interpretation of the TCP urgent pointer field.
420 Most hosts use the older BSD interpretation, so if you turn this on
421 Linux might not communicate correctly with them.
424 tcp_synack_retries - INTEGER
425 Number of times SYNACKs for a passive TCP connection attempt will
426 be retransmitted. Should not be higher than 255. Default value
427 is 5, which corresponds to 31seconds till the last retransmission
428 with the current initial RTO of 1second. With this the final timeout
429 for a passive TCP connection will happen after 63seconds.
431 tcp_syncookies - BOOLEAN
432 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
433 Send out syncookies when the syn backlog queue of a socket
434 overflows. This is to prevent against the common 'SYN flood attack'
437 Note, that syncookies is fallback facility.
438 It MUST NOT be used to help highly loaded servers to stand
439 against legal connection rate. If you see SYN flood warnings
440 in your logs, but investigation shows that they occur
441 because of overload with legal connections, you should tune
442 another parameters until this warning disappear.
443 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
445 syncookies seriously violate TCP protocol, do not allow
446 to use TCP extensions, can result in serious degradation
447 of some services (f.e. SMTP relaying), visible not by you,
448 but your clients and relays, contacting you. While you see
449 SYN flood warnings in logs not being really flooded, your server
450 is seriously misconfigured.
452 tcp_fastopen - INTEGER
453 Enable TCP Fast Open feature (draft-ietf-tcpm-fastopen) to send data
454 in the opening SYN packet. To use this feature, the client application
455 must use sendmsg() or sendto() with MSG_FASTOPEN flag rather than
456 connect() to perform a TCP handshake automatically.
458 The values (bitmap) are
459 1: Enables sending data in the opening SYN on the client.
460 2: Enables TCP Fast Open on the server side, i.e., allowing data in
461 a SYN packet to be accepted and passed to the application before
462 3-way hand shake finishes.
463 4: Send data in the opening SYN regardless of cookie availability and
464 without a cookie option.
465 0x100: Accept SYN data w/o validating the cookie.
466 0x200: Accept data-in-SYN w/o any cookie option present.
467 0x400/0x800: Enable Fast Open on all listeners regardless of the
468 TCP_FASTOPEN socket option. The two different flags designate two
469 different ways of setting max_qlen without the TCP_FASTOPEN socket
474 Note that the client & server side Fast Open flags (1 and 2
475 respectively) must be also enabled before the rest of flags can take
478 See include/net/tcp.h and the code for more details.
480 tcp_fwmark_accept - BOOLEAN
481 If set, incoming connections to listening sockets that do not have a
482 socket mark will set the mark of the accepting socket to the fwmark of
483 the incoming SYN packet. This will cause all packets on that connection
484 (starting from the first SYNACK) to be sent with that fwmark. The
485 listening socket's mark is unchanged. Listening sockets that already
486 have a fwmark set via setsockopt(SOL_SOCKET, SO_MARK, ...) are
490 tcp_syn_retries - INTEGER
491 Number of times initial SYNs for an active TCP connection attempt
492 will be retransmitted. Should not be higher than 255. Default value
493 is 6, which corresponds to 63seconds till the last retransmission
494 with the current initial RTO of 1second. With this the final timeout
495 for an active TCP connection attempt will happen after 127seconds.
497 tcp_timestamps - BOOLEAN
498 Enable timestamps as defined in RFC1323.
500 tcp_tso_win_divisor - INTEGER
501 This allows control over what percentage of the congestion window
502 can be consumed by a single TSO frame.
503 The setting of this parameter is a choice between burstiness and
504 building larger TSO frames.
507 tcp_tw_recycle - BOOLEAN
508 Enable fast recycling TIME-WAIT sockets. Default value is 0.
509 It should not be changed without advice/request of technical
512 tcp_tw_reuse - BOOLEAN
513 Allow to reuse TIME-WAIT sockets for new connections when it is
514 safe from protocol viewpoint. Default value is 0.
515 It should not be changed without advice/request of technical
518 tcp_window_scaling - BOOLEAN
519 Enable window scaling as defined in RFC1323.
521 tcp_wmem - vector of 3 INTEGERs: min, default, max
522 min: Amount of memory reserved for send buffers for TCP sockets.
523 Each TCP socket has rights to use it due to fact of its birth.
526 default: initial size of send buffer used by TCP sockets. This
527 value overrides net.core.wmem_default used by other protocols.
528 It is usually lower than net.core.wmem_default.
531 max: Maximal amount of memory allowed for automatically tuned
532 send buffers for TCP sockets. This value does not override
533 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
534 automatic tuning of that socket's send buffer size, in which case
535 this value is ignored.
536 Default: between 64K and 4MB, depending on RAM size.
538 tcp_workaround_signed_windows - BOOLEAN
539 If set, assume no receipt of a window scaling option means the
540 remote TCP is broken and treats the window as a signed quantity.
