1 /proc/sys/net/ipv4/* Variables:
7 Forward Packets between interfaces.
9 This variable is special, its change resets all configuration
10 parameters to their default state (RFC1122 for hosts, RFC1812
13 ip_default_ttl - INTEGER
14 Default value of TTL field (Time To Live) for outgoing (but not
15 forwarded) IP packets. Should be between 1 and 255 inclusive.
16 Default: 64 (as recommended by RFC1700)
18 ip_no_pmtu_disc - INTEGER
19 Disable Path MTU Discovery. If enabled in mode 1 and a
20 fragmentation-required ICMP is received, the PMTU to this
21 destination will be set to min_pmtu (see below). You will need
22 to raise min_pmtu to the smallest interface MTU on your system
23 manually if you want to avoid locally generated fragments.
25 In mode 2 incoming Path MTU Discovery messages will be
26 discarded. Outgoing frames are handled the same as in mode 1,
27 implicitly setting IP_PMTUDISC_DONT on every created socket.
29 Mode 3 is a hardend pmtu discover mode. The kernel will only
30 accept fragmentation-needed errors if the underlying protocol
31 can verify them besides a plain socket lookup. Current
32 protocols for which pmtu events will be honored are TCP, SCTP
33 and DCCP as they verify e.g. the sequence number or the
34 association. This mode should not be enabled globally but is
35 only intended to secure e.g. name servers in namespaces where
36 TCP path mtu must still work but path MTU information of other
37 protocols should be discarded. If enabled globally this mode
38 could break other protocols.
44 default 552 - minimum discovered Path MTU
46 ip_forward_use_pmtu - BOOLEAN
47 By default we don't trust protocol path MTUs while forwarding
48 because they could be easily forged and can lead to unwanted
49 fragmentation by the router.
50 You only need to enable this if you have user-space software
51 which tries to discover path mtus by itself and depends on the
52 kernel honoring this information. This is normally not the
59 route/max_size - INTEGER
60 Maximum number of routes allowed in the kernel. Increase
61 this when using large numbers of interfaces and/or routes.
63 neigh/default/gc_thresh1 - INTEGER
64 Minimum number of entries to keep. Garbage collector will not
65 purge entries if there are fewer than this number.
68 neigh/default/gc_thresh2 - INTEGER
69 Threshold when garbage collector becomes more aggressive about
70 purging entries. Entries older than 5 seconds will be cleared
71 when over this number.
74 neigh/default/gc_thresh3 - INTEGER
75 Maximum number of neighbor entries allowed. Increase this
76 when using large numbers of interfaces and when communicating
77 with large numbers of directly-connected peers.
80 neigh/default/unres_qlen_bytes - INTEGER
81 The maximum number of bytes which may be used by packets
82 queued for each unresolved address by other network layers.
84 Setting negative value is meaningless and will return error.
85 Default: 65536 Bytes(64KB)
87 neigh/default/unres_qlen - INTEGER
88 The maximum number of packets which may be queued for each
89 unresolved address by other network layers.
90 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
91 Prior to linux 3.3, the default value is 3 which may cause
92 unexpected packet loss. The current default value is calculated
93 according to default value of unres_qlen_bytes and true size of
98 Time, in seconds, that cached PMTU information is kept.
100 min_adv_mss - INTEGER
101 The advertised MSS depends on the first hop route MTU, but will
102 never be lower than this setting.
106 ipfrag_high_thresh - INTEGER
107 Maximum memory used to reassemble IP fragments. When
108 ipfrag_high_thresh bytes of memory is allocated for this purpose,
109 the fragment handler will toss packets until ipfrag_low_thresh
110 is reached. This also serves as a maximum limit to namespaces
111 different from the initial one.
113 ipfrag_low_thresh - INTEGER
114 Maximum memory used to reassemble IP fragments before the kernel
115 begins to remove incomplete fragment queues to free up resources.
116 The kernel still accepts new fragments for defragmentation.
118 ipfrag_time - INTEGER
119 Time in seconds to keep an IP fragment in memory.
121 ipfrag_max_dist - INTEGER
122 ipfrag_max_dist is a non-negative integer value which defines the
123 maximum "disorder" which is allowed among fragments which share a
124 common IP source address. Note that reordering of packets is
125 not unusual, but if a large number of fragments arrive from a source
126 IP address while a particular fragment queue remains incomplete, it
127 probably indicates that one or more fragments belonging to that queue
128 have been lost. When ipfrag_max_dist is positive, an additional check
129 is done on fragments before they are added to a reassembly queue - if
130 ipfrag_max_dist (or more) fragments have arrived from a particular IP
131 address between additions to any IP fragment queue using that source
132 address, it's presumed that one or more fragments in the queue are
133 lost. The existing fragment queue will be dropped, and a new one
134 started. An ipfrag_max_dist value of zero disables this check.
136 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
137 result in unnecessarily dropping fragment queues when normal
138 reordering of packets occurs, which could lead to poor application
139 performance. Using a very large value, e.g. 50000, increases the
140 likelihood of incorrectly reassembling IP fragments that originate
141 from different IP datagrams, which could result in data corruption.
146 inet_peer_threshold - INTEGER
147 The approximate size of the storage. Starting from this threshold
148 entries will be thrown aggressively. This threshold also determines
149 entries' time-to-live and time intervals between garbage collection
150 passes. More entries, less time-to-live, less GC interval.
152 inet_peer_minttl - INTEGER
153 Minimum time-to-live of entries. Should be enough to cover fragment
154 time-to-live on the reassembling side. This minimum time-to-live is
155 guaranteed if the pool size is less than inet_peer_threshold.
158 inet_peer_maxttl - INTEGER
159 Maximum time-to-live of entries. Unused entries will expire after
160 this period of time if there is no memory pressure on the pool (i.e.
161 when the number of entries in the pool is very small).
167 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
168 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
171 tcp_abort_on_overflow - BOOLEAN
172 If listening service is too slow to accept new connections,
173 reset them. Default state is FALSE. It means that if overflow
174 occurred due to a burst, connection will recover. Enable this
175 option _only_ if you are really sure that listening daemon
176 cannot be tuned to accept connections faster. Enabling this
177 option can harm clients of your server.
179 tcp_adv_win_scale - INTEGER
180 Count buffering overhead as bytes/2^tcp_adv_win_scale
181 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
183 Possible values are [-31, 31], inclusive.
186 tcp_allowed_congestion_control - STRING
187 Show/set the congestion control choices available to non-privileged
188 processes. The list is a subset of those listed in
189 tcp_available_congestion_control.
190 Default is "reno" and the default setting (tcp_congestion_control).
192 tcp_app_win - INTEGER
193 Reserve max(window/2^tcp_app_win, mss) of window for application
194 buffer. Value 0 is special, it means that nothing is reserved.
197 tcp_autocorking - BOOLEAN
198 Enable TCP auto corking :
199 When applications do consecutive small write()/sendmsg() system calls,
200 we try to coalesce these small writes as much as possible, to lower
201 total amount of sent packets. This is done if at least one prior
202 packet for the flow is waiting in Qdisc queues or device transmit
203 queue. Applications can still use TCP_CORK for optimal behavior
204 when they know how/when to uncork their sockets.
