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_thresh3 - INTEGER
69 Maximum number of neighbor entries allowed. Increase this
70 when using large numbers of interfaces and when communicating
71 with large numbers of directly-connected peers.
74 neigh/default/unres_qlen_bytes - INTEGER
75 The maximum number of bytes which may be used by packets
76 queued for each unresolved address by other network layers.
78 Setting negative value is meaningless and will return error.
79 Default: 65536 Bytes(64KB)
81 neigh/default/unres_qlen - INTEGER
82 The maximum number of packets which may be queued for each
83 unresolved address by other network layers.
84 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
85 Prior to linux 3.3, the default value is 3 which may cause
86 unexpected packet loss. The current default value is calculated
87 according to default value of unres_qlen_bytes and true size of
92 Time, in seconds, that cached PMTU information is kept.
95 The advertised MSS depends on the first hop route MTU, but will
96 never be lower than this setting.
100 ipfrag_high_thresh - INTEGER
101 Maximum memory used to reassemble IP fragments. When
102 ipfrag_high_thresh bytes of memory is allocated for this purpose,
103 the fragment handler will toss packets until ipfrag_low_thresh
104 is reached. This also serves as a maximum limit to namespaces
105 different from the initial one.
107 ipfrag_low_thresh - INTEGER
108 Maximum memory used to reassemble IP fragments before the kernel
109 begins to remove incomplete fragment queues to free up resources.
110 The kernel still accepts new fragments for defragmentation.
112 ipfrag_time - INTEGER
113 Time in seconds to keep an IP fragment in memory.
115 ipfrag_max_dist - INTEGER
116 ipfrag_max_dist is a non-negative integer value which defines the
117 maximum "disorder" which is allowed among fragments which share a
118 common IP source address. Note that reordering of packets is
119 not unusual, but if a large number of fragments arrive from a source
120 IP address while a particular fragment queue remains incomplete, it
121 probably indicates that one or more fragments belonging to that queue
122 have been lost. When ipfrag_max_dist is positive, an additional check
123 is done on fragments before they are added to a reassembly queue - if
124 ipfrag_max_dist (or more) fragments have arrived from a particular IP
125 address between additions to any IP fragment queue using that source
126 address, it's presumed that one or more fragments in the queue are
127 lost. The existing fragment queue will be dropped, and a new one
128 started. An ipfrag_max_dist value of zero disables this check.
130 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
131 result in unnecessarily dropping fragment queues when normal
132 reordering of packets occurs, which could lead to poor application
133 performance. Using a very large value, e.g. 50000, increases the
134 likelihood of incorrectly reassembling IP fragments that originate
135 from different IP datagrams, which could result in data corruption.
140 inet_peer_threshold - INTEGER
141 The approximate size of the storage. Starting from this threshold
142 entries will be thrown aggressively. This threshold also determines
143 entries' time-to-live and time intervals between garbage collection
144 passes. More entries, less time-to-live, less GC interval.
146 inet_peer_minttl - INTEGER
147 Minimum time-to-live of entries. Should be enough to cover fragment
148 time-to-live on the reassembling side. This minimum time-to-live is
149 guaranteed if the pool size is less than inet_peer_threshold.
152 inet_peer_maxttl - INTEGER
153 Maximum time-to-live of entries. Unused entries will expire after
154 this period of time if there is no memory pressure on the pool (i.e.
155 when the number of entries in the pool is very small).
161 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
162 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
165 tcp_abort_on_overflow - BOOLEAN
166 If listening service is too slow to accept new connections,
167 reset them. Default state is FALSE. It means that if overflow
168 occurred due to a burst, connection will recover. Enable this
169 option _only_ if you are really sure that listening daemon
170 cannot be tuned to accept connections faster. Enabling this
171 option can harm clients of your server.
173 tcp_adv_win_scale - INTEGER
174 Count buffering overhead as bytes/2^tcp_adv_win_scale
175 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
177 Possible values are [-31, 31], inclusive.
180 tcp_allowed_congestion_control - STRING
181 Show/set the congestion control choices available to non-privileged
182 processes. The list is a subset of those listed in
183 tcp_available_congestion_control.
184 Default is "reno" and the default setting (tcp_congestion_control).
186 tcp_app_win - INTEGER
187 Reserve max(window/2^tcp_app_win, mss) of window for application
188 buffer. Value 0 is special, it means that nothing is reserved.
191 tcp_autocorking - BOOLEAN
192 Enable TCP auto corking :
193 When applications do consecutive small write()/sendmsg() system calls,
194 we try to coalesce these small writes as much as possible, to lower
195 total amount of sent packets. This is done if at least one prior
196 packet for the flow is waiting in Qdisc queues or device transmit
197 queue. Applications can still use TCP_CORK for optimal behavior
198 when they know how/when to uncork their sockets.
201 tcp_available_congestion_control - STRING
202 Shows the available congestion control choices that are registered.
203 More congestion control algorithms may be available as modules,
206 tcp_base_mss - INTEGER
207 The initial value of search_low to be used by the packetization layer
208 Path MTU discovery (MTU probing). If MTU probing is enabled,
209 this is the initial MSS used by the connection.
211 tcp_congestion_control - STRING
212 Set the congestion control algorithm to be used for new
213 connections. The algorithm "reno" is always available, but
214 additional choices may be available based on kernel configuration.
215 Default is set as part of kernel configuration.
216 For passive connections, the listener congestion control choice
218 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
221 Allows TCP to send "duplicate" SACKs.
223 tcp_early_retrans - INTEGER
224 Enable Early Retransmit (ER), per RFC 5827. ER lowers the threshold
225 for triggering fast retransmit when the amount of outstanding data is
226 small and when no previously unsent data can be transmitted (such
227 that limited transmit could be used). Also controls the use of
228 Tail loss probe (TLP) that converts RTOs occurring due to tail
229 losses into fast recovery (draft-dukkipati-tcpm-tcp-loss-probe-01).
233 2 enables ER but delays fast recovery and fast retransmit
234 by a fourth of RTT. This mitigates connection falsely
235 recovers when network has a small degree of reordering
236 (less than 3 packets).
237 3 enables delayed ER and TLP.
242 Control use of Explicit Congestion Notification (ECN) by TCP.
243 ECN is used only when both ends of the TCP connection indicate
244 support for it. This feature is useful in avoiding losses due
245 to congestion by allowing supporting routers to signal
246 congestion before having to drop packets.
248 0 Disable ECN. Neither initiate nor accept ECN.
249 1 Enable ECN when requested by incoming connections and
250 also request ECN on outgoing connection attempts.
251 2 Enable ECN when requested by incoming connections
252 but do not request ECN on outgoing connections.
