| blob | 8be76235fe6724c43e0c2b39778f3f741e53b619 |
1 /proc/sys/net/ipv4/* Variables:
3 ip_forward - BOOLEAN
4 0 - disabled (default)
5 not 0 - enabled
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
11 for routers)
13 ip_default_ttl - INTEGER
14 default 64
16 ip_no_pmtu_disc - BOOLEAN
17 Disable Path MTU Discovery.
18 default FALSE
20 min_pmtu - INTEGER
21 default 562 - minimum discovered Path MTU
23 mtu_expires - INTEGER
24 Time, in seconds, that cached PMTU information is kept.
26 min_adv_mss - INTEGER
27 The advertised MSS depends on the first hop route MTU, but will
28 never be lower than this setting.
30 rt_cache_rebuild_count - INTEGER
31 The per net-namespace route cache emergency rebuild threshold.
32 Any net-namespace having its route cache rebuilt due to
33 a hash bucket chain being too long more than this many times
34 will have its route caching disabled
36 IP Fragmentation:
38 ipfrag_high_thresh - INTEGER
39 Maximum memory used to reassemble IP fragments. When
40 ipfrag_high_thresh bytes of memory is allocated for this purpose,
41 the fragment handler will toss packets until ipfrag_low_thresh
42 is reached.
44 ipfrag_low_thresh - INTEGER
45 See ipfrag_high_thresh
47 ipfrag_time - INTEGER
48 Time in seconds to keep an IP fragment in memory.
50 ipfrag_secret_interval - INTEGER
51 Regeneration interval (in seconds) of the hash secret (or lifetime
52 for the hash secret) for IP fragments.
53 Default: 600
55 ipfrag_max_dist - INTEGER
56 ipfrag_max_dist is a non-negative integer value which defines the
57 maximum "disorder" which is allowed among fragments which share a
58 common IP source address. Note that reordering of packets is
59 not unusual, but if a large number of fragments arrive from a source
60 IP address while a particular fragment queue remains incomplete, it
61 probably indicates that one or more fragments belonging to that queue
62 have been lost. When ipfrag_max_dist is positive, an additional check
63 is done on fragments before they are added to a reassembly queue - if
64 ipfrag_max_dist (or more) fragments have arrived from a particular IP
65 address between additions to any IP fragment queue using that source
66 address, it's presumed that one or more fragments in the queue are
67 lost. The existing fragment queue will be dropped, and a new one
68 started. An ipfrag_max_dist value of zero disables this check.
70 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
71 result in unnecessarily dropping fragment queues when normal
72 reordering of packets occurs, which could lead to poor application
73 performance. Using a very large value, e.g. 50000, increases the
74 likelihood of incorrectly reassembling IP fragments that originate
75 from different IP datagrams, which could result in data corruption.
76 Default: 64
78 INET peer storage:
80 inet_peer_threshold - INTEGER
81 The approximate size of the storage. Starting from this threshold
82 entries will be thrown aggressively. This threshold also determines
83 entries' time-to-live and time intervals between garbage collection
84 passes. More entries, less time-to-live, less GC interval.
86 inet_peer_minttl - INTEGER
87 Minimum time-to-live of entries. Should be enough to cover fragment
88 time-to-live on the reassembling side. This minimum time-to-live is
89 guaranteed if the pool size is less than inet_peer_threshold.
90 Measured in seconds.
92 inet_peer_maxttl - INTEGER
93 Maximum time-to-live of entries. Unused entries will expire after
94 this period of time if there is no memory pressure on the pool (i.e.
95 when the number of entries in the pool is very small).
96 Measured in seconds.
98 inet_peer_gc_mintime - INTEGER
99 Minimum interval between garbage collection passes. This interval is
100 in effect under high memory pressure on the pool.
101 Measured in seconds.
103 inet_peer_gc_maxtime - INTEGER
104 Minimum interval between garbage collection passes. This interval is
105 in effect under low (or absent) memory pressure on the pool.
106 Measured in seconds.
108 TCP variables:
110 somaxconn - INTEGER
111 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
112 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
113 for TCP sockets.
115 tcp_abc - INTEGER
116 Controls Appropriate Byte Count (ABC) defined in RFC3465.
117 ABC is a way of increasing congestion window (cwnd) more slowly
118 in response to partial acknowledgments.
119 Possible values are:
120 0 increase cwnd once per acknowledgment (no ABC)
121 1 increase cwnd once per acknowledgment of full sized segment
122 2 allow increase cwnd by two if acknowledgment is
123 of two segments to compensate for delayed acknowledgments.