207 tcp_available_congestion_control - STRING
208 Shows the available congestion control choices that are registered.
209 More congestion control algorithms may be available as modules,
212 tcp_base_mss - INTEGER
213 The initial value of search_low to be used by the packetization layer
214 Path MTU discovery (MTU probing). If MTU probing is enabled,
215 this is the initial MSS used by the connection.
217 tcp_congestion_control - STRING
218 Set the congestion control algorithm to be used for new
219 connections. The algorithm "reno" is always available, but
220 additional choices may be available based on kernel configuration.
221 Default is set as part of kernel configuration.
222 For passive connections, the listener congestion control choice
224 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
227 Allows TCP to send "duplicate" SACKs.
229 tcp_early_retrans - INTEGER
230 Enable Early Retransmit (ER), per RFC 5827. ER lowers the threshold
231 for triggering fast retransmit when the amount of outstanding data is
232 small and when no previously unsent data can be transmitted (such
233 that limited transmit could be used). Also controls the use of
234 Tail loss probe (TLP) that converts RTOs occurring due to tail
235 losses into fast recovery (draft-dukkipati-tcpm-tcp-loss-probe-01).
239 2 enables ER but delays fast recovery and fast retransmit
240 by a fourth of RTT. This mitigates connection falsely
241 recovers when network has a small degree of reordering
242 (less than 3 packets).
243 3 enables delayed ER and TLP.
248 Control use of Explicit Congestion Notification (ECN) by TCP.
249 ECN is used only when both ends of the TCP connection indicate
250 support for it. This feature is useful in avoiding losses due
251 to congestion by allowing supporting routers to signal
252 congestion before having to drop packets.
254 0 Disable ECN. Neither initiate nor accept ECN.
255 1 Enable ECN when requested by incoming connections and
256 also request ECN on outgoing connection attempts.
257 2 Enable ECN when requested by incoming connections
258 but do not request ECN on outgoing connections.
262 Enable FACK congestion avoidance and fast retransmission.
263 The value is not used, if tcp_sack is not enabled.
265 tcp_fin_timeout - INTEGER
266 The length of time an orphaned (no longer referenced by any
267 application) connection will remain in the FIN_WAIT_2 state
268 before it is aborted at the local end. While a perfectly
269 valid "receive only" state for an un-orphaned connection, an
270 orphaned connection in FIN_WAIT_2 state could otherwise wait
271 forever for the remote to close its end of the connection.
276 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
277 F-RTO is an enhanced recovery algorithm for TCP retransmission
278 timeouts. It is particularly beneficial in networks where the
279 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
280 modification. It does not require any support from the peer.
282 By default it's enabled with a non-zero value. 0 disables F-RTO.
284 tcp_keepalive_time - INTEGER
285 How often TCP sends out keepalive messages when keepalive is enabled.
288 tcp_keepalive_probes - INTEGER
289 How many keepalive probes TCP sends out, until it decides that the
290 connection is broken. Default value: 9.
292 tcp_keepalive_intvl - INTEGER
293 How frequently the probes are send out. Multiplied by
294 tcp_keepalive_probes it is time to kill not responding connection,
295 after probes started. Default value: 75sec i.e. connection
296 will be aborted after ~11 minutes of retries.
298 tcp_low_latency - BOOLEAN
299 If set, the TCP stack makes decisions that prefer lower
300 latency as opposed to higher throughput. By default, this
301 option is not set meaning that higher throughput is preferred.
302 An example of an application where this default should be
303 changed would be a Beowulf compute cluster.
306 tcp_max_orphans - INTEGER
307 Maximal number of TCP sockets not attached to any user file handle,
308 held by system. If this number is exceeded orphaned connections are
309 reset immediately and warning is printed. This limit exists
310 only to prevent simple DoS attacks, you _must_ not rely on this
311 or lower the limit artificially, but rather increase it
312 (probably, after increasing installed memory),
313 if network conditions require more than default value,
314 and tune network services to linger and kill such states
315 more aggressively. Let me to remind again: each orphan eats
316 up to ~64K of unswappable memory.
318 tcp_max_syn_backlog - INTEGER
319 Maximal number of remembered connection requests, which have not
320 received an acknowledgment from connecting client.
321 The minimal value is 128 for low memory machines, and it will
322 increase in proportion to the memory of machine.
323 If server suffers from overload, try increasing this number.
325 tcp_max_tw_buckets - INTEGER
326 Maximal number of timewait sockets held by system simultaneously.
327 If this number is exceeded time-wait socket is immediately destroyed
328 and warning is printed. This limit exists only to prevent
329 simple DoS attacks, you _must_ not lower the limit artificially,
330 but rather increase it (probably, after increasing installed memory),
331 if network conditions require more than default value.
333 tcp_mem - vector of 3 INTEGERs: min, pressure, max
334 min: below this number of pages TCP is not bothered about its
337 pressure: when amount of memory allocated by TCP exceeds this number
338 of pages, TCP moderates its memory consumption and enters memory
339 pressure mode, which is exited when memory consumption falls
342 max: number of pages allowed for queueing by all TCP sockets.
344 Defaults are calculated at boot time from amount of available
347 tcp_moderate_rcvbuf - BOOLEAN
348 If set, TCP performs receive buffer auto-tuning, attempting to
349 automatically size the buffer (no greater than tcp_rmem[2]) to
350 match the size required by the path for full throughput. Enabled by
353 tcp_mtu_probing - INTEGER
354 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
357 1 - Disabled by default, enabled when an ICMP black hole detected
358 2 - Always enabled, use initial MSS of tcp_base_mss.
360 tcp_no_metrics_save - BOOLEAN
361 By default, TCP saves various connection metrics in the route cache
362 when the connection closes, so that connections established in the
363 near future can use these to set initial conditions. Usually, this
364 increases overall performance, but may sometimes cause performance
365 degradation. If set, TCP will not cache metrics on closing
368 tcp_orphan_retries - INTEGER
369 This value influences the timeout of a locally closed TCP connection,
370 when RTO retransmissions remain unacknowledged.
371 See tcp_retries2 for more details.
373 The default value is 8.
374 If your machine is a loaded WEB server,
375 you should think about lowering this value, such sockets
376 may consume significant resources. Cf. tcp_max_orphans.
378 tcp_reordering - INTEGER
379 Maximal reordering of packets in a TCP stream.
382 tcp_retrans_collapse - BOOLEAN
383 Bug-to-bug compatibility with some broken printers.
384 On retransmit try to send bigger packets to work around bugs in
387 tcp_retries1 - INTEGER
388 This value influences the time, after which TCP decides, that
389 something is wrong due to unacknowledged RTO retransmissions,
390 and reports this suspicion to the network layer.
391 See tcp_retries2 for more details.
393 RFC 1122 recommends at least 3 retransmissions, which is the
396 tcp_retries2 - INTEGER
397 This value influences the timeout of an alive TCP connection,
398 when RTO retransmissions remain unacknowledged.
399 Given a value of N, a hypothetical TCP connection following
400 exponential backoff with an initial RTO of TCP_RTO_MIN would
401 retransmit N times before killing the connection at the (N+1)th RTO.