256 Enable FACK congestion avoidance and fast retransmission.
257 The value is not used, if tcp_sack is not enabled.
259 tcp_fin_timeout - INTEGER
260 The length of time an orphaned (no longer referenced by any
261 application) connection will remain in the FIN_WAIT_2 state
262 before it is aborted at the local end. While a perfectly
263 valid "receive only" state for an un-orphaned connection, an
264 orphaned connection in FIN_WAIT_2 state could otherwise wait
265 forever for the remote to close its end of the connection.
270 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
271 F-RTO is an enhanced recovery algorithm for TCP retransmission
272 timeouts. It is particularly beneficial in networks where the
273 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
274 modification. It does not require any support from the peer.
276 By default it's enabled with a non-zero value. 0 disables F-RTO.
278 tcp_keepalive_time - INTEGER
279 How often TCP sends out keepalive messages when keepalive is enabled.
282 tcp_keepalive_probes - INTEGER
283 How many keepalive probes TCP sends out, until it decides that the
284 connection is broken. Default value: 9.
286 tcp_keepalive_intvl - INTEGER
287 How frequently the probes are send out. Multiplied by
288 tcp_keepalive_probes it is time to kill not responding connection,
289 after probes started. Default value: 75sec i.e. connection
290 will be aborted after ~11 minutes of retries.
292 tcp_low_latency - BOOLEAN
293 If set, the TCP stack makes decisions that prefer lower
294 latency as opposed to higher throughput. By default, this
295 option is not set meaning that higher throughput is preferred.
296 An example of an application where this default should be
297 changed would be a Beowulf compute cluster.
300 tcp_max_orphans - INTEGER
301 Maximal number of TCP sockets not attached to any user file handle,
302 held by system. If this number is exceeded orphaned connections are
303 reset immediately and warning is printed. This limit exists
304 only to prevent simple DoS attacks, you _must_ not rely on this
305 or lower the limit artificially, but rather increase it
306 (probably, after increasing installed memory),
307 if network conditions require more than default value,
308 and tune network services to linger and kill such states
309 more aggressively. Let me to remind again: each orphan eats
310 up to ~64K of unswappable memory.
312 tcp_max_syn_backlog - INTEGER
313 Maximal number of remembered connection requests, which have not
314 received an acknowledgment from connecting client.
315 The minimal value is 128 for low memory machines, and it will
316 increase in proportion to the memory of machine.
317 If server suffers from overload, try increasing this number.
319 tcp_max_tw_buckets - INTEGER
320 Maximal number of timewait sockets held by system simultaneously.
321 If this number is exceeded time-wait socket is immediately destroyed
322 and warning is printed. This limit exists only to prevent
323 simple DoS attacks, you _must_ not lower the limit artificially,
324 but rather increase it (probably, after increasing installed memory),
325 if network conditions require more than default value.
327 tcp_mem - vector of 3 INTEGERs: min, pressure, max
328 min: below this number of pages TCP is not bothered about its
331 pressure: when amount of memory allocated by TCP exceeds this number
332 of pages, TCP moderates its memory consumption and enters memory
333 pressure mode, which is exited when memory consumption falls
336 max: number of pages allowed for queueing by all TCP sockets.
338 Defaults are calculated at boot time from amount of available
341 tcp_moderate_rcvbuf - BOOLEAN
342 If set, TCP performs receive buffer auto-tuning, attempting to
343 automatically size the buffer (no greater than tcp_rmem[2]) to
344 match the size required by the path for full throughput. Enabled by
347 tcp_mtu_probing - INTEGER
348 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
351 1 - Disabled by default, enabled when an ICMP black hole detected
352 2 - Always enabled, use initial MSS of tcp_base_mss.
354 tcp_no_metrics_save - BOOLEAN
355 By default, TCP saves various connection metrics in the route cache
356 when the connection closes, so that connections established in the
357 near future can use these to set initial conditions. Usually, this
358 increases overall performance, but may sometimes cause performance
359 degradation. If set, TCP will not cache metrics on closing
362 tcp_orphan_retries - INTEGER
363 This value influences the timeout of a locally closed TCP connection,
364 when RTO retransmissions remain unacknowledged.
365 See tcp_retries2 for more details.
367 The default value is 8.
368 If your machine is a loaded WEB server,
369 you should think about lowering this value, such sockets
370 may consume significant resources. Cf. tcp_max_orphans.
372 tcp_reordering - INTEGER
373 Maximal reordering of packets in a TCP stream.
376 tcp_retrans_collapse - BOOLEAN
377 Bug-to-bug compatibility with some broken printers.
378 On retransmit try to send bigger packets to work around bugs in
381 tcp_retries1 - INTEGER
382 This value influences the time, after which TCP decides, that
383 something is wrong due to unacknowledged RTO retransmissions,
384 and reports this suspicion to the network layer.
385 See tcp_retries2 for more details.
387 RFC 1122 recommends at least 3 retransmissions, which is the
390 tcp_retries2 - INTEGER
391 This value influences the timeout of an alive TCP connection,
392 when RTO retransmissions remain unacknowledged.
393 Given a value of N, a hypothetical TCP connection following
394 exponential backoff with an initial RTO of TCP_RTO_MIN would
395 retransmit N times before killing the connection at the (N+1)th RTO.
397 The default value of 15 yields a hypothetical timeout of 924.6
398 seconds and is a lower bound for the effective timeout.
399 TCP will effectively time out at the first RTO which exceeds the
400 hypothetical timeout.
402 RFC 1122 recommends at least 100 seconds for the timeout,
403 which corresponds to a value of at least 8.
405 tcp_rfc1337 - BOOLEAN
406 If set, the TCP stack behaves conforming to RFC1337. If unset,
407 we are not conforming to RFC, but prevent TCP TIME_WAIT
411 tcp_rmem - vector of 3 INTEGERs: min, default, max
412 min: Minimal size of receive buffer used by TCP sockets.
413 It is guaranteed to each TCP socket, even under moderate memory
417 default: initial size of receive buffer used by TCP sockets.
418 This value overrides net.core.rmem_default used by other protocols.
419 Default: 87380 bytes. This value results in window of 65535 with
420 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
421 less for default tcp_app_win. See below about these variables.
423 max: maximal size of receive buffer allowed for automatically
424 selected receiver buffers for TCP socket. This value does not override
425 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
426 automatic tuning of that socket's receive buffer size, in which
427 case this value is ignored.
428 Default: between 87380B and 6MB, depending on RAM size.
431 Enable select acknowledgments (SACKS).
433 tcp_slow_start_after_idle - BOOLEAN
434 If set, provide RFC2861 behavior and time out the congestion
435 window after an idle period. An idle period is defined at
436 the current RTO. If unset, the congestion window will not
437 be timed out after an idle period.