124 Default: 0 (off)
126 tcp_abort_on_overflow - BOOLEAN
127 If listening service is too slow to accept new connections,
128 reset them. Default state is FALSE. It means that if overflow
129 occurred due to a burst, connection will recover. Enable this
130 option _only_ if you are really sure that listening daemon
131 cannot be tuned to accept connections faster. Enabling this
132 option can harm clients of your server.
134 tcp_adv_win_scale - INTEGER
135 Count buffering overhead as bytes/2^tcp_adv_win_scale
136 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
137 if it is <= 0.
138 Default: 2
140 tcp_allowed_congestion_control - STRING
141 Show/set the congestion control choices available to non-privileged
142 processes. The list is a subset of those listed in
143 tcp_available_congestion_control.
144 Default is "reno" and the default setting (tcp_congestion_control).
146 tcp_app_win - INTEGER
147 Reserve max(window/2^tcp_app_win, mss) of window for application
148 buffer. Value 0 is special, it means that nothing is reserved.
149 Default: 31
151 tcp_available_congestion_control - STRING
152 Shows the available congestion control choices that are registered.
153 More congestion control algorithms may be available as modules,
154 but not loaded.
156 tcp_base_mss - INTEGER
157 The initial value of search_low to be used by the packetization layer
158 Path MTU discovery (MTU probing). If MTU probing is enabled,
159 this is the initial MSS used by the connection.
161 tcp_congestion_control - STRING
162 Set the congestion control algorithm to be used for new
163 connections. The algorithm "reno" is always available, but
164 additional choices may be available based on kernel configuration.
165 Default is set as part of kernel configuration.
167 tcp_dsack - BOOLEAN
168 Allows TCP to send "duplicate" SACKs.
170 tcp_ecn - BOOLEAN
171 Enable Explicit Congestion Notification (ECN) in TCP. ECN is only
172 used when both ends of the TCP flow support it. It is useful to
173 avoid losses due to congestion (when the bottleneck router supports
174 ECN).
175 Possible values are:
176 0 disable ECN
177 1 ECN enabled
178 2 Only server-side ECN enabled. If the other end does
179 not support ECN, behavior is like with ECN disabled.
180 Default: 2
182 tcp_fack - BOOLEAN
183 Enable FACK congestion avoidance and fast retransmission.
184 The value is not used, if tcp_sack is not enabled.
186 tcp_fin_timeout - INTEGER
187 Time to hold socket in state FIN-WAIT-2, if it was closed
188 by our side. Peer can be broken and never close its side,
189 or even died unexpectedly. Default value is 60sec.
190 Usual value used in 2.2 was 180 seconds, you may restore
191 it, but remember that if your machine is even underloaded WEB server,
192 you risk to overflow memory with kilotons of dead sockets,
193 FIN-WAIT-2 sockets are less dangerous than FIN-WAIT-1,
194 because they eat maximum 1.5K of memory, but they tend
195 to live longer. Cf. tcp_max_orphans.
197 tcp_frto - INTEGER
198 Enables Forward RTO-Recovery (F-RTO) defined in RFC4138.
199 F-RTO is an enhanced recovery algorithm for TCP retransmission
200 timeouts. It is particularly beneficial in wireless environments
201 where packet loss is typically due to random radio interference
202 rather than intermediate router congestion. F-RTO is sender-side
203 only modification. Therefore it does not require any support from
204 the peer.
206 If set to 1, basic version is enabled. 2 enables SACK enhanced
207 F-RTO if flow uses SACK. The basic version can be used also when
208 SACK is in use though scenario(s) with it exists where F-RTO
209 interacts badly with the packet counting of the SACK enabled TCP
210 flow.
212 tcp_frto_response - INTEGER
213 When F-RTO has detected that a TCP retransmission timeout was
214 spurious (i.e, the timeout would have been avoided had TCP set a
215 longer retransmission timeout), TCP has several options what to do
216 next. Possible values are:
217 0 Rate halving based; a smooth and conservative response,
218 results in halved cwnd and ssthresh after one RTT
219 1 Very conservative response; not recommended because even
220 though being valid, it interacts poorly with the rest of
221 Linux TCP, halves cwnd and ssthresh immediately
222 2 Aggressive response; undoes congestion control measures
223 that are now known to be unnecessary (ignoring the
224 possibility of a lost retransmission that would require
225 TCP to be more cautious), cwnd and ssthresh are restored
226 to the values prior timeout
227 Default: 0 (rate halving based)
229 tcp_keepalive_time - INTEGER
230 How often TCP sends out keepalive messages when keepalive is enabled.
231 Default: 2hours.
233 tcp_keepalive_probes - INTEGER
234 How many keepalive probes TCP sends out, until it decides that the
235 connection is broken. Default value: 9.