403 The default value of 15 yields a hypothetical timeout of 924.6
404 seconds and is a lower bound for the effective timeout.
405 TCP will effectively time out at the first RTO which exceeds the
406 hypothetical timeout.
408 RFC 1122 recommends at least 100 seconds for the timeout,
409 which corresponds to a value of at least 8.
411 tcp_rfc1337 - BOOLEAN
412 If set, the TCP stack behaves conforming to RFC1337. If unset,
413 we are not conforming to RFC, but prevent TCP TIME_WAIT
417 tcp_rmem - vector of 3 INTEGERs: min, default, max
418 min: Minimal size of receive buffer used by TCP sockets.
419 It is guaranteed to each TCP socket, even under moderate memory
423 default: initial size of receive buffer used by TCP sockets.
424 This value overrides net.core.rmem_default used by other protocols.
425 Default: 87380 bytes. This value results in window of 65535 with
426 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
427 less for default tcp_app_win. See below about these variables.
429 max: maximal size of receive buffer allowed for automatically
430 selected receiver buffers for TCP socket. This value does not override
431 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
432 automatic tuning of that socket's receive buffer size, in which
433 case this value is ignored.
434 Default: between 87380B and 6MB, depending on RAM size.
437 Enable select acknowledgments (SACKS).
439 tcp_slow_start_after_idle - BOOLEAN
440 If set, provide RFC2861 behavior and time out the congestion
441 window after an idle period. An idle period is defined at
442 the current RTO. If unset, the congestion window will not
443 be timed out after an idle period.
447 Use the Host requirements interpretation of the TCP urgent pointer field.
448 Most hosts use the older BSD interpretation, so if you turn this on
449 Linux might not communicate correctly with them.
452 tcp_synack_retries - INTEGER
453 Number of times SYNACKs for a passive TCP connection attempt will
454 be retransmitted. Should not be higher than 255. Default value
455 is 5, which corresponds to 31seconds till the last retransmission
456 with the current initial RTO of 1second. With this the final timeout
457 for a passive TCP connection will happen after 63seconds.
459 tcp_syncookies - BOOLEAN
460 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
461 Send out syncookies when the syn backlog queue of a socket
462 overflows. This is to prevent against the common 'SYN flood attack'
465 Note, that syncookies is fallback facility.
466 It MUST NOT be used to help highly loaded servers to stand
467 against legal connection rate. If you see SYN flood warnings
468 in your logs, but investigation shows that they occur
469 because of overload with legal connections, you should tune
470 another parameters until this warning disappear.
471 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
473 syncookies seriously violate TCP protocol, do not allow
474 to use TCP extensions, can result in serious degradation
475 of some services (f.e. SMTP relaying), visible not by you,
476 but your clients and relays, contacting you. While you see
477 SYN flood warnings in logs not being really flooded, your server
478 is seriously misconfigured.
480 If you want to test which effects syncookies have to your
481 network connections you can set this knob to 2 to enable
482 unconditionally generation of syncookies.
484 tcp_fastopen - INTEGER
485 Enable TCP Fast Open feature (draft-ietf-tcpm-fastopen) to send data
486 in the opening SYN packet. To use this feature, the client application
487 must use sendmsg() or sendto() with MSG_FASTOPEN flag rather than
488 connect() to perform a TCP handshake automatically.
490 The values (bitmap) are
491 1: Enables sending data in the opening SYN on the client w/ MSG_FASTOPEN.
492 2: Enables TCP Fast Open on the server side, i.e., allowing data in
493 a SYN packet to be accepted and passed to the application before
494 3-way hand shake finishes.
495 4: Send data in the opening SYN regardless of cookie availability and
496 without a cookie option.
497 0x100: Accept SYN data w/o validating the cookie.
498 0x200: Accept data-in-SYN w/o any cookie option present.
499 0x400/0x800: Enable Fast Open on all listeners regardless of the
500 TCP_FASTOPEN socket option. The two different flags designate two
501 different ways of setting max_qlen without the TCP_FASTOPEN socket
506 Note that the client & server side Fast Open flags (1 and 2
507 respectively) must be also enabled before the rest of flags can take
510 See include/net/tcp.h and the code for more details.
512 tcp_syn_retries - INTEGER
513 Number of times initial SYNs for an active TCP connection attempt
514 will be retransmitted. Should not be higher than 255. Default value
515 is 6, which corresponds to 63seconds till the last retransmission
516 with the current initial RTO of 1second. With this the final timeout
517 for an active TCP connection attempt will happen after 127seconds.
519 tcp_timestamps - BOOLEAN
520 Enable timestamps as defined in RFC1323.
522 tcp_min_tso_segs - INTEGER
523 Minimal number of segments per TSO frame.
524 Since linux-3.12, TCP does an automatic sizing of TSO frames,
525 depending on flow rate, instead of filling 64Kbytes packets.
526 For specific usages, it's possible to force TCP to build big
527 TSO frames. Note that TCP stack might split too big TSO packets
528 if available window is too small.
531 tcp_tso_win_divisor - INTEGER
532 This allows control over what percentage of the congestion window
533 can be consumed by a single TSO frame.
534 The setting of this parameter is a choice between burstiness and
535 building larger TSO frames.
538 tcp_tw_recycle - BOOLEAN
539 Enable fast recycling TIME-WAIT sockets. Default value is 0.
540 It should not be changed without advice/request of technical
543 tcp_tw_reuse - BOOLEAN
544 Allow to reuse TIME-WAIT sockets for new connections when it is
545 safe from protocol viewpoint. Default value is 0.
546 It should not be changed without advice/request of technical
549 tcp_window_scaling - BOOLEAN
550 Enable window scaling as defined in RFC1323.
552 tcp_wmem - vector of 3 INTEGERs: min, default, max
553 min: Amount of memory reserved for send buffers for TCP sockets.
554 Each TCP socket has rights to use it due to fact of its birth.
557 default: initial size of send buffer used by TCP sockets. This
558 value overrides net.core.wmem_default used by other protocols.
559 It is usually lower than net.core.wmem_default.
562 max: Maximal amount of memory allowed for automatically tuned
563 send buffers for TCP sockets. This value does not override
564 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
565 automatic tuning of that socket's send buffer size, in which case
566 this value is ignored.
567 Default: between 64K and 4MB, depending on RAM size.
569 tcp_notsent_lowat - UNSIGNED INTEGER
570 A TCP socket can control the amount of unsent bytes in its write queue,
571 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
572 reports POLLOUT events if the amount of unsent bytes is below a per
573 socket value, and if the write queue is not full. sendmsg() will
574 also not add new buffers if the limit is hit.
576 This global variable controls the amount of unsent data for
577 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
578 to the global variable has immediate effect.
580 Default: UINT_MAX (0xFFFFFFFF)
582 tcp_workaround_signed_windows - BOOLEAN
583 If set, assume no receipt of a window scaling option means the
584 remote TCP is broken and treats the window as a signed quantity.
585 If unset, assume the remote TCP is not broken even if we do
586 not receive a window scaling option from them.