441 Use the Host requirements interpretation of the TCP urgent pointer field.
442 Most hosts use the older BSD interpretation, so if you turn this on
443 Linux might not communicate correctly with them.
446 tcp_synack_retries - INTEGER
447 Number of times SYNACKs for a passive TCP connection attempt will
448 be retransmitted. Should not be higher than 255. Default value
449 is 5, which corresponds to 31seconds till the last retransmission
450 with the current initial RTO of 1second. With this the final timeout
451 for a passive TCP connection will happen after 63seconds.
453 tcp_syncookies - BOOLEAN
454 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
455 Send out syncookies when the syn backlog queue of a socket
456 overflows. This is to prevent against the common 'SYN flood attack'
459 Note, that syncookies is fallback facility.
460 It MUST NOT be used to help highly loaded servers to stand
461 against legal connection rate. If you see SYN flood warnings
462 in your logs, but investigation shows that they occur
463 because of overload with legal connections, you should tune
464 another parameters until this warning disappear.
465 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
467 syncookies seriously violate TCP protocol, do not allow
468 to use TCP extensions, can result in serious degradation
469 of some services (f.e. SMTP relaying), visible not by you,
470 but your clients and relays, contacting you. While you see
471 SYN flood warnings in logs not being really flooded, your server
472 is seriously misconfigured.
474 If you want to test which effects syncookies have to your
475 network connections you can set this knob to 2 to enable
476 unconditionally generation of syncookies.
478 tcp_fastopen - INTEGER
479 Enable TCP Fast Open feature (draft-ietf-tcpm-fastopen) to send data
480 in the opening SYN packet. To use this feature, the client application
481 must use sendmsg() or sendto() with MSG_FASTOPEN flag rather than
482 connect() to perform a TCP handshake automatically.
484 The values (bitmap) are
485 1: Enables sending data in the opening SYN on the client w/ MSG_FASTOPEN.
486 2: Enables TCP Fast Open on the server side, i.e., allowing data in
487 a SYN packet to be accepted and passed to the application before
488 3-way hand shake finishes.
489 4: Send data in the opening SYN regardless of cookie availability and
490 without a cookie option.
491 0x100: Accept SYN data w/o validating the cookie.
492 0x200: Accept data-in-SYN w/o any cookie option present.
493 0x400/0x800: Enable Fast Open on all listeners regardless of the
494 TCP_FASTOPEN socket option. The two different flags designate two
495 different ways of setting max_qlen without the TCP_FASTOPEN socket
500 Note that the client & server side Fast Open flags (1 and 2
501 respectively) must be also enabled before the rest of flags can take
504 See include/net/tcp.h and the code for more details.
506 tcp_syn_retries - INTEGER
507 Number of times initial SYNs for an active TCP connection attempt
508 will be retransmitted. Should not be higher than 255. Default value
509 is 6, which corresponds to 63seconds till the last retransmission
510 with the current initial RTO of 1second. With this the final timeout
511 for an active TCP connection attempt will happen after 127seconds.
513 tcp_timestamps - BOOLEAN
514 Enable timestamps as defined in RFC1323.
516 tcp_min_tso_segs - INTEGER
517 Minimal number of segments per TSO frame.
518 Since linux-3.12, TCP does an automatic sizing of TSO frames,
519 depending on flow rate, instead of filling 64Kbytes packets.
520 For specific usages, it's possible to force TCP to build big
521 TSO frames. Note that TCP stack might split too big TSO packets
522 if available window is too small.
525 tcp_tso_win_divisor - INTEGER
526 This allows control over what percentage of the congestion window
527 can be consumed by a single TSO frame.
528 The setting of this parameter is a choice between burstiness and
529 building larger TSO frames.
532 tcp_tw_recycle - BOOLEAN
533 Enable fast recycling TIME-WAIT sockets. Default value is 0.
534 It should not be changed without advice/request of technical
537 tcp_tw_reuse - BOOLEAN
538 Allow to reuse TIME-WAIT sockets for new connections when it is
539 safe from protocol viewpoint. Default value is 0.
540 It should not be changed without advice/request of technical
543 tcp_window_scaling - BOOLEAN
544 Enable window scaling as defined in RFC1323.
546 tcp_wmem - vector of 3 INTEGERs: min, default, max
547 min: Amount of memory reserved for send buffers for TCP sockets.
548 Each TCP socket has rights to use it due to fact of its birth.
551 default: initial size of send buffer used by TCP sockets. This
552 value overrides net.core.wmem_default used by other protocols.
553 It is usually lower than net.core.wmem_default.
556 max: Maximal amount of memory allowed for automatically tuned
557 send buffers for TCP sockets. This value does not override
558 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
559 automatic tuning of that socket's send buffer size, in which case
560 this value is ignored.
561 Default: between 64K and 4MB, depending on RAM size.
563 tcp_notsent_lowat - UNSIGNED INTEGER
564 A TCP socket can control the amount of unsent bytes in its write queue,
565 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
566 reports POLLOUT events if the amount of unsent bytes is below a per
567 socket value, and if the write queue is not full. sendmsg() will
568 also not add new buffers if the limit is hit.
570 This global variable controls the amount of unsent data for
571 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
572 to the global variable has immediate effect.
574 Default: UINT_MAX (0xFFFFFFFF)
576 tcp_workaround_signed_windows - BOOLEAN
577 If set, assume no receipt of a window scaling option means the
578 remote TCP is broken and treats the window as a signed quantity.
579 If unset, assume the remote TCP is not broken even if we do
580 not receive a window scaling option from them.
583 tcp_dma_copybreak - INTEGER
584 Lower limit, in bytes, of the size of socket reads that will be
585 offloaded to a DMA copy engine, if one is present in the system
586 and CONFIG_NET_DMA is enabled.
589 tcp_thin_linear_timeouts - BOOLEAN
590 Enable dynamic triggering of linear timeouts for thin streams.
591 If set, a check is performed upon retransmission by timeout to
592 determine if the stream is thin (less than 4 packets in flight).
593 As long as the stream is found to be thin, up to 6 linear
594 timeouts may be performed before exponential backoff mode is
595 initiated. This improves retransmission latency for
596 non-aggressive thin streams, often found to be time-dependent.
597 For more information on thin streams, see
598 Documentation/networking/tcp-thin.txt
601 tcp_thin_dupack - BOOLEAN
602 Enable dynamic triggering of retransmissions after one dupACK
603 for thin streams. If set, a check is performed upon reception
604 of a dupACK to determine if the stream is thin (less than 4
605 packets in flight). As long as the stream is found to be thin,
606 data is retransmitted on the first received dupACK. This
607 improves retransmission latency for non-aggressive thin
608 streams, often found to be time-dependent.
609 For more information on thin streams, see
610 Documentation/networking/tcp-thin.txt
613 tcp_limit_output_bytes - INTEGER
614 Controls TCP Small Queue limit per tcp socket.
615 TCP bulk sender tends to increase packets in flight until it
616 gets losses notifications. With SNDBUF autotuning, this can
617 result in a large amount of packets queued in qdisc/device
618 on the local machine, hurting latency of other flows, for
619 typical pfifo_fast qdiscs.