237 tcp_keepalive_intvl - INTEGER
238 How frequently the probes are send out. Multiplied by
239 tcp_keepalive_probes it is time to kill not responding connection,
240 after probes started. Default value: 75sec i.e. connection
241 will be aborted after ~11 minutes of retries.
243 tcp_low_latency - BOOLEAN
244 If set, the TCP stack makes decisions that prefer lower
245 latency as opposed to higher throughput. By default, this
246 option is not set meaning that higher throughput is preferred.
247 An example of an application where this default should be
248 changed would be a Beowulf compute cluster.
249 Default: 0
251 tcp_max_orphans - INTEGER
252 Maximal number of TCP sockets not attached to any user file handle,
253 held by system. If this number is exceeded orphaned connections are
254 reset immediately and warning is printed. This limit exists
255 only to prevent simple DoS attacks, you _must_ not rely on this
256 or lower the limit artificially, but rather increase it
257 (probably, after increasing installed memory),
258 if network conditions require more than default value,
259 and tune network services to linger and kill such states
260 more aggressively. Let me to remind again: each orphan eats
261 up to ~64K of unswappable memory.
263 tcp_max_syn_backlog - INTEGER
264 Maximal number of remembered connection requests, which are
265 still did not receive an acknowledgment from connecting client.
266 Default value is 1024 for systems with more than 128Mb of memory,
267 and 128 for low memory machines. If server suffers of overload,
268 try to increase this number.
270 tcp_max_tw_buckets - INTEGER
271 Maximal number of timewait sockets held by system simultaneously.
272 If this number is exceeded time-wait socket is immediately destroyed
273 and warning is printed. This limit exists only to prevent
274 simple DoS attacks, you _must_ not lower the limit artificially,
275 but rather increase it (probably, after increasing installed memory),
276 if network conditions require more than default value.
278 tcp_mem - vector of 3 INTEGERs: min, pressure, max
279 min: below this number of pages TCP is not bothered about its
280 memory appetite.
282 pressure: when amount of memory allocated by TCP exceeds this number
283 of pages, TCP moderates its memory consumption and enters memory
284 pressure mode, which is exited when memory consumption falls
285 under "min".
287 max: number of pages allowed for queueing by all TCP sockets.
289 Defaults are calculated at boot time from amount of available
290 memory.
292 tcp_moderate_rcvbuf - BOOLEAN
293 If set, TCP performs receive buffer auto-tuning, attempting to
294 automatically size the buffer (no greater than tcp_rmem[2]) to
295 match the size required by the path for full throughput. Enabled by
296 default.
298 tcp_mtu_probing - INTEGER
299 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
300 values:
301 0 - Disabled
302 1 - Disabled by default, enabled when an ICMP black hole detected
303 2 - Always enabled, use initial MSS of tcp_base_mss.
305 tcp_no_metrics_save - BOOLEAN
306 By default, TCP saves various connection metrics in the route cache
307 when the connection closes, so that connections established in the
308 near future can use these to set initial conditions. Usually, this
309 increases overall performance, but may sometimes cause performance
310 degradation. If set, TCP will not cache metrics on closing
311 connections.
313 tcp_orphan_retries - INTEGER
314 How may times to retry before killing TCP connection, closed
315 by our side. Default value 7 corresponds to ~50sec-16min
316 depending on RTO. If you machine is loaded WEB server,
317 you should think about lowering this value, such sockets
318 may consume significant resources. Cf. tcp_max_orphans.
320 tcp_reordering - INTEGER
321 Maximal reordering of packets in a TCP stream.
322 Default: 3
324 tcp_retrans_collapse - BOOLEAN
325 Bug-to-bug compatibility with some broken printers.
326 On retransmit try to send bigger packets to work around bugs in
327 certain TCP stacks.
329 tcp_retries1 - INTEGER
330 How many times to retry before deciding that something is wrong
331 and it is necessary to report this suspicion to network layer.
332 Minimal RFC value is 3, it is default, which corresponds
333 to ~3sec-8min depending on RTO.
335 tcp_retries2 - INTEGER
336 How may times to retry before killing alive TCP connection.
337 RFC1122 says that the limit should be longer than 100 sec.
338 It is too small number. Default value 15 corresponds to ~13-30min
339 depending on RTO.
341 tcp_rfc1337 - BOOLEAN
342 If set, the TCP stack behaves conforming to RFC1337. If unset,
343 we are not conforming to RFC, but prevent TCP TIME_WAIT
344 assassination.
345 Default: 0
347 tcp_rmem - vector of 3 INTEGERs: min, default, max
348 min: Minimal size of receive buffer used by TCP sockets.
349 It is guaranteed to each TCP socket, even under moderate memory
350 pressure.
351 Default: 8K
353 default: initial size of receive buffer used by TCP sockets.