589 tcp_dma_copybreak - INTEGER
590 Lower limit, in bytes, of the size of socket reads that will be
591 offloaded to a DMA copy engine, if one is present in the system
592 and CONFIG_NET_DMA is enabled.
595 tcp_thin_linear_timeouts - BOOLEAN
596 Enable dynamic triggering of linear timeouts for thin streams.
597 If set, a check is performed upon retransmission by timeout to
598 determine if the stream is thin (less than 4 packets in flight).
599 As long as the stream is found to be thin, up to 6 linear
600 timeouts may be performed before exponential backoff mode is
601 initiated. This improves retransmission latency for
602 non-aggressive thin streams, often found to be time-dependent.
603 For more information on thin streams, see
604 Documentation/networking/tcp-thin.txt
607 tcp_thin_dupack - BOOLEAN
608 Enable dynamic triggering of retransmissions after one dupACK
609 for thin streams. If set, a check is performed upon reception
610 of a dupACK to determine if the stream is thin (less than 4
611 packets in flight). As long as the stream is found to be thin,
612 data is retransmitted on the first received dupACK. This
613 improves retransmission latency for non-aggressive thin
614 streams, often found to be time-dependent.
615 For more information on thin streams, see
616 Documentation/networking/tcp-thin.txt
619 tcp_limit_output_bytes - INTEGER
620 Controls TCP Small Queue limit per tcp socket.
621 TCP bulk sender tends to increase packets in flight until it
622 gets losses notifications. With SNDBUF autotuning, this can
623 result in a large amount of packets queued in qdisc/device
624 on the local machine, hurting latency of other flows, for
625 typical pfifo_fast qdiscs.
626 tcp_limit_output_bytes limits the number of bytes on qdisc
627 or device to reduce artificial RTT/cwnd and reduce bufferbloat.
630 tcp_challenge_ack_limit - INTEGER
631 Limits number of Challenge ACK sent per second, as recommended
632 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
637 udp_mem - vector of 3 INTEGERs: min, pressure, max
638 Number of pages allowed for queueing by all UDP sockets.
640 min: Below this number of pages UDP is not bothered about its
641 memory appetite. When amount of memory allocated by UDP exceeds
642 this number, UDP starts to moderate memory usage.
644 pressure: This value was introduced to follow format of tcp_mem.
646 max: Number of pages allowed for queueing by all UDP sockets.
648 Default is calculated at boot time from amount of available memory.
650 udp_rmem_min - INTEGER
651 Minimal size of receive buffer used by UDP sockets in moderation.
652 Each UDP socket is able to use the size for receiving data, even if
653 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
656 udp_wmem_min - INTEGER
657 Minimal size of send buffer used by UDP sockets in moderation.
658 Each UDP socket is able to use the size for sending data, even if
659 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
664 cipso_cache_enable - BOOLEAN
665 If set, enable additions to and lookups from the CIPSO label mapping
666 cache. If unset, additions are ignored and lookups always result in a
667 miss. However, regardless of the setting the cache is still
668 invalidated when required when means you can safely toggle this on and
669 off and the cache will always be "safe".
672 cipso_cache_bucket_size - INTEGER
673 The CIPSO label cache consists of a fixed size hash table with each
674 hash bucket containing a number of cache entries. This variable limits
675 the number of entries in each hash bucket; the larger the value the
676 more CIPSO label mappings that can be cached. When the number of
677 entries in a given hash bucket reaches this limit adding new entries
678 causes the oldest entry in the bucket to be removed to make room.
681 cipso_rbm_optfmt - BOOLEAN
682 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
683 the CIPSO draft specification (see Documentation/netlabel for details).
684 This means that when set the CIPSO tag will be padded with empty
685 categories in order to make the packet data 32-bit aligned.
688 cipso_rbm_structvalid - BOOLEAN
689 If set, do a very strict check of the CIPSO option when
690 ip_options_compile() is called. If unset, relax the checks done during
691 ip_options_compile(). Either way is "safe" as errors are caught else
692 where in the CIPSO processing code but setting this to 0 (False) should
693 result in less work (i.e. it should be faster) but could cause problems
694 with other implementations that require strict checking.
699 ip_local_port_range - 2 INTEGERS
700 Defines the local port range that is used by TCP and UDP to
701 choose the local port. The first number is the first, the
702 second the last local port number. The default values are
703 32768 and 61000 respectively.
705 ip_local_reserved_ports - list of comma separated ranges
706 Specify the ports which are reserved for known third-party
707 applications. These ports will not be used by automatic port
708 assignments (e.g. when calling connect() or bind() with port
709 number 0). Explicit port allocation behavior is unchanged.
711 The format used for both input and output is a comma separated
712 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
713 10). Writing to the file will clear all previously reserved
714 ports and update the current list with the one given in the
717 Note that ip_local_port_range and ip_local_reserved_ports
718 settings are independent and both are considered by the kernel
719 when determining which ports are available for automatic port
722 You can reserve ports which are not in the current
723 ip_local_port_range, e.g.:
725 $ cat /proc/sys/net/ipv4/ip_local_port_range
727 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
730 although this is redundant. However such a setting is useful
731 if later the port range is changed to a value that will
732 include the reserved ports.
736 ip_nonlocal_bind - BOOLEAN
737 If set, allows processes to bind() to non-local IP addresses,
738 which can be quite useful - but may break some applications.
742 If set non-zero, enables support for dynamic addresses.
743 If set to a non-zero value larger than 1, a kernel log
744 message will be printed when dynamic address rewriting
748 ip_early_demux - BOOLEAN
749 Optimize input packet processing down to one demux for
750 certain kinds of local sockets. Currently we only do this
751 for established TCP sockets.
753 It may add an additional cost for pure routing workloads that
754 reduces overall throughput, in such case you should disable it.
757 icmp_echo_ignore_all - BOOLEAN
758 If set non-zero, then the kernel will ignore all ICMP ECHO
762 icmp_echo_ignore_broadcasts - BOOLEAN
763 If set non-zero, then the kernel will ignore all ICMP ECHO and
764 TIMESTAMP requests sent to it via broadcast/multicast.
767 icmp_ratelimit - INTEGER
768 Limit the maximal rates for sending ICMP packets whose type matches
769 icmp_ratemask (see below) to specific targets.
770 0 to disable any limiting,
771 otherwise the minimal space between responses in milliseconds.
774 icmp_ratemask - INTEGER
775 Mask made of ICMP types for which rates are being limited.
776 Significant bits: IHGFEDCBA9876543210
777 Default mask: 0000001100000011000 (6168)
779 Bit definitions (see include/linux/icmp.h):
781 3 Destination Unreachable *
786 C Parameter Problem *
791 H Address Mask Request
794 * These are rate limited by default (see default mask above)
796 icmp_ignore_bogus_error_responses - BOOLEAN
797 Some routers violate RFC1122 by sending bogus responses to broadcast
798 frames. Such violations are normally logged via a kernel warning.
799 If this is set to TRUE, the kernel will not give such warnings, which
800 will avoid log file clutter.
803 icmp_errors_use_inbound_ifaddr - BOOLEAN
805 If zero, icmp error messages are sent with the primary address of
806 the exiting interface.