541 If unset, assume the remote TCP is not broken even if we do
542 not receive a window scaling option from them.
545 tcp_dma_copybreak - INTEGER
546 Lower limit, in bytes, of the size of socket reads that will be
547 offloaded to a DMA copy engine, if one is present in the system
548 and CONFIG_NET_DMA is enabled.
551 tcp_thin_linear_timeouts - BOOLEAN
552 Enable dynamic triggering of linear timeouts for thin streams.
553 If set, a check is performed upon retransmission by timeout to
554 determine if the stream is thin (less than 4 packets in flight).
555 As long as the stream is found to be thin, up to 6 linear
556 timeouts may be performed before exponential backoff mode is
557 initiated. This improves retransmission latency for
558 non-aggressive thin streams, often found to be time-dependent.
559 For more information on thin streams, see
560 Documentation/networking/tcp-thin.txt
563 tcp_thin_dupack - BOOLEAN
564 Enable dynamic triggering of retransmissions after one dupACK
565 for thin streams. If set, a check is performed upon reception
566 of a dupACK to determine if the stream is thin (less than 4
567 packets in flight). As long as the stream is found to be thin,
568 data is retransmitted on the first received dupACK. This
569 improves retransmission latency for non-aggressive thin
570 streams, often found to be time-dependent.
571 For more information on thin streams, see
572 Documentation/networking/tcp-thin.txt
575 tcp_limit_output_bytes - INTEGER
576 Controls TCP Small Queue limit per tcp socket.
577 TCP bulk sender tends to increase packets in flight until it
578 gets losses notifications. With SNDBUF autotuning, this can
579 result in a large amount of packets queued in qdisc/device
580 on the local machine, hurting latency of other flows, for
581 typical pfifo_fast qdiscs.
582 tcp_limit_output_bytes limits the number of bytes on qdisc
583 or device to reduce artificial RTT/cwnd and reduce bufferbloat.
584 Note: For GSO/TSO enabled flows, we try to have at least two
585 packets in flight. Reducing tcp_limit_output_bytes might also
586 reduce the size of individual GSO packet (64KB being the max)
589 tcp_challenge_ack_limit - INTEGER
590 Limits number of Challenge ACK sent per second, as recommended
591 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
596 udp_mem - vector of 3 INTEGERs: min, pressure, max
597 Number of pages allowed for queueing by all UDP sockets.
599 min: Below this number of pages UDP is not bothered about its
600 memory appetite. When amount of memory allocated by UDP exceeds
601 this number, UDP starts to moderate memory usage.
603 pressure: This value was introduced to follow format of tcp_mem.
605 max: Number of pages allowed for queueing by all UDP sockets.
607 Default is calculated at boot time from amount of available memory.
609 udp_rmem_min - INTEGER
610 Minimal size of receive buffer used by UDP sockets in moderation.
611 Each UDP socket is able to use the size for receiving data, even if
612 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
615 udp_wmem_min - INTEGER
616 Minimal size of send buffer used by UDP sockets in moderation.
617 Each UDP socket is able to use the size for sending data, even if
618 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
623 cipso_cache_enable - BOOLEAN
624 If set, enable additions to and lookups from the CIPSO label mapping
625 cache. If unset, additions are ignored and lookups always result in a
626 miss. However, regardless of the setting the cache is still
627 invalidated when required when means you can safely toggle this on and
628 off and the cache will always be "safe".
631 cipso_cache_bucket_size - INTEGER
632 The CIPSO label cache consists of a fixed size hash table with each
633 hash bucket containing a number of cache entries. This variable limits
634 the number of entries in each hash bucket; the larger the value the
635 more CIPSO label mappings that can be cached. When the number of
636 entries in a given hash bucket reaches this limit adding new entries
637 causes the oldest entry in the bucket to be removed to make room.
640 cipso_rbm_optfmt - BOOLEAN
641 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
642 the CIPSO draft specification (see Documentation/netlabel for details).
643 This means that when set the CIPSO tag will be padded with empty
644 categories in order to make the packet data 32-bit aligned.
647 cipso_rbm_structvalid - BOOLEAN
648 If set, do a very strict check of the CIPSO option when
649 ip_options_compile() is called. If unset, relax the checks done during
650 ip_options_compile(). Either way is "safe" as errors are caught else
651 where in the CIPSO processing code but setting this to 0 (False) should
652 result in less work (i.e. it should be faster) but could cause problems
653 with other implementations that require strict checking.
658 ip_local_port_range - 2 INTEGERS
659 Defines the local port range that is used by TCP and UDP to
660 choose the local port. The first number is the first, the
661 second the last local port number. The default values are
662 32768 and 61000 respectively.
664 ip_local_reserved_ports - list of comma separated ranges
665 Specify the ports which are reserved for known third-party
666 applications. These ports will not be used by automatic port
667 assignments (e.g. when calling connect() or bind() with port
668 number 0). Explicit port allocation behavior is unchanged.