354 This value overrides net.core.rmem_default used by other protocols.
355 Default: 87380 bytes. This value results in window of 65535 with
356 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
357 less for default tcp_app_win. See below about these variables.
359 max: maximal size of receive buffer allowed for automatically
360 selected receiver buffers for TCP socket. This value does not override
361 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
362 automatic tuning of that socket's receive buffer size, in which
363 case this value is ignored.
364 Default: between 87380B and 4MB, depending on RAM size.
366 tcp_sack - BOOLEAN
367 Enable select acknowledgments (SACKS).
369 tcp_slow_start_after_idle - BOOLEAN
370 If set, provide RFC2861 behavior and time out the congestion
371 window after an idle period. An idle period is defined at
372 the current RTO. If unset, the congestion window will not
373 be timed out after an idle period.
374 Default: 1
376 tcp_stdurg - BOOLEAN
377 Use the Host requirements interpretation of the TCP urgent pointer field.
378 Most hosts use the older BSD interpretation, so if you turn this on
379 Linux might not communicate correctly with them.
380 Default: FALSE
382 tcp_synack_retries - INTEGER
383 Number of times SYNACKs for a passive TCP connection attempt will
384 be retransmitted. Should not be higher than 255. Default value
385 is 5, which corresponds to ~180seconds.
387 tcp_syncookies - BOOLEAN
388 Only valid when the kernel was compiled with CONFIG_SYNCOOKIES
389 Send out syncookies when the syn backlog queue of a socket
390 overflows. This is to prevent against the common 'SYN flood attack'
391 Default: FALSE
393 Note, that syncookies is fallback facility.
394 It MUST NOT be used to help highly loaded servers to stand
395 against legal connection rate. If you see SYN flood warnings
396 in your logs, but investigation shows that they occur
397 because of overload with legal connections, you should tune
398 another parameters until this warning disappear.
399 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
401 syncookies seriously violate TCP protocol, do not allow
402 to use TCP extensions, can result in serious degradation
403 of some services (f.e. SMTP relaying), visible not by you,
404 but your clients and relays, contacting you. While you see
405 SYN flood warnings in logs not being really flooded, your server
406 is seriously misconfigured.
408 tcp_syn_retries - INTEGER
409 Number of times initial SYNs for an active TCP connection attempt
410 will be retransmitted. Should not be higher than 255. Default value
411 is 5, which corresponds to ~180seconds.
413 tcp_timestamps - BOOLEAN
414 Enable timestamps as defined in RFC1323.
416 tcp_tso_win_divisor - INTEGER
417 This allows control over what percentage of the congestion window
418 can be consumed by a single TSO frame.
419 The setting of this parameter is a choice between burstiness and
420 building larger TSO frames.
421 Default: 3
423 tcp_tw_recycle - BOOLEAN
424 Enable fast recycling TIME-WAIT sockets. Default value is 0.
425 It should not be changed without advice/request of technical
426 experts.
428 tcp_tw_reuse - BOOLEAN
429 Allow to reuse TIME-WAIT sockets for new connections when it is
430 safe from protocol viewpoint. Default value is 0.
431 It should not be changed without advice/request of technical
432 experts.
434 tcp_window_scaling - BOOLEAN
435 Enable window scaling as defined in RFC1323.
437 tcp_wmem - vector of 3 INTEGERs: min, default, max
438 min: Amount of memory reserved for send buffers for TCP sockets.
439 Each TCP socket has rights to use it due to fact of its birth.
440 Default: 4K
442 default: initial size of send buffer used by TCP sockets. This
443 value overrides net.core.wmem_default used by other protocols.
444 It is usually lower than net.core.wmem_default.
445 Default: 16K
447 max: Maximal amount of memory allowed for automatically tuned
448 send buffers for TCP sockets. This value does not override
449 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
450 automatic tuning of that socket's send buffer size, in which case
451 this value is ignored.
452 Default: between 64K and 4MB, depending on RAM size.
454 tcp_workaround_signed_windows - BOOLEAN
455 If set, assume no receipt of a window scaling option means the
456 remote TCP is broken and treats the window as a signed quantity.
457 If unset, assume the remote TCP is not broken even if we do
458 not receive a window scaling option from them.
459 Default: 0
461 tcp_dma_copybreak - INTEGER
462 Lower limit, in bytes, of the size of socket reads that will be
463 offloaded to a DMA copy engine, if one is present in the system
464 and CONFIG_NET_DMA is enabled.
465 Default: 4096
467 UDP variables:
469 udp_mem - vector of 3 INTEGERs: min, pressure, max
470 Number of pages allowed for queueing by all UDP sockets.
472 min: Below this number of pages UDP is not bothered about its
473 memory appetite. When amount of memory allocated by UDP exceeds
474 this number, UDP starts to moderate memory usage.