808 If non-zero, the message will be sent with the primary address of
809 the interface that received the packet that caused the icmp error.
810 This is the behaviour network many administrators will expect from
811 a router. And it can make debugging complicated network layouts
814 Note that if no primary address exists for the interface selected,
815 then the primary address of the first non-loopback interface that
816 has one will be used regardless of this setting.
820 igmp_max_memberships - INTEGER
821 Change the maximum number of multicast groups we can subscribe to.
824 Theoretical maximum value is bounded by having to send a membership
825 report in a single datagram (i.e. the report can't span multiple
826 datagrams, or risk confusing the switch and leaving groups you don't
829 The number of supported groups 'M' is bounded by the number of group
830 report entries you can fit into a single datagram of 65535 bytes.
832 M = 65536-sizeof (ip header)/(sizeof(Group record))
834 Group records are variable length, with a minimum of 12 bytes.
835 So net.ipv4.igmp_max_memberships should not be set higher than:
837 (65536-24) / 12 = 5459
839 The value 5459 assumes no IP header options, so in practice
840 this number may be lower.
842 conf/interface/* changes special settings per interface (where
843 "interface" is the name of your network interface)
845 conf/all/* is special, changes the settings for all interfaces
847 log_martians - BOOLEAN
848 Log packets with impossible addresses to kernel log.
849 log_martians for the interface will be enabled if at least one of
850 conf/{all,interface}/log_martians is set to TRUE,
851 it will be disabled otherwise
853 accept_redirects - BOOLEAN
854 Accept ICMP redirect messages.
855 accept_redirects for the interface will be enabled if:
856 - both conf/{all,interface}/accept_redirects are TRUE in the case
857 forwarding for the interface is enabled
859 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
860 case forwarding for the interface is disabled
861 accept_redirects for the interface will be disabled otherwise
866 Enable IP forwarding on this interface.
868 mc_forwarding - BOOLEAN
869 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
870 and a multicast routing daemon is required.
871 conf/all/mc_forwarding must also be set to TRUE to enable multicast
872 routing for the interface
875 Integer value used to differentiate the devices by the medium they
876 are attached to. Two devices can have different id values when
877 the broadcast packets are received only on one of them.
878 The default value 0 means that the device is the only interface
879 to its medium, value of -1 means that medium is not known.
881 Currently, it is used to change the proxy_arp behavior:
882 the proxy_arp feature is enabled for packets forwarded between
883 two devices attached to different media.
887 proxy_arp for the interface will be enabled if at least one of
888 conf/{all,interface}/proxy_arp is set to TRUE,
889 it will be disabled otherwise
891 proxy_arp_pvlan - BOOLEAN
892 Private VLAN proxy arp.
893 Basically allow proxy arp replies back to the same interface
894 (from which the ARP request/solicitation was received).
896 This is done to support (ethernet) switch features, like RFC
897 3069, where the individual ports are NOT allowed to
898 communicate with each other, but they are allowed to talk to
899 the upstream router. As described in RFC 3069, it is possible
900 to allow these hosts to communicate through the upstream
901 router by proxy_arp'ing. Don't need to be used together with
904 This technology is known by different names:
905 In RFC 3069 it is called VLAN Aggregation.
906 Cisco and Allied Telesyn call it Private VLAN.
907 Hewlett-Packard call it Source-Port filtering or port-isolation.
908 Ericsson call it MAC-Forced Forwarding (RFC Draft).
910 shared_media - BOOLEAN
911 Send(router) or accept(host) RFC1620 shared media redirects.
912 Overrides ip_secure_redirects.
913 shared_media for the interface will be enabled if at least one of
914 conf/{all,interface}/shared_media is set to TRUE,
915 it will be disabled otherwise
918 secure_redirects - BOOLEAN
919 Accept ICMP redirect messages only for gateways,
920 listed in default gateway list.
921 secure_redirects for the interface will be enabled if at least one of
922 conf/{all,interface}/secure_redirects is set to TRUE,
923 it will be disabled otherwise
926 send_redirects - BOOLEAN
927 Send redirects, if router.
928 send_redirects for the interface will be enabled if at least one of
929 conf/{all,interface}/send_redirects is set to TRUE,
930 it will be disabled otherwise
933 bootp_relay - BOOLEAN
934 Accept packets with source address 0.b.c.d destined
935 not to this host as local ones. It is supposed, that
936 BOOTP relay daemon will catch and forward such packets.
937 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
942 accept_source_route - BOOLEAN
943 Accept packets with SRR option.
944 conf/all/accept_source_route must also be set to TRUE to accept packets
945 with SRR option on the interface
946 default TRUE (router)
949 accept_local - BOOLEAN
950 Accept packets with local source addresses. In combination
951 with suitable routing, this can be used to direct packets
952 between two local interfaces over the wire and have them
955 rp_filter must be set to a non-zero value in order for
956 accept_local to have an effect.
960 route_localnet - BOOLEAN
961 Do not consider loopback addresses as martian source or destination
962 while routing. This enables the use of 127/8 for local routing purposes.
966 0 - No source validation.
967 1 - Strict mode as defined in RFC3704 Strict Reverse Path
968 Each incoming packet is tested against the FIB and if the interface
969 is not the best reverse path the packet check will fail.
970 By default failed packets are discarded.
971 2 - Loose mode as defined in RFC3704 Loose Reverse Path
972 Each incoming packet's source address is also tested against the FIB
973 and if the source address is not reachable via any interface
974 the packet check will fail.
976 Current recommended practice in RFC3704 is to enable strict mode
977 to prevent IP spoofing from DDos attacks. If using asymmetric routing
978 or other complicated routing, then loose mode is recommended.
980 The max value from conf/{all,interface}/rp_filter is used
981 when doing source validation on the {interface}.
983 Default value is 0. Note that some distributions enable it
987 1 - Allows you to have multiple network interfaces on the same
988 subnet, and have the ARPs for each interface be answered
989 based on whether or not the kernel would route a packet from
990 the ARP'd IP out that interface (therefore you must use source
991 based routing for this to work). In other words it allows control
992 of which cards (usually 1) will respond to an arp request.
994 0 - (default) The kernel can respond to arp requests with addresses
995 from other interfaces. This may seem wrong but it usually makes
996 sense, because it increases the chance of successful communication.
997 IP addresses are owned by the complete host on Linux, not by
998 particular interfaces. Only for more complex setups like load-
999 balancing, does this behaviour cause problems.
1001 arp_filter for the interface will be enabled if at least one of
1002 conf/{all,interface}/arp_filter is set to TRUE,
1003 it will be disabled otherwise
1005 arp_announce - INTEGER
1006 Define different restriction levels for announcing the local
1007 source IP address from IP packets in ARP requests sent on
1009 0 - (default) Use any local address, configured on any interface
1010 1 - Try to avoid local addresses that are not in the target's
1011 subnet for this interface. This mode is useful when target
1012 hosts reachable via this interface require the source IP
1013 address in ARP requests to be part of their logical network
1014 configured on the receiving interface. When we generate the
1015 request we will check all our subnets that include the
1016 target IP and will preserve the source address if it is from
1017 such subnet. If there is no such subnet we select source
1018 address according to the rules for level 2.