620 tcp_limit_output_bytes limits the number of bytes on qdisc
621 or device to reduce artificial RTT/cwnd and reduce bufferbloat.
624 tcp_challenge_ack_limit - INTEGER
625 Limits number of Challenge ACK sent per second, as recommended
626 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
631 udp_mem - vector of 3 INTEGERs: min, pressure, max
632 Number of pages allowed for queueing by all UDP sockets.
634 min: Below this number of pages UDP is not bothered about its
635 memory appetite. When amount of memory allocated by UDP exceeds
636 this number, UDP starts to moderate memory usage.
638 pressure: This value was introduced to follow format of tcp_mem.
640 max: Number of pages allowed for queueing by all UDP sockets.
642 Default is calculated at boot time from amount of available memory.
644 udp_rmem_min - INTEGER
645 Minimal size of receive buffer used by UDP sockets in moderation.
646 Each UDP socket is able to use the size for receiving data, even if
647 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
650 udp_wmem_min - INTEGER
651 Minimal size of send buffer used by UDP sockets in moderation.
652 Each UDP socket is able to use the size for sending data, even if
653 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
658 cipso_cache_enable - BOOLEAN
659 If set, enable additions to and lookups from the CIPSO label mapping
660 cache. If unset, additions are ignored and lookups always result in a
661 miss. However, regardless of the setting the cache is still
662 invalidated when required when means you can safely toggle this on and
663 off and the cache will always be "safe".
666 cipso_cache_bucket_size - INTEGER
667 The CIPSO label cache consists of a fixed size hash table with each
668 hash bucket containing a number of cache entries. This variable limits
669 the number of entries in each hash bucket; the larger the value the
670 more CIPSO label mappings that can be cached. When the number of
671 entries in a given hash bucket reaches this limit adding new entries
672 causes the oldest entry in the bucket to be removed to make room.
675 cipso_rbm_optfmt - BOOLEAN
676 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
677 the CIPSO draft specification (see Documentation/netlabel for details).
678 This means that when set the CIPSO tag will be padded with empty
679 categories in order to make the packet data 32-bit aligned.
682 cipso_rbm_structvalid - BOOLEAN
683 If set, do a very strict check of the CIPSO option when
684 ip_options_compile() is called. If unset, relax the checks done during
685 ip_options_compile(). Either way is "safe" as errors are caught else
686 where in the CIPSO processing code but setting this to 0 (False) should
687 result in less work (i.e. it should be faster) but could cause problems
688 with other implementations that require strict checking.
693 ip_local_port_range - 2 INTEGERS
694 Defines the local port range that is used by TCP and UDP to
695 choose the local port. The first number is the first, the
696 second the last local port number. The default values are
697 32768 and 61000 respectively.
699 ip_local_reserved_ports - list of comma separated ranges
700 Specify the ports which are reserved for known third-party
701 applications. These ports will not be used by automatic port
702 assignments (e.g. when calling connect() or bind() with port
703 number 0). Explicit port allocation behavior is unchanged.
705 The format used for both input and output is a comma separated
706 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
707 10). Writing to the file will clear all previously reserved
708 ports and update the current list with the one given in the
711 Note that ip_local_port_range and ip_local_reserved_ports
712 settings are independent and both are considered by the kernel
713 when determining which ports are available for automatic port
716 You can reserve ports which are not in the current
717 ip_local_port_range, e.g.:
719 $ cat /proc/sys/net/ipv4/ip_local_port_range
721 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
724 although this is redundant. However such a setting is useful
725 if later the port range is changed to a value that will
726 include the reserved ports.
730 ip_nonlocal_bind - BOOLEAN
731 If set, allows processes to bind() to non-local IP addresses,
732 which can be quite useful - but may break some applications.
736 If set non-zero, enables support for dynamic addresses.
737 If set to a non-zero value larger than 1, a kernel log
738 message will be printed when dynamic address rewriting
742 ip_early_demux - BOOLEAN
743 Optimize input packet processing down to one demux for
744 certain kinds of local sockets. Currently we only do this
745 for established TCP sockets.
747 It may add an additional cost for pure routing workloads that
748 reduces overall throughput, in such case you should disable it.
751 icmp_echo_ignore_all - BOOLEAN
752 If set non-zero, then the kernel will ignore all ICMP ECHO
756 icmp_echo_ignore_broadcasts - BOOLEAN
757 If set non-zero, then the kernel will ignore all ICMP ECHO and
758 TIMESTAMP requests sent to it via broadcast/multicast.
761 icmp_ratelimit - INTEGER
762 Limit the maximal rates for sending ICMP packets whose type matches
763 icmp_ratemask (see below) to specific targets.
764 0 to disable any limiting,
765 otherwise the minimal space between responses in milliseconds.
768 icmp_ratemask - INTEGER
769 Mask made of ICMP types for which rates are being limited.
770 Significant bits: IHGFEDCBA9876543210
771 Default mask: 0000001100000011000 (6168)
773 Bit definitions (see include/linux/icmp.h):
775 3 Destination Unreachable *
780 C Parameter Problem *
785 H Address Mask Request
788 * These are rate limited by default (see default mask above)
790 icmp_ignore_bogus_error_responses - BOOLEAN
791 Some routers violate RFC1122 by sending bogus responses to broadcast
792 frames. Such violations are normally logged via a kernel warning.
793 If this is set to TRUE, the kernel will not give such warnings, which
794 will avoid log file clutter.
797 icmp_errors_use_inbound_ifaddr - BOOLEAN
799 If zero, icmp error messages are sent with the primary address of
800 the exiting interface.
802 If non-zero, the message will be sent with the primary address of
803 the interface that received the packet that caused the icmp error.
804 This is the behaviour network many administrators will expect from
805 a router. And it can make debugging complicated network layouts
808 Note that if no primary address exists for the interface selected,
809 then the primary address of the first non-loopback interface that
810 has one will be used regardless of this setting.
814 igmp_max_memberships - INTEGER
815 Change the maximum number of multicast groups we can subscribe to.
818 Theoretical maximum value is bounded by having to send a membership
819 report in a single datagram (i.e. the report can't span multiple
820 datagrams, or risk confusing the switch and leaving groups you don't
823 The number of supported groups 'M' is bounded by the number of group
824 report entries you can fit into a single datagram of 65535 bytes.
826 M = 65536-sizeof (ip header)/(sizeof(Group record))
828 Group records are variable length, with a minimum of 12 bytes.
829 So net.ipv4.igmp_max_memberships should not be set higher than:
831 (65536-24) / 12 = 5459
833 The value 5459 assumes no IP header options, so in practice
834 this number may be lower.