476 pressure: This value was introduced to follow format of tcp_mem.
478 max: Number of pages allowed for queueing by all UDP sockets.
480 Default is calculated at boot time from amount of available memory.
482 udp_rmem_min - INTEGER
483 Minimal size of receive buffer used by UDP sockets in moderation.
484 Each UDP socket is able to use the size for receiving data, even if
485 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
486 Default: 4096
488 udp_wmem_min - INTEGER
489 Minimal size of send buffer used by UDP sockets in moderation.
490 Each UDP socket is able to use the size for sending data, even if
491 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
492 Default: 4096
494 CIPSOv4 Variables:
496 cipso_cache_enable - BOOLEAN
497 If set, enable additions to and lookups from the CIPSO label mapping
498 cache. If unset, additions are ignored and lookups always result in a
499 miss. However, regardless of the setting the cache is still
500 invalidated when required when means you can safely toggle this on and
501 off and the cache will always be "safe".
502 Default: 1
504 cipso_cache_bucket_size - INTEGER
505 The CIPSO label cache consists of a fixed size hash table with each
506 hash bucket containing a number of cache entries. This variable limits
507 the number of entries in each hash bucket; the larger the value the
508 more CIPSO label mappings that can be cached. When the number of
509 entries in a given hash bucket reaches this limit adding new entries
510 causes the oldest entry in the bucket to be removed to make room.
511 Default: 10
513 cipso_rbm_optfmt - BOOLEAN
514 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
515 the CIPSO draft specification (see Documentation/netlabel for details).
516 This means that when set the CIPSO tag will be padded with empty
517 categories in order to make the packet data 32-bit aligned.
518 Default: 0
520 cipso_rbm_structvalid - BOOLEAN
521 If set, do a very strict check of the CIPSO option when
522 ip_options_compile() is called. If unset, relax the checks done during
523 ip_options_compile(). Either way is "safe" as errors are caught else
524 where in the CIPSO processing code but setting this to 0 (False) should
525 result in less work (i.e. it should be faster) but could cause problems
526 with other implementations that require strict checking.
527 Default: 0
529 IP Variables:
531 ip_local_port_range - 2 INTEGERS
532 Defines the local port range that is used by TCP and UDP to
533 choose the local port. The first number is the first, the
534 second the last local port number. Default value depends on
535 amount of memory available on the system:
536 > 128Mb 32768-61000
537 < 128Mb 1024-4999 or even less.
538 This number defines number of active connections, which this
539 system can issue simultaneously to systems not supporting
540 TCP extensions (timestamps). With tcp_tw_recycle enabled
541 (i.e. by default) range 1024-4999 is enough to issue up to
542 2000 connections per second to systems supporting timestamps.
544 ip_nonlocal_bind - BOOLEAN
545 If set, allows processes to bind() to non-local IP addresses,
546 which can be quite useful - but may break some applications.
547 Default: 0
549 ip_dynaddr - BOOLEAN
550 If set non-zero, enables support for dynamic addresses.
551 If set to a non-zero value larger than 1, a kernel log
552 message will be printed when dynamic address rewriting
553 occurs.
554 Default: 0
556 icmp_echo_ignore_all - BOOLEAN
557 If set non-zero, then the kernel will ignore all ICMP ECHO
558 requests sent to it.
559 Default: 0
561 icmp_echo_ignore_broadcasts - BOOLEAN
562 If set non-zero, then the kernel will ignore all ICMP ECHO and
563 TIMESTAMP requests sent to it via broadcast/multicast.
564 Default: 1
566 icmp_ratelimit - INTEGER
567 Limit the maximal rates for sending ICMP packets whose type matches
568 icmp_ratemask (see below) to specific targets.
569 0 to disable any limiting,
570 otherwise the minimal space between responses in milliseconds.
571 Default: 1000
573 icmp_ratemask - INTEGER
574 Mask made of ICMP types for which rates are being limited.
575 Significant bits: IHGFEDCBA9876543210
576 Default mask: 0000001100000011000 (6168)
578 Bit definitions (see include/linux/icmp.h):
579 0 Echo Reply
580 3 Destination Unreachable *
581 4 Source Quench *
582 5 Redirect
583 8 Echo Request
584 B Time Exceeded *
585 C Parameter Problem *
586 D Timestamp Request
587 E Timestamp Reply
588 F Info Request
589 G Info Reply
590 H Address Mask Request
591 I Address Mask Reply
593 * These are rate limited by default (see default mask above)
595 icmp_ignore_bogus_error_responses - BOOLEAN
596 Some routers violate RFC1122 by sending bogus responses to broadcast
597 frames. Such violations are normally logged via a kernel warning.