1019 2 - Always use the best local address for this target.
1020 In this mode we ignore the source address in the IP packet
1021 and try to select local address that we prefer for talks with
1022 the target host. Such local address is selected by looking
1023 for primary IP addresses on all our subnets on the outgoing
1024 interface that include the target IP address. If no suitable
1025 local address is found we select the first local address
1026 we have on the outgoing interface or on all other interfaces,
1027 with the hope we will receive reply for our request and
1028 even sometimes no matter the source IP address we announce.
1030 The max value from conf/{all,interface}/arp_announce is used.
1032 Increasing the restriction level gives more chance for
1033 receiving answer from the resolved target while decreasing
1034 the level announces more valid sender's information.
1036 arp_ignore - INTEGER
1037 Define different modes for sending replies in response to
1038 received ARP requests that resolve local target IP addresses:
1039 0 - (default): reply for any local target IP address, configured
1041 1 - reply only if the target IP address is local address
1042 configured on the incoming interface
1043 2 - reply only if the target IP address is local address
1044 configured on the incoming interface and both with the
1045 sender's IP address are part from same subnet on this interface
1046 3 - do not reply for local addresses configured with scope host,
1047 only resolutions for global and link addresses are replied
1049 8 - do not reply for all local addresses
1051 The max value from conf/{all,interface}/arp_ignore is used
1052 when ARP request is received on the {interface}
1054 arp_notify - BOOLEAN
1055 Define mode for notification of address and device changes.
1056 0 - (default): do nothing
1057 1 - Generate gratuitous arp requests when device is brought up
1058 or hardware address changes.
1060 arp_accept - BOOLEAN
1061 Define behavior for gratuitous ARP frames who's IP is not
1062 already present in the ARP table:
1063 0 - don't create new entries in the ARP table
1064 1 - create new entries in the ARP table
1066 Both replies and requests type gratuitous arp will trigger the
1067 ARP table to be updated, if this setting is on.
1069 If the ARP table already contains the IP address of the
1070 gratuitous arp frame, the arp table will be updated regardless
1071 if this setting is on or off.
1074 app_solicit - INTEGER
1075 The maximum number of probes to send to the user space ARP daemon
1076 via netlink before dropping back to multicast probes (see
1077 mcast_solicit). Defaults to 0.
1079 disable_policy - BOOLEAN
1080 Disable IPSEC policy (SPD) for this interface
1082 disable_xfrm - BOOLEAN
1083 Disable IPSEC encryption on this interface, whatever the policy
1085 igmpv2_unsolicited_report_interval - INTEGER
1086 The interval in milliseconds in which the next unsolicited
1087 IGMPv1 or IGMPv2 report retransmit will take place.
1088 Default: 10000 (10 seconds)
1090 igmpv3_unsolicited_report_interval - INTEGER
1091 The interval in milliseconds in which the next unsolicited
1092 IGMPv3 report retransmit will take place.
1093 Default: 1000 (1 seconds)
1095 promote_secondaries - BOOLEAN
1096 When a primary IP address is removed from this interface
1097 promote a corresponding secondary IP address instead of
1098 removing all the corresponding secondary IP addresses.
1102 Allows you to write a number, which can be used as required.
1106 kuznet@ms2.inr.ac.ru
1112 delon.nicolas@wanadoo.fr
1117 /proc/sys/net/ipv6/* Variables:
1119 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1120 apply to IPv6 [XXX?].
1122 bindv6only - BOOLEAN
1123 Default value for IPV6_V6ONLY socket option,
1124 which restricts use of the IPv6 socket to IPv6 communication
1126 TRUE: disable IPv4-mapped address feature
1127 FALSE: enable IPv4-mapped address feature
1129 Default: FALSE (as specified in RFC3493)
1131 flowlabel_consistency - BOOLEAN
1132 Protect the consistency (and unicity) of flow label.
1133 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1139 auto_flowlabels - BOOLEAN
1140 Automatically generate flow labels based based on a flow hash
1141 of the packet. This allows intermediate devices, such as routers,
1142 to idenfify packet flows for mechanisms like Equal Cost Multipath
1143 Routing (see RFC 6438).
1148 anycast_src_echo_reply - BOOLEAN
1149 Controls the use of anycast addresses as source addresses for ICMPv6
1157 ip6frag_high_thresh - INTEGER
1158 Maximum memory used to reassemble IPv6 fragments. When
1159 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1160 the fragment handler will toss packets until ip6frag_low_thresh
1163 ip6frag_low_thresh - INTEGER
1164 See ip6frag_high_thresh
1166 ip6frag_time - INTEGER
1167 Time in seconds to keep an IPv6 fragment in memory.
1170 Change the interface-specific default settings.
1174 Change all the interface-specific settings.
1176 [XXX: Other special features than forwarding?]
1178 conf/all/forwarding - BOOLEAN
1179 Enable global IPv6 forwarding between all interfaces.
1181 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1182 to control which interfaces may forward packets and which not.
1184 This also sets all interfaces' Host/Router setting
1185 'forwarding' to the specified value. See below for details.
1187 This referred to as global forwarding.
1193 Change special settings per interface.
1195 The functional behaviour for certain settings is different
1196 depending on whether local forwarding is enabled or not.
1199 Accept Router Advertisements; autoconfigure using them.
1201 It also determines whether or not to transmit Router
1202 Solicitations. If and only if the functional setting is to
1203 accept Router Advertisements, Router Solicitations will be
1206 Possible values are:
1207 0 Do not accept Router Advertisements.
1208 1 Accept Router Advertisements if forwarding is disabled.
1209 2 Overrule forwarding behaviour. Accept Router Advertisements
1210 even if forwarding is enabled.
1212 Functional default: enabled if local forwarding is disabled.
1213 disabled if local forwarding is enabled.
1215 accept_ra_defrtr - BOOLEAN
1216 Learn default router in Router Advertisement.
1218 Functional default: enabled if accept_ra is enabled.
1219 disabled if accept_ra is disabled.
1221 accept_ra_from_local - BOOLEAN
1222 Accept RA with source-address that is found on local machine
1223 if the RA is otherwise proper and able to be accepted.
1224 Default is to NOT accept these as it may be an un-intended
1228 enabled if accept_ra_from_local is enabled
1229 on a specific interface.
1230 disabled if accept_ra_from_local is disabled
1231 on a specific interface.
1233 accept_ra_pinfo - BOOLEAN
1234 Learn Prefix Information in Router Advertisement.
1236 Functional default: enabled if accept_ra is enabled.
1237 disabled if accept_ra is disabled.
1239 accept_ra_rt_info_max_plen - INTEGER
1240 Maximum prefix length of Route Information in RA.
1242 Route Information w/ prefix larger than or equal to this
1243 variable shall be ignored.
1245 Functional default: 0 if accept_ra_rtr_pref is enabled.
1246 -1 if accept_ra_rtr_pref is disabled.
1248 accept_ra_rtr_pref - BOOLEAN
1249 Accept Router Preference in RA.
1251 Functional default: enabled if accept_ra is enabled.
1252 disabled if accept_ra is disabled.