670 The format used for both input and output is a comma separated
671 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
672 10). Writing to the file will clear all previously reserved
673 ports and update the current list with the one given in the
676 Note that ip_local_port_range and ip_local_reserved_ports
677 settings are independent and both are considered by the kernel
678 when determining which ports are available for automatic port
681 You can reserve ports which are not in the current
682 ip_local_port_range, e.g.:
684 $ cat /proc/sys/net/ipv4/ip_local_port_range
686 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
689 although this is redundant. However such a setting is useful
690 if later the port range is changed to a value that will
691 include the reserved ports.
695 ip_nonlocal_bind - BOOLEAN
696 If set, allows processes to bind() to non-local IP addresses,
697 which can be quite useful - but may break some applications.
701 If set non-zero, enables support for dynamic addresses.
702 If set to a non-zero value larger than 1, a kernel log
703 message will be printed when dynamic address rewriting
707 icmp_echo_ignore_all - BOOLEAN
708 If set non-zero, then the kernel will ignore all ICMP ECHO
712 icmp_echo_ignore_broadcasts - BOOLEAN
713 If set non-zero, then the kernel will ignore all ICMP ECHO and
714 TIMESTAMP requests sent to it via broadcast/multicast.
717 icmp_ratelimit - INTEGER
718 Limit the maximal rates for sending ICMP packets whose type matches
719 icmp_ratemask (see below) to specific targets.
720 0 to disable any limiting,
721 otherwise the minimal space between responses in milliseconds.
724 icmp_ratemask - INTEGER
725 Mask made of ICMP types for which rates are being limited.
726 Significant bits: IHGFEDCBA9876543210
727 Default mask: 0000001100000011000 (6168)
729 Bit definitions (see include/linux/icmp.h):
731 3 Destination Unreachable *
736 C Parameter Problem *
741 H Address Mask Request
744 * These are rate limited by default (see default mask above)
746 icmp_ignore_bogus_error_responses - BOOLEAN
747 Some routers violate RFC1122 by sending bogus responses to broadcast
748 frames. Such violations are normally logged via a kernel warning.
749 If this is set to TRUE, the kernel will not give such warnings, which
750 will avoid log file clutter.
753 icmp_errors_use_inbound_ifaddr - BOOLEAN
755 If zero, icmp error messages are sent with the primary address of
756 the exiting interface.
758 If non-zero, the message will be sent with the primary address of
759 the interface that received the packet that caused the icmp error.
760 This is the behaviour network many administrators will expect from
761 a router. And it can make debugging complicated network layouts
764 Note that if no primary address exists for the interface selected,
765 then the primary address of the first non-loopback interface that
766 has one will be used regardless of this setting.
770 igmp_max_memberships - INTEGER
771 Change the maximum number of multicast groups we can subscribe to.
774 Theoretical maximum value is bounded by having to send a membership
775 report in a single datagram (i.e. the report can't span multiple
776 datagrams, or risk confusing the switch and leaving groups you don't
779 The number of supported groups 'M' is bounded by the number of group
780 report entries you can fit into a single datagram of 65535 bytes.
782 M = 65536-sizeof (ip header)/(sizeof(Group record))
784 Group records are variable length, with a minimum of 12 bytes.
785 So net.ipv4.igmp_max_memberships should not be set higher than:
787 (65536-24) / 12 = 5459
789 The value 5459 assumes no IP header options, so in practice
790 this number may be lower.
792 conf/interface/* changes special settings per interface (where
793 "interface" is the name of your network interface)
795 conf/all/* is special, changes the settings for all interfaces
797 log_martians - BOOLEAN
798 Log packets with impossible addresses to kernel log.
799 log_martians for the interface will be enabled if at least one of
800 conf/{all,interface}/log_martians is set to TRUE,
801 it will be disabled otherwise
803 accept_redirects - BOOLEAN
804 Accept ICMP redirect messages.
805 accept_redirects for the interface will be enabled if:
806 - both conf/{all,interface}/accept_redirects are TRUE in the case
807 forwarding for the interface is enabled
809 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
810 case forwarding for the interface is disabled
811 accept_redirects for the interface will be disabled otherwise
816 Enable IP forwarding on this interface.
818 mc_forwarding - BOOLEAN
819 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
820 and a multicast routing daemon is required.
821 conf/all/mc_forwarding must also be set to TRUE to enable multicast
822 routing for the interface
825 Integer value used to differentiate the devices by the medium they
826 are attached to. Two devices can have different id values when
827 the broadcast packets are received only on one of them.
828 The default value 0 means that the device is the only interface
829 to its medium, value of -1 means that medium is not known.
831 Currently, it is used to change the proxy_arp behavior:
832 the proxy_arp feature is enabled for packets forwarded between
833 two devices attached to different media.