598 If this is set to TRUE, the kernel will not give such warnings, which
599 will avoid log file clutter.
600 Default: FALSE
602 icmp_errors_use_inbound_ifaddr - BOOLEAN
604 If zero, icmp error messages are sent with the primary address of
605 the exiting interface.
607 If non-zero, the message will be sent with the primary address of
608 the interface that received the packet that caused the icmp error.
609 This is the behaviour network many administrators will expect from
610 a router. And it can make debugging complicated network layouts
611 much easier.
613 Note that if no primary address exists for the interface selected,
614 then the primary address of the first non-loopback interface that
615 has one will be used regardless of this setting.
617 Default: 0
619 igmp_max_memberships - INTEGER
620 Change the maximum number of multicast groups we can subscribe to.
621 Default: 20
623 conf/interface/* changes special settings per interface (where "interface" is
624 the name of your network interface)
625 conf/all/* is special, changes the settings for all interfaces
628 log_martians - BOOLEAN
629 Log packets with impossible addresses to kernel log.
630 log_martians for the interface will be enabled if at least one of
631 conf/{all,interface}/log_martians is set to TRUE,
632 it will be disabled otherwise
634 accept_redirects - BOOLEAN
635 Accept ICMP redirect messages.
636 accept_redirects for the interface will be enabled if:
637 - both conf/{all,interface}/accept_redirects are TRUE in the case
638 forwarding for the interface is enabled
639 or
640 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
641 case forwarding for the interface is disabled
642 accept_redirects for the interface will be disabled otherwise
643 default TRUE (host)
644 FALSE (router)
646 forwarding - BOOLEAN
647 Enable IP forwarding on this interface.
649 mc_forwarding - BOOLEAN
650 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
651 and a multicast routing daemon is required.
652 conf/all/mc_forwarding must also be set to TRUE to enable multicast
653 routing for the interface
655 medium_id - INTEGER
656 Integer value used to differentiate the devices by the medium they
657 are attached to. Two devices can have different id values when
658 the broadcast packets are received only on one of them.
659 The default value 0 means that the device is the only interface
660 to its medium, value of -1 means that medium is not known.
662 Currently, it is used to change the proxy_arp behavior:
663 the proxy_arp feature is enabled for packets forwarded between
664 two devices attached to different media.
666 proxy_arp - BOOLEAN
667 Do proxy arp.
668 proxy_arp for the interface will be enabled if at least one of
669 conf/{all,interface}/proxy_arp is set to TRUE,
670 it will be disabled otherwise
672 shared_media - BOOLEAN
673 Send(router) or accept(host) RFC1620 shared media redirects.
674 Overrides ip_secure_redirects.
675 shared_media for the interface will be enabled if at least one of
676 conf/{all,interface}/shared_media is set to TRUE,
677 it will be disabled otherwise
678 default TRUE
680 secure_redirects - BOOLEAN
681 Accept ICMP redirect messages only for gateways,
682 listed in default gateway list.
683 secure_redirects for the interface will be enabled if at least one of
684 conf/{all,interface}/secure_redirects is set to TRUE,
685 it will be disabled otherwise
686 default TRUE
688 send_redirects - BOOLEAN
689 Send redirects, if router.
690 send_redirects for the interface will be enabled if at least one of
691 conf/{all,interface}/send_redirects is set to TRUE,
692 it will be disabled otherwise
693 Default: TRUE
695 bootp_relay - BOOLEAN
696 Accept packets with source address 0.b.c.d destined
697 not to this host as local ones. It is supposed, that
698 BOOTP relay daemon will catch and forward such packets.
699 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
700 for the interface
701 default FALSE
702 Not Implemented Yet.
704 accept_source_route - BOOLEAN
705 Accept packets with SRR option.
706 conf/all/accept_source_route must also be set to TRUE to accept packets
707 with SRR option on the interface
708 default TRUE (router)
709 FALSE (host)
711 rp_filter - INTEGER
712 0 - No source validation.
713 1 - Strict mode as defined in RFC3704 Strict Reverse Path
714 Each incoming packet is tested against the FIB and if the interface
715 is not the best reverse path the packet check will fail.
716 By default failed packets are discarded.
717 2 - Loose mode as defined in RFC3704 Loose Reverse Path
718 Each incoming packet's source address is also tested against the FIB
719 and if the source address is not reachable via any interface
720 the packet check will fail.
722 Current recommended practice in RFC3704 is to enable strict mode
723 to prevent IP spoofing from DDos attacks. If using asymmetric routing
724 or other complicated routing, then loose mode is recommended.
726 conf/all/rp_filter must also be set to non-zero to do source validation
727 on the interface
729 Default value is 0. Note that some distributions enable it
730 in startup scripts.