1254 accept_redirects - BOOLEAN
1257 Functional default: enabled if local forwarding is disabled.
1258 disabled if local forwarding is enabled.
1260 accept_source_route - INTEGER
1261 Accept source routing (routing extension header).
1263 >= 0: Accept only routing header type 2.
1264 < 0: Do not accept routing header.
1269 Autoconfigure addresses using Prefix Information in Router
1272 Functional default: enabled if accept_ra_pinfo is enabled.
1273 disabled if accept_ra_pinfo is disabled.
1275 dad_transmits - INTEGER
1276 The amount of Duplicate Address Detection probes to send.
1279 forwarding - INTEGER
1280 Configure interface-specific Host/Router behaviour.
1282 Note: It is recommended to have the same setting on all
1283 interfaces; mixed router/host scenarios are rather uncommon.
1285 Possible values are:
1286 0 Forwarding disabled
1287 1 Forwarding enabled
1291 By default, Host behaviour is assumed. This means:
1293 1. IsRouter flag is not set in Neighbour Advertisements.
1294 2. If accept_ra is TRUE (default), transmit Router
1296 3. If accept_ra is TRUE (default), accept Router
1297 Advertisements (and do autoconfiguration).
1298 4. If accept_redirects is TRUE (default), accept Redirects.
1302 If local forwarding is enabled, Router behaviour is assumed.
1303 This means exactly the reverse from the above:
1305 1. IsRouter flag is set in Neighbour Advertisements.
1306 2. Router Solicitations are not sent unless accept_ra is 2.
1307 3. Router Advertisements are ignored unless accept_ra is 2.
1308 4. Redirects are ignored.
1310 Default: 0 (disabled) if global forwarding is disabled (default),
1311 otherwise 1 (enabled).
1314 Default Hop Limit to set.
1318 Default Maximum Transfer Unit
1319 Default: 1280 (IPv6 required minimum)
1321 router_probe_interval - INTEGER
1322 Minimum interval (in seconds) between Router Probing described
1327 router_solicitation_delay - INTEGER
1328 Number of seconds to wait after interface is brought up
1329 before sending Router Solicitations.
1332 router_solicitation_interval - INTEGER
1333 Number of seconds to wait between Router Solicitations.
1336 router_solicitations - INTEGER
1337 Number of Router Solicitations to send until assuming no
1338 routers are present.
1341 use_tempaddr - INTEGER
1342 Preference for Privacy Extensions (RFC3041).
1343 <= 0 : disable Privacy Extensions
1344 == 1 : enable Privacy Extensions, but prefer public
1345 addresses over temporary addresses.
1346 > 1 : enable Privacy Extensions and prefer temporary
1347 addresses over public addresses.
1348 Default: 0 (for most devices)
1349 -1 (for point-to-point devices and loopback devices)
1351 temp_valid_lft - INTEGER
1352 valid lifetime (in seconds) for temporary addresses.
1353 Default: 604800 (7 days)
1355 temp_prefered_lft - INTEGER
1356 Preferred lifetime (in seconds) for temporary addresses.
1357 Default: 86400 (1 day)
1359 max_desync_factor - INTEGER
1360 Maximum value for DESYNC_FACTOR, which is a random value
1361 that ensures that clients don't synchronize with each
1362 other and generate new addresses at exactly the same time.
1363 value is in seconds.
1366 regen_max_retry - INTEGER
1367 Number of attempts before give up attempting to generate
1368 valid temporary addresses.
1371 max_addresses - INTEGER
1372 Maximum number of autoconfigured addresses per interface. Setting
1373 to zero disables the limitation. It is not recommended to set this
1374 value too large (or to zero) because it would be an easy way to
1375 crash the kernel by allowing too many addresses to be created.
1378 disable_ipv6 - BOOLEAN
1379 Disable IPv6 operation. If accept_dad is set to 2, this value
1380 will be dynamically set to TRUE if DAD fails for the link-local
1382 Default: FALSE (enable IPv6 operation)
1384 When this value is changed from 1 to 0 (IPv6 is being enabled),
1385 it will dynamically create a link-local address on the given
1386 interface and start Duplicate Address Detection, if necessary.
1388 When this value is changed from 0 to 1 (IPv6 is being disabled),
1389 it will dynamically delete all address on the given interface.
1391 accept_dad - INTEGER
1392 Whether to accept DAD (Duplicate Address Detection).
1394 1: Enable DAD (default)
1395 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1396 link-local address has been found.
1398 force_tllao - BOOLEAN
1399 Enable sending the target link-layer address option even when
1400 responding to a unicast neighbor solicitation.
1403 Quoting from RFC 2461, section 4.4, Target link-layer address:
1405 "The option MUST be included for multicast solicitations in order to
1406 avoid infinite Neighbor Solicitation "recursion" when the peer node
1407 does not have a cache entry to return a Neighbor Advertisements
1408 message. When responding to unicast solicitations, the option can be
1409 omitted since the sender of the solicitation has the correct link-
1410 layer address; otherwise it would not have be able to send the unicast
1411 solicitation in the first place. However, including the link-layer
1412 address in this case adds little overhead and eliminates a potential
1413 race condition where the sender deletes the cached link-layer address
1414 prior to receiving a response to a previous solicitation."
1416 ndisc_notify - BOOLEAN
1417 Define mode for notification of address and device changes.
1418 0 - (default): do nothing
1419 1 - Generate unsolicited neighbour advertisements when device is brought
1420 up or hardware address changes.
1422 mldv1_unsolicited_report_interval - INTEGER
1423 The interval in milliseconds in which the next unsolicited
1424 MLDv1 report retransmit will take place.
1425 Default: 10000 (10 seconds)
1427 mldv2_unsolicited_report_interval - INTEGER
1428 The interval in milliseconds in which the next unsolicited
1429 MLDv2 report retransmit will take place.
1430 Default: 1000 (1 second)
1432 force_mld_version - INTEGER
1433 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
1434 1 - Enforce to use MLD version 1
1435 2 - Enforce to use MLD version 2
1437 suppress_frag_ndisc - INTEGER
1438 Control RFC 6980 (Security Implications of IPv6 Fragmentation
1439 with IPv6 Neighbor Discovery) behavior:
1440 1 - (default) discard fragmented neighbor discovery packets
1441 0 - allow fragmented neighbor discovery packets
1445 Limit the maximal rates for sending ICMPv6 packets.
1446 0 to disable any limiting,
1447 otherwise the minimal space between responses in milliseconds.
1452 Pekka Savola <pekkas@netcore.fi>
1453 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1456 /proc/sys/net/bridge/* Variables:
1458 bridge-nf-call-arptables - BOOLEAN
1459 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1463 bridge-nf-call-iptables - BOOLEAN
1464 1 : pass bridged IPv4 traffic to iptables' chains.
1468 bridge-nf-call-ip6tables - BOOLEAN
1469 1 : pass bridged IPv6 traffic to ip6tables' chains.