837 proxy_arp for the interface will be enabled if at least one of
838 conf/{all,interface}/proxy_arp is set to TRUE,
839 it will be disabled otherwise
841 proxy_arp_pvlan - BOOLEAN
842 Private VLAN proxy arp.
843 Basically allow proxy arp replies back to the same interface
844 (from which the ARP request/solicitation was received).
846 This is done to support (ethernet) switch features, like RFC
847 3069, where the individual ports are NOT allowed to
848 communicate with each other, but they are allowed to talk to
849 the upstream router. As described in RFC 3069, it is possible
850 to allow these hosts to communicate through the upstream
851 router by proxy_arp'ing. Don't need to be used together with
854 This technology is known by different names:
855 In RFC 3069 it is called VLAN Aggregation.
856 Cisco and Allied Telesyn call it Private VLAN.
857 Hewlett-Packard call it Source-Port filtering or port-isolation.
858 Ericsson call it MAC-Forced Forwarding (RFC Draft).
860 shared_media - BOOLEAN
861 Send(router) or accept(host) RFC1620 shared media redirects.
862 Overrides ip_secure_redirects.
863 shared_media for the interface will be enabled if at least one of
864 conf/{all,interface}/shared_media is set to TRUE,
865 it will be disabled otherwise
868 secure_redirects - BOOLEAN
869 Accept ICMP redirect messages only for gateways,
870 listed in default gateway list.
871 secure_redirects for the interface will be enabled if at least one of
872 conf/{all,interface}/secure_redirects is set to TRUE,
873 it will be disabled otherwise
876 send_redirects - BOOLEAN
877 Send redirects, if router.
878 send_redirects for the interface will be enabled if at least one of
879 conf/{all,interface}/send_redirects is set to TRUE,
880 it will be disabled otherwise
883 bootp_relay - BOOLEAN
884 Accept packets with source address 0.b.c.d destined
885 not to this host as local ones. It is supposed, that
886 BOOTP relay daemon will catch and forward such packets.
887 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
892 accept_source_route - BOOLEAN
893 Accept packets with SRR option.
894 conf/all/accept_source_route must also be set to TRUE to accept packets
895 with SRR option on the interface
896 default TRUE (router)
899 accept_local - BOOLEAN
900 Accept packets with local source addresses. In combination
901 with suitable routing, this can be used to direct packets
902 between two local interfaces over the wire and have them
905 rp_filter must be set to a non-zero value in order for
906 accept_local to have an effect.
910 route_localnet - BOOLEAN
911 Do not consider loopback addresses as martian source or destination
912 while routing. This enables the use of 127/8 for local routing purposes.
916 0 - No source validation.
917 1 - Strict mode as defined in RFC3704 Strict Reverse Path
918 Each incoming packet is tested against the FIB and if the interface
919 is not the best reverse path the packet check will fail.
920 By default failed packets are discarded.
921 2 - Loose mode as defined in RFC3704 Loose Reverse Path
922 Each incoming packet's source address is also tested against the FIB
923 and if the source address is not reachable via any interface
924 the packet check will fail.
926 Current recommended practice in RFC3704 is to enable strict mode
927 to prevent IP spoofing from DDos attacks. If using asymmetric routing
928 or other complicated routing, then loose mode is recommended.
930 The max value from conf/{all,interface}/rp_filter is used
931 when doing source validation on the {interface}.
933 Default value is 0. Note that some distributions enable it
937 1 - Allows you to have multiple network interfaces on the same
938 subnet, and have the ARPs for each interface be answered
939 based on whether or not the kernel would route a packet from
940 the ARP'd IP out that interface (therefore you must use source
941 based routing for this to work). In other words it allows control
942 of which cards (usually 1) will respond to an arp request.
944 0 - (default) The kernel can respond to arp requests with addresses
945 from other interfaces. This may seem wrong but it usually makes
946 sense, because it increases the chance of successful communication.
947 IP addresses are owned by the complete host on Linux, not by
948 particular interfaces. Only for more complex setups like load-
949 balancing, does this behaviour cause problems.
951 arp_filter for the interface will be enabled if at least one of
952 conf/{all,interface}/arp_filter is set to TRUE,
953 it will be disabled otherwise
955 arp_announce - INTEGER
956 Define different restriction levels for announcing the local
957 source IP address from IP packets in ARP requests sent on
959 0 - (default) Use any local address, configured on any interface
960 1 - Try to avoid local addresses that are not in the target's
961 subnet for this interface. This mode is useful when target
962 hosts reachable via this interface require the source IP
963 address in ARP requests to be part of their logical network
964 configured on the receiving interface. When we generate the
965 request we will check all our subnets that include the
966 target IP and will preserve the source address if it is from
967 such subnet. If there is no such subnet we select source
968 address according to the rules for level 2.
969 2 - Always use the best local address for this target.