732 arp_filter - BOOLEAN
733 1 - Allows you to have multiple network interfaces on the same
734 subnet, and have the ARPs for each interface be answered
735 based on whether or not the kernel would route a packet from
736 the ARP'd IP out that interface (therefore you must use source
737 based routing for this to work). In other words it allows control
738 of which cards (usually 1) will respond to an arp request.
740 0 - (default) The kernel can respond to arp requests with addresses
741 from other interfaces. This may seem wrong but it usually makes
742 sense, because it increases the chance of successful communication.
743 IP addresses are owned by the complete host on Linux, not by
744 particular interfaces. Only for more complex setups like load-
745 balancing, does this behaviour cause problems.
747 arp_filter for the interface will be enabled if at least one of
748 conf/{all,interface}/arp_filter is set to TRUE,
749 it will be disabled otherwise
751 arp_announce - INTEGER
752 Define different restriction levels for announcing the local
753 source IP address from IP packets in ARP requests sent on
754 interface:
755 0 - (default) Use any local address, configured on any interface
756 1 - Try to avoid local addresses that are not in the target's
757 subnet for this interface. This mode is useful when target
758 hosts reachable via this interface require the source IP
759 address in ARP requests to be part of their logical network
760 configured on the receiving interface. When we generate the
761 request we will check all our subnets that include the
762 target IP and will preserve the source address if it is from
763 such subnet. If there is no such subnet we select source
764 address according to the rules for level 2.
765 2 - Always use the best local address for this target.
766 In this mode we ignore the source address in the IP packet
767 and try to select local address that we prefer for talks with
768 the target host. Such local address is selected by looking
769 for primary IP addresses on all our subnets on the outgoing
770 interface that include the target IP address. If no suitable
771 local address is found we select the first local address
772 we have on the outgoing interface or on all other interfaces,
773 with the hope we will receive reply for our request and
774 even sometimes no matter the source IP address we announce.
776 The max value from conf/{all,interface}/arp_announce is used.
778 Increasing the restriction level gives more chance for
779 receiving answer from the resolved target while decreasing
780 the level announces more valid sender's information.
782 arp_ignore - INTEGER
783 Define different modes for sending replies in response to
784 received ARP requests that resolve local target IP addresses:
785 0 - (default): reply for any local target IP address, configured
786 on any interface
787 1 - reply only if the target IP address is local address
788 configured on the incoming interface
789 2 - reply only if the target IP address is local address
790 configured on the incoming interface and both with the
791 sender's IP address are part from same subnet on this interface
792 3 - do not reply for local addresses configured with scope host,
793 only resolutions for global and link addresses are replied
794 4-7 - reserved
795 8 - do not reply for all local addresses
797 The max value from conf/{all,interface}/arp_ignore is used
798 when ARP request is received on the {interface}
800 arp_notify - BOOLEAN
801 Define mode for notification of address and device changes.
802 0 - (default): do nothing
803 1 - Generate gratuitous arp replies when device is brought up
804 or hardware address changes.
806 arp_accept - BOOLEAN
807 Define behavior when gratuitous arp replies are received:
808 0 - drop gratuitous arp frames
809 1 - accept gratuitous arp frames
811 app_solicit - INTEGER
812 The maximum number of probes to send to the user space ARP daemon
813 via netlink before dropping back to multicast probes (see
814 mcast_solicit). Defaults to 0.
816 disable_policy - BOOLEAN
817 Disable IPSEC policy (SPD) for this interface
819 disable_xfrm - BOOLEAN
820 Disable IPSEC encryption on this interface, whatever the policy
824 tag - INTEGER
825 Allows you to write a number, which can be used as required.
826 Default value is 0.
828 Alexey Kuznetsov.
829 kuznet@ms2.inr.ac.ru
831 Updated by:
832 Andi Kleen
833 ak@muc.de
834 Nicolas Delon
835 delon.nicolas@wanadoo.fr
840 /proc/sys/net/ipv6/* Variables:
842 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
843 apply to IPv6 [XXX?].
845 bindv6only - BOOLEAN
846 Default value for IPV6_V6ONLY socket option,
847 which restricts use of the IPv6 socket to IPv6 communication
848 only.
849 TRUE: disable IPv4-mapped address feature
850 FALSE: enable IPv4-mapped address feature
852 Default: FALSE (as specified in RFC2553bis)
854 IPv6 Fragmentation:
856 ip6frag_high_thresh - INTEGER
857 Maximum memory used to reassemble IPv6 fragments. When
858 ip6frag_high_thresh bytes of memory is allocated for this purpose,
859 the fragment handler will toss packets until ip6frag_low_thresh
860 is reached.