1473 bridge-nf-filter-vlan-tagged - BOOLEAN
1474 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1478 bridge-nf-filter-pppoe-tagged - BOOLEAN
1479 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1483 bridge-nf-pass-vlan-input-dev - BOOLEAN
1484 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
1485 interface on the bridge and set the netfilter input device to the vlan.
1486 This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
1487 target work with vlan-on-top-of-bridge interfaces. When no matching
1488 vlan interface is found, or this switch is off, the input device is
1489 set to the bridge interface.
1490 0: disable bridge netfilter vlan interface lookup.
1493 proc/sys/net/sctp/* Variables:
1495 addip_enable - BOOLEAN
1496 Enable or disable extension of Dynamic Address Reconfiguration
1497 (ADD-IP) functionality specified in RFC5061. This extension provides
1498 the ability to dynamically add and remove new addresses for the SCTP
1501 1: Enable extension.
1503 0: Disable extension.
1507 addip_noauth_enable - BOOLEAN
1508 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1509 authentication to protect the operations of adding or removing new
1510 addresses. This requirement is mandated so that unauthorized hosts
1511 would not be able to hijack associations. However, older
1512 implementations may not have implemented this requirement while
1513 allowing the ADD-IP extension. For reasons of interoperability,
1514 we provide this variable to control the enforcement of the
1515 authentication requirement.
1517 1: Allow ADD-IP extension to be used without authentication. This
1518 should only be set in a closed environment for interoperability
1519 with older implementations.
1521 0: Enforce the authentication requirement
1525 auth_enable - BOOLEAN
1526 Enable or disable Authenticated Chunks extension. This extension
1527 provides the ability to send and receive authenticated chunks and is
1528 required for secure operation of Dynamic Address Reconfiguration
1531 1: Enable this extension.
1532 0: Disable this extension.
1536 prsctp_enable - BOOLEAN
1537 Enable or disable the Partial Reliability extension (RFC3758) which
1538 is used to notify peers that a given DATA should no longer be expected.
1546 The limit of the number of new packets that can be initially sent. It
1547 controls how bursty the generated traffic can be.
1551 association_max_retrans - INTEGER
1552 Set the maximum number for retransmissions that an association can
1553 attempt deciding that the remote end is unreachable. If this value
1554 is exceeded, the association is terminated.
1558 max_init_retransmits - INTEGER
1559 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1560 that an association will attempt before declaring the destination
1561 unreachable and terminating.
1565 path_max_retrans - INTEGER
1566 The maximum number of retransmissions that will be attempted on a given
1567 path. Once this threshold is exceeded, the path is considered
1568 unreachable, and new traffic will use a different path when the
1569 association is multihomed.
1573 pf_retrans - INTEGER
1574 The number of retransmissions that will be attempted on a given path
1575 before traffic is redirected to an alternate transport (should one
1576 exist). Note this is distinct from path_max_retrans, as a path that
1577 passes the pf_retrans threshold can still be used. Its only
1578 deprioritized when a transmission path is selected by the stack. This
1579 setting is primarily used to enable fast failover mechanisms without
1580 having to reduce path_max_retrans to a very low value. See:
1581 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
1582 for details. Note also that a value of pf_retrans > path_max_retrans
1583 disables this feature
1587 rto_initial - INTEGER
1588 The initial round trip timeout value in milliseconds that will be used
1589 in calculating round trip times. This is the initial time interval
1590 for retransmissions.
1595 The maximum value (in milliseconds) of the round trip timeout. This
1596 is the largest time interval that can elapse between retransmissions.
1601 The minimum value (in milliseconds) of the round trip timeout. This
1602 is the smallest time interval the can elapse between retransmissions.
1606 hb_interval - INTEGER
1607 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
1608 are sent at the specified interval on idle paths to probe the state of
1609 a given path between 2 associations.
1613 sack_timeout - INTEGER
1614 The amount of time (in milliseconds) that the implementation will wait
1619 valid_cookie_life - INTEGER
1620 The default lifetime of the SCTP cookie (in milliseconds). The cookie
1621 is used during association establishment.
1625 cookie_preserve_enable - BOOLEAN
1626 Enable or disable the ability to extend the lifetime of the SCTP cookie
1627 that is used during the establishment phase of SCTP association
1629 1: Enable cookie lifetime extension.
1634 cookie_hmac_alg - STRING
1635 Select the hmac algorithm used when generating the cookie value sent by
1636 a listening sctp socket to a connecting client in the INIT-ACK chunk.
1641 Ability to assign md5 or sha1 as the selected alg is predicated on the
1642 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
1643 CONFIG_CRYPTO_SHA1).
1645 Default: Dependent on configuration. MD5 if available, else SHA1 if
1646 available, else none.
1648 rcvbuf_policy - INTEGER
1649 Determines if the receive buffer is attributed to the socket or to
1650 association. SCTP supports the capability to create multiple
1651 associations on a single socket. When using this capability, it is
1652 possible that a single stalled association that's buffering a lot
1653 of data may block other associations from delivering their data by
1654 consuming all of the receive buffer space. To work around this,
1655 the rcvbuf_policy could be set to attribute the receiver buffer space
1656 to each association instead of the socket. This prevents the described
1659 1: rcvbuf space is per association
1660 0: rcvbuf space is per socket
1664 sndbuf_policy - INTEGER
1665 Similar to rcvbuf_policy above, this applies to send buffer space.
1667 1: Send buffer is tracked per association
1668 0: Send buffer is tracked per socket.
1672 sctp_mem - vector of 3 INTEGERs: min, pressure, max
1673 Number of pages allowed for queueing by all SCTP sockets.
1675 min: Below this number of pages SCTP is not bothered about its
1676 memory appetite. When amount of memory allocated by SCTP exceeds
1677 this number, SCTP starts to moderate memory usage.
1679 pressure: This value was introduced to follow format of tcp_mem.
1681 max: Number of pages allowed for queueing by all SCTP sockets.
1683 Default is calculated at boot time from amount of available memory.
1685 sctp_rmem - vector of 3 INTEGERs: min, default, max
1686 Only the first value ("min") is used, "default" and "max" are
1689 min: Minimal size of receive buffer used by SCTP socket.
1690 It is guaranteed to each SCTP socket (but not association) even
1691 under moderate memory pressure.
1695 sctp_wmem - vector of 3 INTEGERs: min, default, max
1696 Currently this tunable has no effect.
1698 addr_scope_policy - INTEGER
1699 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
1701 0 - Disable IPv4 address scoping
1702 1 - Enable IPv4 address scoping
1703 2 - Follow draft but allow IPv4 private addresses
1704 3 - Follow draft but allow IPv4 link local addresses
1709 /proc/sys/net/core/*
1710 Please see: Documentation/sysctl/net.txt for descriptions of these entries.
1713 /proc/sys/net/unix/*
1714 max_dgram_qlen - INTEGER
1715 The maximum length of dgram socket receive queue
1722 /proc/sys/net/irda/*
1723 fast_poll_increase FIXME
1724 warn_noreply_time FIXME
1725 discovery_slots FIXME
1728 discovery_timeout FIXME
1729 lap_keepalive_time FIXME
1730 max_noreply_time FIXME
1731 max_tx_data_size FIXME
1733 min_tx_turn_time FIXME