970 In this mode we ignore the source address in the IP packet
971 and try to select local address that we prefer for talks with
972 the target host. Such local address is selected by looking
973 for primary IP addresses on all our subnets on the outgoing
974 interface that include the target IP address. If no suitable
975 local address is found we select the first local address
976 we have on the outgoing interface or on all other interfaces,
977 with the hope we will receive reply for our request and
978 even sometimes no matter the source IP address we announce.
980 The max value from conf/{all,interface}/arp_announce is used.
982 Increasing the restriction level gives more chance for
983 receiving answer from the resolved target while decreasing
984 the level announces more valid sender's information.
987 Define different modes for sending replies in response to
988 received ARP requests that resolve local target IP addresses:
989 0 - (default): reply for any local target IP address, configured
991 1 - reply only if the target IP address is local address
992 configured on the incoming interface
993 2 - reply only if the target IP address is local address
994 configured on the incoming interface and both with the
995 sender's IP address are part from same subnet on this interface
996 3 - do not reply for local addresses configured with scope host,
997 only resolutions for global and link addresses are replied
999 8 - do not reply for all local addresses
1001 The max value from conf/{all,interface}/arp_ignore is used
1002 when ARP request is received on the {interface}
1004 arp_notify - BOOLEAN
1005 Define mode for notification of address and device changes.
1006 0 - (default): do nothing
1007 1 - Generate gratuitous arp requests when device is brought up
1008 or hardware address changes.
1010 arp_accept - BOOLEAN
1011 Define behavior for gratuitous ARP frames who's IP is not
1012 already present in the ARP table:
1013 0 - don't create new entries in the ARP table
1014 1 - create new entries in the ARP table
1016 Both replies and requests type gratuitous arp will trigger the
1017 ARP table to be updated, if this setting is on.
1019 If the ARP table already contains the IP address of the
1020 gratuitous arp frame, the arp table will be updated regardless
1021 if this setting is on or off.
1024 app_solicit - INTEGER
1025 The maximum number of probes to send to the user space ARP daemon
1026 via netlink before dropping back to multicast probes (see
1027 mcast_solicit). Defaults to 0.
1029 disable_policy - BOOLEAN
1030 Disable IPSEC policy (SPD) for this interface
1032 disable_xfrm - BOOLEAN
1033 Disable IPSEC encryption on this interface, whatever the policy
1038 Allows you to write a number, which can be used as required.
1042 kuznet@ms2.inr.ac.ru
1048 delon.nicolas@wanadoo.fr
1053 /proc/sys/net/ipv6/* Variables:
1055 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1056 apply to IPv6 [XXX?].
1058 bindv6only - BOOLEAN
1059 Default value for IPV6_V6ONLY socket option,
1060 which restricts use of the IPv6 socket to IPv6 communication
1062 TRUE: disable IPv4-mapped address feature
1063 FALSE: enable IPv4-mapped address feature
1065 Default: FALSE (as specified in RFC3493)
1069 ip6frag_high_thresh - INTEGER
1070 Maximum memory used to reassemble IPv6 fragments. When
1071 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1072 the fragment handler will toss packets until ip6frag_low_thresh
1075 ip6frag_low_thresh - INTEGER
1076 See ip6frag_high_thresh
1078 ip6frag_time - INTEGER
1079 Time in seconds to keep an IPv6 fragment in memory.
1081 ip6frag_secret_interval - INTEGER
1082 Regeneration interval (in seconds) of the hash secret (or lifetime
1083 for the hash secret) for IPv6 fragments.
1087 Change the interface-specific default settings.
1091 Change all the interface-specific settings.
1093 [XXX: Other special features than forwarding?]
1095 conf/all/forwarding - BOOLEAN
1096 Enable global IPv6 forwarding between all interfaces.
1098 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1099 to control which interfaces may forward packets and which not.
1101 This also sets all interfaces' Host/Router setting
1102 'forwarding' to the specified value. See below for details.
1104 This referred to as global forwarding.
1109 fwmark_reflect - BOOLEAN
1110 Controls the fwmark of kernel-generated IPv6 reply packets that are not
1111 associated with a socket for example, TCP RSTs or ICMPv6 echo replies).
1112 If unset, these packets have a fwmark of zero. If set, they have the
1113 fwmark of the packet they are replying to. Similarly affects the fwmark
1114 used by internal routing lookups triggered by incoming packets, such as
1115 the ones used for Path MTU Discovery.
1119 Change special settings per interface.
1121 The functional behaviour for certain settings is different
1122 depending on whether local forwarding is enabled or not.
1125 Accept Router Advertisements; autoconfigure using them.
1127 It also determines whether or not to transmit Router
1128 Solicitations. If and only if the functional setting is to
1129 accept Router Advertisements, Router Solicitations will be
1132 Possible values are:
1133 0 Do not accept Router Advertisements.
1134 1 Accept Router Advertisements if forwarding is disabled.