836 conf/interface/* changes special settings per interface (where
837 "interface" is the name of your network interface)
839 conf/all/* is special, changes the settings for all interfaces
841 log_martians - BOOLEAN
842 Log packets with impossible addresses to kernel log.
843 log_martians for the interface will be enabled if at least one of
844 conf/{all,interface}/log_martians is set to TRUE,
845 it will be disabled otherwise
847 accept_redirects - BOOLEAN
848 Accept ICMP redirect messages.
849 accept_redirects for the interface will be enabled if:
850 - both conf/{all,interface}/accept_redirects are TRUE in the case
851 forwarding for the interface is enabled
853 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
854 case forwarding for the interface is disabled
855 accept_redirects for the interface will be disabled otherwise
860 Enable IP forwarding on this interface.
862 mc_forwarding - BOOLEAN
863 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
864 and a multicast routing daemon is required.
865 conf/all/mc_forwarding must also be set to TRUE to enable multicast
866 routing for the interface
869 Integer value used to differentiate the devices by the medium they
870 are attached to. Two devices can have different id values when
871 the broadcast packets are received only on one of them.
872 The default value 0 means that the device is the only interface
873 to its medium, value of -1 means that medium is not known.
875 Currently, it is used to change the proxy_arp behavior:
876 the proxy_arp feature is enabled for packets forwarded between
877 two devices attached to different media.
881 proxy_arp for the interface will be enabled if at least one of
882 conf/{all,interface}/proxy_arp is set to TRUE,
883 it will be disabled otherwise
885 proxy_arp_pvlan - BOOLEAN
886 Private VLAN proxy arp.
887 Basically allow proxy arp replies back to the same interface
888 (from which the ARP request/solicitation was received).
890 This is done to support (ethernet) switch features, like RFC
891 3069, where the individual ports are NOT allowed to
892 communicate with each other, but they are allowed to talk to
893 the upstream router. As described in RFC 3069, it is possible
894 to allow these hosts to communicate through the upstream
895 router by proxy_arp'ing. Don't need to be used together with
898 This technology is known by different names:
899 In RFC 3069 it is called VLAN Aggregation.
900 Cisco and Allied Telesyn call it Private VLAN.
901 Hewlett-Packard call it Source-Port filtering or port-isolation.
902 Ericsson call it MAC-Forced Forwarding (RFC Draft).
904 shared_media - BOOLEAN
905 Send(router) or accept(host) RFC1620 shared media redirects.
906 Overrides ip_secure_redirects.
907 shared_media for the interface will be enabled if at least one of
908 conf/{all,interface}/shared_media is set to TRUE,
909 it will be disabled otherwise
912 secure_redirects - BOOLEAN
913 Accept ICMP redirect messages only for gateways,
914 listed in default gateway list.
915 secure_redirects for the interface will be enabled if at least one of
916 conf/{all,interface}/secure_redirects is set to TRUE,
917 it will be disabled otherwise
920 send_redirects - BOOLEAN
921 Send redirects, if router.
922 send_redirects for the interface will be enabled if at least one of
923 conf/{all,interface}/send_redirects is set to TRUE,
924 it will be disabled otherwise
927 bootp_relay - BOOLEAN
928 Accept packets with source address 0.b.c.d destined
929 not to this host as local ones. It is supposed, that
930 BOOTP relay daemon will catch and forward such packets.
931 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
936 accept_source_route - BOOLEAN
937 Accept packets with SRR option.
938 conf/all/accept_source_route must also be set to TRUE to accept packets
939 with SRR option on the interface
940 default TRUE (router)
943 accept_local - BOOLEAN
944 Accept packets with local source addresses. In combination
945 with suitable routing, this can be used to direct packets
946 between two local interfaces over the wire and have them
949 rp_filter must be set to a non-zero value in order for
950 accept_local to have an effect.
954 route_localnet - BOOLEAN
955 Do not consider loopback addresses as martian source or destination
956 while routing. This enables the use of 127/8 for local routing purposes.
960 0 - No source validation.
961 1 - Strict mode as defined in RFC3704 Strict Reverse Path
962 Each incoming packet is tested against the FIB and if the interface
963 is not the best reverse path the packet check will fail.
964 By default failed packets are discarded.
965 2 - Loose mode as defined in RFC3704 Loose Reverse Path
966 Each incoming packet's source address is also tested against the FIB
967 and if the source address is not reachable via any interface
968 the packet check will fail.
970 Current recommended practice in RFC3704 is to enable strict mode
971 to prevent IP spoofing from DDos attacks. If using asymmetric routing
972 or other complicated routing, then loose mode is recommended.
974 The max value from conf/{all,interface}/rp_filter is used
975 when doing source validation on the {interface}.
977 Default value is 0. Note that some distributions enable it
981 1 - Allows you to have multiple network interfaces on the same
982 subnet, and have the ARPs for each interface be answered
983 based on whether or not the kernel would route a packet from
984 the ARP'd IP out that interface (therefore you must use source
985 based routing for this to work). In other words it allows control
986 of which cards (usually 1) will respond to an arp request.
988 0 - (default) The kernel can respond to arp requests with addresses
989 from other interfaces. This may seem wrong but it usually makes
990 sense, because it increases the chance of successful communication.
991 IP addresses are owned by the complete host on Linux, not by
992 particular interfaces. Only for more complex setups like load-
993 balancing, does this behaviour cause problems.
995 arp_filter for the interface will be enabled if at least one of
996 conf/{all,interface}/arp_filter is set to TRUE,
997 it will be disabled otherwise
999 arp_announce - INTEGER
1000 Define different restriction levels for announcing the local
1001 source IP address from IP packets in ARP requests sent on
1003 0 - (default) Use any local address, configured on any interface
1004 1 - Try to avoid local addresses that are not in the target's
1005 subnet for this interface. This mode is useful when target
1006 hosts reachable via this interface require the source IP
1007 address in ARP requests to be part of their logical network
1008 configured on the receiving interface. When we generate the
1009 request we will check all our subnets that include the
1010 target IP and will preserve the source address if it is from
1011 such subnet. If there is no such subnet we select source
1012 address according to the rules for level 2.
1013 2 - Always use the best local address for this target.
1014 In this mode we ignore the source address in the IP packet
1015 and try to select local address that we prefer for talks with
1016 the target host. Such local address is selected by looking
1017 for primary IP addresses on all our subnets on the outgoing
1018 interface that include the target IP address. If no suitable
1019 local address is found we select the first local address
1020 we have on the outgoing interface or on all other interfaces,
1021 with the hope we will receive reply for our request and
1022 even sometimes no matter the source IP address we announce.
1024 The max value from conf/{all,interface}/arp_announce is used.
1026 Increasing the restriction level gives more chance for
1027 receiving answer from the resolved target while decreasing
1028 the level announces more valid sender's information.