1135 2 Overrule forwarding behaviour. Accept Router Advertisements
1136 even if forwarding is enabled.
1138 Functional default: enabled if local forwarding is disabled.
1139 disabled if local forwarding is enabled.
1141 accept_ra_defrtr - BOOLEAN
1142 Learn default router in Router Advertisement.
1144 Functional default: enabled if accept_ra is enabled.
1145 disabled if accept_ra is disabled.
1147 accept_ra_pinfo - BOOLEAN
1148 Learn Prefix Information in Router Advertisement.
1150 Functional default: enabled if accept_ra is enabled.
1151 disabled if accept_ra is disabled.
1153 accept_ra_rt_info_max_plen - INTEGER
1154 Maximum prefix length of Route Information in RA.
1156 Route Information w/ prefix larger than or equal to this
1157 variable shall be ignored.
1159 Functional default: 0 if accept_ra_rtr_pref is enabled.
1160 -1 if accept_ra_rtr_pref is disabled.
1162 accept_ra_rtr_pref - BOOLEAN
1163 Accept Router Preference in RA.
1165 Functional default: enabled if accept_ra is enabled.
1166 disabled if accept_ra is disabled.
1168 accept_redirects - BOOLEAN
1171 Functional default: enabled if local forwarding is disabled.
1172 disabled if local forwarding is enabled.
1174 accept_source_route - INTEGER
1175 Accept source routing (routing extension header).
1177 >= 0: Accept only routing header type 2.
1178 < 0: Do not accept routing header.
1183 Autoconfigure addresses using Prefix Information in Router
1186 Functional default: enabled if accept_ra_pinfo is enabled.
1187 disabled if accept_ra_pinfo is disabled.
1189 dad_transmits - INTEGER
1190 The amount of Duplicate Address Detection probes to send.
1193 forwarding - INTEGER
1194 Configure interface-specific Host/Router behaviour.
1196 Note: It is recommended to have the same setting on all
1197 interfaces; mixed router/host scenarios are rather uncommon.
1199 Possible values are:
1200 0 Forwarding disabled
1201 1 Forwarding enabled
1205 By default, Host behaviour is assumed. This means:
1207 1. IsRouter flag is not set in Neighbour Advertisements.
1208 2. If accept_ra is TRUE (default), transmit Router
1210 3. If accept_ra is TRUE (default), accept Router
1211 Advertisements (and do autoconfiguration).
1212 4. If accept_redirects is TRUE (default), accept Redirects.
1216 If local forwarding is enabled, Router behaviour is assumed.
1217 This means exactly the reverse from the above:
1219 1. IsRouter flag is set in Neighbour Advertisements.
1220 2. Router Solicitations are not sent unless accept_ra is 2.
1221 3. Router Advertisements are ignored unless accept_ra is 2.
1222 4. Redirects are ignored.
1224 Default: 0 (disabled) if global forwarding is disabled (default),
1225 otherwise 1 (enabled).
1228 Default Hop Limit to set.
1232 Default Maximum Transfer Unit
1233 Default: 1280 (IPv6 required minimum)
1235 router_probe_interval - INTEGER
1236 Minimum interval (in seconds) between Router Probing described
1241 router_solicitation_delay - INTEGER
1242 Number of seconds to wait after interface is brought up
1243 before sending Router Solicitations.
1246 router_solicitation_interval - INTEGER
1247 Number of seconds to wait between Router Solicitations.
1250 router_solicitations - INTEGER
1251 Number of Router Solicitations to send until assuming no
1252 routers are present.
1255 use_tempaddr - INTEGER
1256 Preference for Privacy Extensions (RFC3041).
1257 <= 0 : disable Privacy Extensions
1258 == 1 : enable Privacy Extensions, but prefer public
1259 addresses over temporary addresses.
1260 > 1 : enable Privacy Extensions and prefer temporary
1261 addresses over public addresses.
1262 Default: 0 (for most devices)
1263 -1 (for point-to-point devices and loopback devices)
1265 temp_valid_lft - INTEGER
1266 valid lifetime (in seconds) for temporary addresses.
1267 Default: 604800 (7 days)
1269 temp_prefered_lft - INTEGER
1270 Preferred lifetime (in seconds) for temporary addresses.
1271 Default: 86400 (1 day)
1273 max_desync_factor - INTEGER
1274 Maximum value for DESYNC_FACTOR, which is a random value
1275 that ensures that clients don't synchronize with each
1276 other and generate new addresses at exactly the same time.
1277 value is in seconds.
1280 regen_max_retry - INTEGER
1281 Number of attempts before give up attempting to generate
1282 valid temporary addresses.
1285 max_addresses - INTEGER
1286 Maximum number of autoconfigured addresses per interface. Setting
1287 to zero disables the limitation. It is not recommended to set this
1288 value too large (or to zero) because it would be an easy way to
1289 crash the kernel by allowing too many addresses to be created.
1292 disable_ipv6 - BOOLEAN
1293 Disable IPv6 operation. If accept_dad is set to 2, this value
1294 will be dynamically set to TRUE if DAD fails for the link-local
1296 Default: FALSE (enable IPv6 operation)
1298 When this value is changed from 1 to 0 (IPv6 is being enabled),
1299 it will dynamically create a link-local address on the given
1300 interface and start Duplicate Address Detection, if necessary.
1302 When this value is changed from 0 to 1 (IPv6 is being disabled),
1303 it will dynamically delete all address on the given interface.
1305 accept_dad - INTEGER
1306 Whether to accept DAD (Duplicate Address Detection).
1308 1: Enable DAD (default)
1309 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1310 link-local address has been found.
1312 force_tllao - BOOLEAN
1313 Enable sending the target link-layer address option even when
1314 responding to a unicast neighbor solicitation.
1317 Quoting from RFC 2461, section 4.4, Target link-layer address:
1319 "The option MUST be included for multicast solicitations in order to
1320 avoid infinite Neighbor Solicitation "recursion" when the peer node
1321 does not have a cache entry to return a Neighbor Advertisements
1322 message. When responding to unicast solicitations, the option can be
1323 omitted since the sender of the solicitation has the correct link-
1324 layer address; otherwise it would not have be able to send the unicast
1325 solicitation in the first place. However, including the link-layer
1326 address in this case adds little overhead and eliminates a potential
1327 race condition where the sender deletes the cached link-layer address
1328 prior to receiving a response to a previous solicitation."
1330 ndisc_notify - BOOLEAN
1331 Define mode for notification of address and device changes.
1332 0 - (default): do nothing
1333 1 - Generate unsolicited neighbour advertisements when device is brought
1334 up or hardware address changes.
1338 Limit the maximal rates for sending ICMPv6 packets.
1339 0 to disable any limiting,
1340 otherwise the minimal space between responses in milliseconds.
1345 Pekka Savola <pekkas@netcore.fi>
1346 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1349 /proc/sys/net/bridge/* Variables:
1351 bridge-nf-call-arptables - BOOLEAN
1352 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1356 bridge-nf-call-iptables - BOOLEAN
1357 1 : pass bridged IPv4 traffic to iptables' chains.
1361 bridge-nf-call-ip6tables - BOOLEAN
1362 1 : pass bridged IPv6 traffic to ip6tables' chains.
1366 bridge-nf-filter-vlan-tagged - BOOLEAN
1367 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1371 bridge-nf-filter-pppoe-tagged - BOOLEAN
1372 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1376 bridge-nf-pass-vlan-input-dev - BOOLEAN
1377 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
1378 interface on the bridge and set the netfilter input device to the vlan.
1379 This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
1380 target work with vlan-on-top-of-bridge interfaces. When no matching
1381 vlan interface is found, or this switch is off, the input device is
1382 set to the bridge interface.
1383 0: disable bridge netfilter vlan interface lookup.
1386 proc/sys/net/sctp/* Variables:
1388 addip_enable - BOOLEAN
1389 Enable or disable extension of Dynamic Address Reconfiguration
1390 (ADD-IP) functionality specified in RFC5061. This extension provides
1391 the ability to dynamically add and remove new addresses for the SCTP
1394 1: Enable extension.
1396 0: Disable extension.
1400 addip_noauth_enable - BOOLEAN
1401 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1402 authentication to protect the operations of adding or removing new
1403 addresses. This requirement is mandated so that unauthorized hosts
1404 would not be able to hijack associations. However, older
1405 implementations may not have implemented this requirement while
1406 allowing the ADD-IP extension. For reasons of interoperability,
1407 we provide this variable to control the enforcement of the
1408 authentication requirement.
1410 1: Allow ADD-IP extension to be used without authentication. This
1411 should only be set in a closed environment for interoperability
1412 with older implementations.
1414 0: Enforce the authentication requirement
1418 auth_enable - BOOLEAN
1419 Enable or disable Authenticated Chunks extension. This extension
1420 provides the ability to send and receive authenticated chunks and is
1421 required for secure operation of Dynamic Address Reconfiguration
1424 1: Enable this extension.
1425 0: Disable this extension.
1429 prsctp_enable - BOOLEAN
1430 Enable or disable the Partial Reliability extension (RFC3758) which
1431 is used to notify peers that a given DATA should no longer be expected.
1439 The limit of the number of new packets that can be initially sent. It
1440 controls how bursty the generated traffic can be.
1444 association_max_retrans - INTEGER
1445 Set the maximum number for retransmissions that an association can
1446 attempt deciding that the remote end is unreachable. If this value
1447 is exceeded, the association is terminated.
1451 max_init_retransmits - INTEGER
1452 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1453 that an association will attempt before declaring the destination
1454 unreachable and terminating.