1030 arp_ignore - INTEGER
1031 Define different modes for sending replies in response to
1032 received ARP requests that resolve local target IP addresses:
1033 0 - (default): reply for any local target IP address, configured
1035 1 - reply only if the target IP address is local address
1036 configured on the incoming interface
1037 2 - reply only if the target IP address is local address
1038 configured on the incoming interface and both with the
1039 sender's IP address are part from same subnet on this interface
1040 3 - do not reply for local addresses configured with scope host,
1041 only resolutions for global and link addresses are replied
1043 8 - do not reply for all local addresses
1045 The max value from conf/{all,interface}/arp_ignore is used
1046 when ARP request is received on the {interface}
1048 arp_notify - BOOLEAN
1049 Define mode for notification of address and device changes.
1050 0 - (default): do nothing
1051 1 - Generate gratuitous arp requests when device is brought up
1052 or hardware address changes.
1054 arp_accept - BOOLEAN
1055 Define behavior for gratuitous ARP frames who's IP is not
1056 already present in the ARP table:
1057 0 - don't create new entries in the ARP table
1058 1 - create new entries in the ARP table
1060 Both replies and requests type gratuitous arp will trigger the
1061 ARP table to be updated, if this setting is on.
1063 If the ARP table already contains the IP address of the
1064 gratuitous arp frame, the arp table will be updated regardless
1065 if this setting is on or off.
1068 app_solicit - INTEGER
1069 The maximum number of probes to send to the user space ARP daemon
1070 via netlink before dropping back to multicast probes (see
1071 mcast_solicit). Defaults to 0.
1073 disable_policy - BOOLEAN
1074 Disable IPSEC policy (SPD) for this interface
1076 disable_xfrm - BOOLEAN
1077 Disable IPSEC encryption on this interface, whatever the policy
1079 igmpv2_unsolicited_report_interval - INTEGER
1080 The interval in milliseconds in which the next unsolicited
1081 IGMPv1 or IGMPv2 report retransmit will take place.
1082 Default: 10000 (10 seconds)
1084 igmpv3_unsolicited_report_interval - INTEGER
1085 The interval in milliseconds in which the next unsolicited
1086 IGMPv3 report retransmit will take place.
1087 Default: 1000 (1 seconds)
1089 promote_secondaries - BOOLEAN
1090 When a primary IP address is removed from this interface
1091 promote a corresponding secondary IP address instead of
1092 removing all the corresponding secondary IP addresses.
1096 Allows you to write a number, which can be used as required.
1100 kuznet@ms2.inr.ac.ru
1106 delon.nicolas@wanadoo.fr
1111 /proc/sys/net/ipv6/* Variables:
1113 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1114 apply to IPv6 [XXX?].
1116 bindv6only - BOOLEAN
1117 Default value for IPV6_V6ONLY socket option,
1118 which restricts use of the IPv6 socket to IPv6 communication
1120 TRUE: disable IPv4-mapped address feature
1121 FALSE: enable IPv4-mapped address feature
1123 Default: FALSE (as specified in RFC3493)
1125 flowlabel_consistency - BOOLEAN
1126 Protect the consistency (and unicity) of flow label.
1127 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1133 auto_flowlabels - BOOLEAN
1134 Automatically generate flow labels based based on a flow hash
1135 of the packet. This allows intermediate devices, such as routers,
1136 to idenfify packet flows for mechanisms like Equal Cost Multipath
1137 Routing (see RFC 6438).
1142 anycast_src_echo_reply - BOOLEAN
1143 Controls the use of anycast addresses as source addresses for ICMPv6
1151 ip6frag_high_thresh - INTEGER
1152 Maximum memory used to reassemble IPv6 fragments. When
1153 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1154 the fragment handler will toss packets until ip6frag_low_thresh
1157 ip6frag_low_thresh - INTEGER
1158 See ip6frag_high_thresh
1160 ip6frag_time - INTEGER
1161 Time in seconds to keep an IPv6 fragment in memory.
1164 Change the interface-specific default settings.
1168 Change all the interface-specific settings.
1170 [XXX: Other special features than forwarding?]
1172 conf/all/forwarding - BOOLEAN
1173 Enable global IPv6 forwarding between all interfaces.
1175 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1176 to control which interfaces may forward packets and which not.
1178 This also sets all interfaces' Host/Router setting
1179 'forwarding' to the specified value. See below for details.
1181 This referred to as global forwarding.
1187 Change special settings per interface.
1189 The functional behaviour for certain settings is different
1190 depending on whether local forwarding is enabled or not.
1193 Accept Router Advertisements; autoconfigure using them.
1195 It also determines whether or not to transmit Router
1196 Solicitations. If and only if the functional setting is to
1197 accept Router Advertisements, Router Solicitations will be
1200 Possible values are:
1201 0 Do not accept Router Advertisements.
1202 1 Accept Router Advertisements if forwarding is disabled.
1203 2 Overrule forwarding behaviour. Accept Router Advertisements
1204 even if forwarding is enabled.
1206 Functional default: enabled if local forwarding is disabled.
1207 disabled if local forwarding is enabled.
1209 accept_ra_defrtr - BOOLEAN
1210 Learn default router in Router Advertisement.
1212 Functional default: enabled if accept_ra is enabled.
1213 disabled if accept_ra is disabled.
1215 accept_ra_from_local - BOOLEAN
1216 Accept RA with source-address that is found on local machine
1217 if the RA is otherwise proper and able to be accepted.
1218 Default is to NOT accept these as it may be an un-intended
1222 enabled if accept_ra_from_local is enabled
1223 on a specific interface.
1224 disabled if accept_ra_from_local is disabled
1225 on a specific interface.
1227 accept_ra_pinfo - BOOLEAN
1228 Learn Prefix Information in Router Advertisement.
1230 Functional default: enabled if accept_ra is enabled.
1231 disabled if accept_ra is disabled.
1233 accept_ra_rt_info_max_plen - INTEGER
1234 Maximum prefix length of Route Information in RA.
1236 Route Information w/ prefix larger than or equal to this
1237 variable shall be ignored.
1239 Functional default: 0 if accept_ra_rtr_pref is enabled.
1240 -1 if accept_ra_rtr_pref is disabled.
1242 accept_ra_rtr_pref - BOOLEAN
1243 Accept Router Preference in RA.
1245 Functional default: enabled if accept_ra is enabled.
1246 disabled if accept_ra is disabled.
1248 accept_redirects - BOOLEAN
1251 Functional default: enabled if local forwarding is disabled.
1252 disabled if local forwarding is enabled.
1254 accept_source_route - INTEGER
1255 Accept source routing (routing extension header).
1257 >= 0: Accept only routing header type 2.
1258 < 0: Do not accept routing header.
1263 Autoconfigure addresses using Prefix Information in Router
1266 Functional default: enabled if accept_ra_pinfo is enabled.
1267 disabled if accept_ra_pinfo is disabled.