1458 path_max_retrans - INTEGER
1459 The maximum number of retransmissions that will be attempted on a given
1460 path. Once this threshold is exceeded, the path is considered
1461 unreachable, and new traffic will use a different path when the
1462 association is multihomed.
1466 pf_retrans - INTEGER
1467 The number of retransmissions that will be attempted on a given path
1468 before traffic is redirected to an alternate transport (should one
1469 exist). Note this is distinct from path_max_retrans, as a path that
1470 passes the pf_retrans threshold can still be used. Its only
1471 deprioritized when a transmission path is selected by the stack. This
1472 setting is primarily used to enable fast failover mechanisms without
1473 having to reduce path_max_retrans to a very low value. See:
1474 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
1475 for details. Note also that a value of pf_retrans > path_max_retrans
1476 disables this feature
1480 rto_initial - INTEGER
1481 The initial round trip timeout value in milliseconds that will be used
1482 in calculating round trip times. This is the initial time interval
1483 for retransmissions.
1488 The maximum value (in milliseconds) of the round trip timeout. This
1489 is the largest time interval that can elapse between retransmissions.
1494 The minimum value (in milliseconds) of the round trip timeout. This
1495 is the smallest time interval the can elapse between retransmissions.
1499 hb_interval - INTEGER
1500 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
1501 are sent at the specified interval on idle paths to probe the state of
1502 a given path between 2 associations.
1506 sack_timeout - INTEGER
1507 The amount of time (in milliseconds) that the implementation will wait
1512 valid_cookie_life - INTEGER
1513 The default lifetime of the SCTP cookie (in milliseconds). The cookie
1514 is used during association establishment.
1518 cookie_preserve_enable - BOOLEAN
1519 Enable or disable the ability to extend the lifetime of the SCTP cookie
1520 that is used during the establishment phase of SCTP association
1522 1: Enable cookie lifetime extension.
1527 cookie_hmac_alg - STRING
1528 Select the hmac algorithm used when generating the cookie value sent by
1529 a listening sctp socket to a connecting client in the INIT-ACK chunk.
1534 Ability to assign md5 or sha1 as the selected alg is predicated on the
1535 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
1536 CONFIG_CRYPTO_SHA1).
1538 Default: Dependent on configuration. MD5 if available, else SHA1 if
1539 available, else none.
1541 rcvbuf_policy - INTEGER
1542 Determines if the receive buffer is attributed to the socket or to
1543 association. SCTP supports the capability to create multiple
1544 associations on a single socket. When using this capability, it is
1545 possible that a single stalled association that's buffering a lot
1546 of data may block other associations from delivering their data by
1547 consuming all of the receive buffer space. To work around this,
1548 the rcvbuf_policy could be set to attribute the receiver buffer space
1549 to each association instead of the socket. This prevents the described
1552 1: rcvbuf space is per association
1553 0: rcvbuf space is per socket
1557 sndbuf_policy - INTEGER
1558 Similar to rcvbuf_policy above, this applies to send buffer space.
1560 1: Send buffer is tracked per association
1561 0: Send buffer is tracked per socket.
1565 sctp_mem - vector of 3 INTEGERs: min, pressure, max
1566 Number of pages allowed for queueing by all SCTP sockets.
1568 min: Below this number of pages SCTP is not bothered about its
1569 memory appetite. When amount of memory allocated by SCTP exceeds
1570 this number, SCTP starts to moderate memory usage.
1572 pressure: This value was introduced to follow format of tcp_mem.
1574 max: Number of pages allowed for queueing by all SCTP sockets.
1576 Default is calculated at boot time from amount of available memory.
1578 sctp_rmem - vector of 3 INTEGERs: min, default, max
1579 Only the first value ("min") is used, "default" and "max" are
1582 min: Minimal size of receive buffer used by SCTP socket.
1583 It is guaranteed to each SCTP socket (but not association) even
1584 under moderate memory pressure.
1588 sctp_wmem - vector of 3 INTEGERs: min, default, max
1589 Currently this tunable has no effect.
1591 addr_scope_policy - INTEGER
1592 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
1594 0 - Disable IPv4 address scoping
1595 1 - Enable IPv4 address scoping
1596 2 - Follow draft but allow IPv4 private addresses
1597 3 - Follow draft but allow IPv4 link local addresses
1602 /proc/sys/net/core/*
1603 Please see: Documentation/sysctl/net.txt for descriptions of these entries.
1606 /proc/sys/net/unix/*
1607 max_dgram_qlen - INTEGER
1608 The maximum length of dgram socket receive queue
1615 /proc/sys/net/irda/*
1616 fast_poll_increase FIXME
1617 warn_noreply_time FIXME
1618 discovery_slots FIXME
1621 discovery_timeout FIXME
1622 lap_keepalive_time FIXME
1623 max_noreply_time FIXME
1624 max_tx_data_size FIXME
1626 min_tx_turn_time FIXME