1269 dad_transmits - INTEGER
1270 The amount of Duplicate Address Detection probes to send.
1273 forwarding - INTEGER
1274 Configure interface-specific Host/Router behaviour.
1276 Note: It is recommended to have the same setting on all
1277 interfaces; mixed router/host scenarios are rather uncommon.
1279 Possible values are:
1280 0 Forwarding disabled
1281 1 Forwarding enabled
1285 By default, Host behaviour is assumed. This means:
1287 1. IsRouter flag is not set in Neighbour Advertisements.
1288 2. If accept_ra is TRUE (default), transmit Router
1290 3. If accept_ra is TRUE (default), accept Router
1291 Advertisements (and do autoconfiguration).
1292 4. If accept_redirects is TRUE (default), accept Redirects.
1296 If local forwarding is enabled, Router behaviour is assumed.
1297 This means exactly the reverse from the above:
1299 1. IsRouter flag is set in Neighbour Advertisements.
1300 2. Router Solicitations are not sent unless accept_ra is 2.
1301 3. Router Advertisements are ignored unless accept_ra is 2.
1302 4. Redirects are ignored.
1304 Default: 0 (disabled) if global forwarding is disabled (default),
1305 otherwise 1 (enabled).
1308 Default Hop Limit to set.
1312 Default Maximum Transfer Unit
1313 Default: 1280 (IPv6 required minimum)
1315 router_probe_interval - INTEGER
1316 Minimum interval (in seconds) between Router Probing described
1321 router_solicitation_delay - INTEGER
1322 Number of seconds to wait after interface is brought up
1323 before sending Router Solicitations.
1326 router_solicitation_interval - INTEGER
1327 Number of seconds to wait between Router Solicitations.
1330 router_solicitations - INTEGER
1331 Number of Router Solicitations to send until assuming no
1332 routers are present.
1335 use_tempaddr - INTEGER
1336 Preference for Privacy Extensions (RFC3041).
1337 <= 0 : disable Privacy Extensions
1338 == 1 : enable Privacy Extensions, but prefer public
1339 addresses over temporary addresses.
1340 > 1 : enable Privacy Extensions and prefer temporary
1341 addresses over public addresses.
1342 Default: 0 (for most devices)
1343 -1 (for point-to-point devices and loopback devices)
1345 temp_valid_lft - INTEGER
1346 valid lifetime (in seconds) for temporary addresses.
1347 Default: 604800 (7 days)
1349 temp_prefered_lft - INTEGER
1350 Preferred lifetime (in seconds) for temporary addresses.
1351 Default: 86400 (1 day)
1353 max_desync_factor - INTEGER
1354 Maximum value for DESYNC_FACTOR, which is a random value
1355 that ensures that clients don't synchronize with each
1356 other and generate new addresses at exactly the same time.
1357 value is in seconds.
1360 regen_max_retry - INTEGER
1361 Number of attempts before give up attempting to generate
1362 valid temporary addresses.
1365 max_addresses - INTEGER
1366 Maximum number of autoconfigured addresses per interface. Setting
1367 to zero disables the limitation. It is not recommended to set this
1368 value too large (or to zero) because it would be an easy way to
1369 crash the kernel by allowing too many addresses to be created.
1372 disable_ipv6 - BOOLEAN
1373 Disable IPv6 operation. If accept_dad is set to 2, this value
1374 will be dynamically set to TRUE if DAD fails for the link-local
1376 Default: FALSE (enable IPv6 operation)
1378 When this value is changed from 1 to 0 (IPv6 is being enabled),
1379 it will dynamically create a link-local address on the given
1380 interface and start Duplicate Address Detection, if necessary.
1382 When this value is changed from 0 to 1 (IPv6 is being disabled),
1383 it will dynamically delete all address on the given interface.
1385 accept_dad - INTEGER
1386 Whether to accept DAD (Duplicate Address Detection).
1388 1: Enable DAD (default)
1389 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1390 link-local address has been found.
1392 force_tllao - BOOLEAN
1393 Enable sending the target link-layer address option even when
1394 responding to a unicast neighbor solicitation.
1397 Quoting from RFC 2461, section 4.4, Target link-layer address:
1399 "The option MUST be included for multicast solicitations in order to
1400 avoid infinite Neighbor Solicitation "recursion" when the peer node
1401 does not have a cache entry to return a Neighbor Advertisements
1402 message. When responding to unicast solicitations, the option can be
1403 omitted since the sender of the solicitation has the correct link-
1404 layer address; otherwise it would not have be able to send the unicast
1405 solicitation in the first place. However, including the link-layer
1406 address in this case adds little overhead and eliminates a potential
1407 race condition where the sender deletes the cached link-layer address
1408 prior to receiving a response to a previous solicitation."
1410 ndisc_notify - BOOLEAN
1411 Define mode for notification of address and device changes.
1412 0 - (default): do nothing
1413 1 - Generate unsolicited neighbour advertisements when device is brought
1414 up or hardware address changes.
1416 mldv1_unsolicited_report_interval - INTEGER
1417 The interval in milliseconds in which the next unsolicited
1418 MLDv1 report retransmit will take place.
1419 Default: 10000 (10 seconds)
1421 mldv2_unsolicited_report_interval - INTEGER
1422 The interval in milliseconds in which the next unsolicited
1423 MLDv2 report retransmit will take place.
1424 Default: 1000 (1 second)
1426 force_mld_version - INTEGER
1427 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
1428 1 - Enforce to use MLD version 1
1429 2 - Enforce to use MLD version 2
1431 suppress_frag_ndisc - INTEGER
1432 Control RFC 6980 (Security Implications of IPv6 Fragmentation
1433 with IPv6 Neighbor Discovery) behavior:
1434 1 - (default) discard fragmented neighbor discovery packets
1435 0 - allow fragmented neighbor discovery packets
1439 Limit the maximal rates for sending ICMPv6 packets.
1440 0 to disable any limiting,
1441 otherwise the minimal space between responses in milliseconds.
1446 Pekka Savola <pekkas@netcore.fi>
1447 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1450 /proc/sys/net/bridge/* Variables:
1452 bridge-nf-call-arptables - BOOLEAN
1453 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1457 bridge-nf-call-iptables - BOOLEAN
1458 1 : pass bridged IPv4 traffic to iptables' chains.
1462 bridge-nf-call-ip6tables - BOOLEAN
1463 1 : pass bridged IPv6 traffic to ip6tables' chains.
1467 bridge-nf-filter-vlan-tagged - BOOLEAN
1468 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1472 bridge-nf-filter-pppoe-tagged - BOOLEAN
1473 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1477 bridge-nf-pass-vlan-input-dev - BOOLEAN
1478 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
1479 interface on the bridge and set the netfilter input device to the vlan.
1480 This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
1481 target work with vlan-on-top-of-bridge interfaces. When no matching
1482 vlan interface is found, or this switch is off, the input device is
1483 set to the bridge interface.
1484 0: disable bridge netfilter vlan interface lookup.
1487 proc/sys/net/sctp/* Variables:
1489 addip_enable - BOOLEAN
1490 Enable or disable extension of Dynamic Address Reconfiguration
1491 (ADD-IP) functionality specified in RFC5061. This extension provides
1492 the ability to dynamically add and remove new addresses for the SCTP
1495 1: Enable extension.
1497 0: Disable extension.
1501 addip_noauth_enable - BOOLEAN
1502 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1503 authentication to protect the operations of adding or removing new
1504 addresses. This requirement is mandated so that unauthorized hosts
1505 would not be able to hijack associations. However, older
1506 implementations may not have implemented this requirement while
1507 allowing the ADD-IP extension. For reasons of interoperability,
1508 we provide this variable to control the enforcement of the
1509 authentication requirement.
1511 1: Allow ADD-IP extension to be used without authentication. This
1512 should only be set in a closed environment for interoperability
1513 with older implementations.
1515 0: Enforce the authentication requirement
1519 auth_enable - BOOLEAN
1520 Enable or disable Authenticated Chunks extension. This extension
1521 provides the ability to send and receive authenticated chunks and is
1522 required for secure operation of Dynamic Address Reconfiguration
1525 1: Enable this extension.
1526 0: Disable this extension.
1530 prsctp_enable - BOOLEAN
1531 Enable or disable the Partial Reliability extension (RFC3758) which
1532 is used to notify peers that a given DATA should no longer be expected.
1540 The limit of the number of new packets that can be initially sent. It
1541 controls how bursty the generated traffic can be.
1545 association_max_retrans - INTEGER
1546 Set the maximum number for retransmissions that an association can
1547 attempt deciding that the remote end is unreachable. If this value
1548 is exceeded, the association is terminated.
1552 max_init_retransmits - INTEGER
1553 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1554 that an association will attempt before declaring the destination
1555 unreachable and terminating.
1559 path_max_retrans - INTEGER
1560 The maximum number of retransmissions that will be attempted on a given
1561 path. Once this threshold is exceeded, the path is considered
1562 unreachable, and new traffic will use a different path when the
1563 association is multihomed.
1567 pf_retrans - INTEGER
1568 The number of retransmissions that will be attempted on a given path
1569 before traffic is redirected to an alternate transport (should one
1570 exist). Note this is distinct from path_max_retrans, as a path that
1571 passes the pf_retrans threshold can still be used. Its only
1572 deprioritized when a transmission path is selected by the stack. This
1573 setting is primarily used to enable fast failover mechanisms without
1574 having to reduce path_max_retrans to a very low value. See:
1575 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
1576 for details. Note also that a value of pf_retrans > path_max_retrans
1577 disables this feature
1581 rto_initial - INTEGER
1582 The initial round trip timeout value in milliseconds that will be used
1583 in calculating round trip times. This is the initial time interval
1584 for retransmissions.
1589 The maximum value (in milliseconds) of the round trip timeout. This
1590 is the largest time interval that can elapse between retransmissions.
1595 The minimum value (in milliseconds) of the round trip timeout. This
1596 is the smallest time interval the can elapse between retransmissions.
1600 hb_interval - INTEGER
1601 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
1602 are sent at the specified interval on idle paths to probe the state of
1603 a given path between 2 associations.
1607 sack_timeout - INTEGER
1608 The amount of time (in milliseconds) that the implementation will wait
1613 valid_cookie_life - INTEGER
1614 The default lifetime of the SCTP cookie (in milliseconds). The cookie
1615 is used during association establishment.
1619 cookie_preserve_enable - BOOLEAN
1620 Enable or disable the ability to extend the lifetime of the SCTP cookie
1621 that is used during the establishment phase of SCTP association
1623 1: Enable cookie lifetime extension.
1628 cookie_hmac_alg - STRING
1629 Select the hmac algorithm used when generating the cookie value sent by
1630 a listening sctp socket to a connecting client in the INIT-ACK chunk.
1635 Ability to assign md5 or sha1 as the selected alg is predicated on the
1636 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
1637 CONFIG_CRYPTO_SHA1).
1639 Default: Dependent on configuration. MD5 if available, else SHA1 if
1640 available, else none.
1642 rcvbuf_policy - INTEGER
1643 Determines if the receive buffer is attributed to the socket or to
1644 association. SCTP supports the capability to create multiple
1645 associations on a single socket. When using this capability, it is
1646 possible that a single stalled association that's buffering a lot
1647 of data may block other associations from delivering their data by
1648 consuming all of the receive buffer space. To work around this,
1649 the rcvbuf_policy could be set to attribute the receiver buffer space
1650 to each association instead of the socket. This prevents the described
1653 1: rcvbuf space is per association
1654 0: rcvbuf space is per socket
1658 sndbuf_policy - INTEGER
1659 Similar to rcvbuf_policy above, this applies to send buffer space.
1661 1: Send buffer is tracked per association
1662 0: Send buffer is tracked per socket.
1666 sctp_mem - vector of 3 INTEGERs: min, pressure, max
1667 Number of pages allowed for queueing by all SCTP sockets.
1669 min: Below this number of pages SCTP is not bothered about its
1670 memory appetite. When amount of memory allocated by SCTP exceeds
1671 this number, SCTP starts to moderate memory usage.
1673 pressure: This value was introduced to follow format of tcp_mem.
1675 max: Number of pages allowed for queueing by all SCTP sockets.
1677 Default is calculated at boot time from amount of available memory.
1679 sctp_rmem - vector of 3 INTEGERs: min, default, max
1680 Only the first value ("min") is used, "default" and "max" are
1683 min: Minimal size of receive buffer used by SCTP socket.
1684 It is guaranteed to each SCTP socket (but not association) even
1685 under moderate memory pressure.
1689 sctp_wmem - vector of 3 INTEGERs: min, default, max
1690 Currently this tunable has no effect.
1692 addr_scope_policy - INTEGER
1693 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
1695 0 - Disable IPv4 address scoping
1696 1 - Enable IPv4 address scoping
1697 2 - Follow draft but allow IPv4 private addresses
1698 3 - Follow draft but allow IPv4 link local addresses
1703 /proc/sys/net/core/*
1704 Please see: Documentation/sysctl/net.txt for descriptions of these entries.
1707 /proc/sys/net/unix/*
1708 max_dgram_qlen - INTEGER
1709 The maximum length of dgram socket receive queue
1716 /proc/sys/net/irda/*
1717 fast_poll_increase FIXME
1718 warn_noreply_time FIXME
1719 discovery_slots FIXME
1722 discovery_timeout FIXME
1723 lap_keepalive_time FIXME
1724 max_noreply_time FIXME
1725 max_tx_data_size FIXME
1727 min_tx_turn_time FIXME