862 ip6frag_low_thresh - INTEGER
863 See ip6frag_high_thresh
865 ip6frag_time - INTEGER
866 Time in seconds to keep an IPv6 fragment in memory.
868 ip6frag_secret_interval - INTEGER
869 Regeneration interval (in seconds) of the hash secret (or lifetime
870 for the hash secret) for IPv6 fragments.
871 Default: 600
873 conf/default/*:
874 Change the interface-specific default settings.
877 conf/all/*:
878 Change all the interface-specific settings.
880 [XXX: Other special features than forwarding?]
882 conf/all/forwarding - BOOLEAN
883 Enable global IPv6 forwarding between all interfaces.
885 IPv4 and IPv6 work differently here; e.g. netfilter must be used
886 to control which interfaces may forward packets and which not.
888 This also sets all interfaces' Host/Router setting
889 'forwarding' to the specified value. See below for details.
891 This referred to as global forwarding.
893 proxy_ndp - BOOLEAN
894 Do proxy ndp.
896 conf/interface/*:
897 Change special settings per interface.
899 The functional behaviour for certain settings is different
900 depending on whether local forwarding is enabled or not.
902 accept_ra - BOOLEAN
903 Accept Router Advertisements; autoconfigure using them.
905 Functional default: enabled if local forwarding is disabled.
906 disabled if local forwarding is enabled.
908 accept_ra_defrtr - BOOLEAN
909 Learn default router in Router Advertisement.
911 Functional default: enabled if accept_ra is enabled.
912 disabled if accept_ra is disabled.
914 accept_ra_pinfo - BOOLEAN
915 Learn Prefix Information in Router Advertisement.
917 Functional default: enabled if accept_ra is enabled.
918 disabled if accept_ra is disabled.
920 accept_ra_rt_info_max_plen - INTEGER
921 Maximum prefix length of Route Information in RA.
923 Route Information w/ prefix larger than or equal to this
924 variable shall be ignored.
926 Functional default: 0 if accept_ra_rtr_pref is enabled.
927 -1 if accept_ra_rtr_pref is disabled.
929 accept_ra_rtr_pref - BOOLEAN
930 Accept Router Preference in RA.
932 Functional default: enabled if accept_ra is enabled.
933 disabled if accept_ra is disabled.
935 accept_redirects - BOOLEAN
936 Accept Redirects.
938 Functional default: enabled if local forwarding is disabled.
939 disabled if local forwarding is enabled.
941 accept_source_route - INTEGER
942 Accept source routing (routing extension header).
944 >= 0: Accept only routing header type 2.
945 < 0: Do not accept routing header.
947 Default: 0
949 autoconf - BOOLEAN
950 Autoconfigure addresses using Prefix Information in Router
951 Advertisements.
953 Functional default: enabled if accept_ra_pinfo is enabled.
954 disabled if accept_ra_pinfo is disabled.
956 dad_transmits - INTEGER
957 The amount of Duplicate Address Detection probes to send.
958 Default: 1
960 forwarding - BOOLEAN
961 Configure interface-specific Host/Router behaviour.
963 Note: It is recommended to have the same setting on all
964 interfaces; mixed router/host scenarios are rather uncommon.
966 FALSE:
968 By default, Host behaviour is assumed. This means:
970 1. IsRouter flag is not set in Neighbour Advertisements.
971 2. Router Solicitations are being sent when necessary.
972 3. If accept_ra is TRUE (default), accept Router
973 Advertisements (and do autoconfiguration).
974 4. If accept_redirects is TRUE (default), accept Redirects.
976 TRUE:
978 If local forwarding is enabled, Router behaviour is assumed.
979 This means exactly the reverse from the above:
981 1. IsRouter flag is set in Neighbour Advertisements.
982 2. Router Solicitations are not sent.
983 3. Router Advertisements are ignored.
984 4. Redirects are ignored.
986 Default: FALSE if global forwarding is disabled (default),
987 otherwise TRUE.
989 hop_limit - INTEGER
990 Default Hop Limit to set.
991 Default: 64
993 mtu - INTEGER
994 Default Maximum Transfer Unit
995 Default: 1280 (IPv6 required minimum)
997 router_probe_interval - INTEGER
998 Minimum interval (in seconds) between Router Probing described
999 in RFC4191.
1001 Default: 60
1003 router_solicitation_delay - INTEGER
1004 Number of seconds to wait after interface is brought up
1005 before sending Router Solicitations.
1006 Default: 1
1008 router_solicitation_interval - INTEGER
1009 Number of seconds to wait between Router Solicitations.
1010 Default: 4
1012 router_solicitations - INTEGER
1013 Number of Router Solicitations to send until assuming no
1014 routers are present.
1015 Default: 3
1017 use_tempaddr - INTEGER
1018 Preference for Privacy Extensions (RFC3041).
1019 <= 0 : disable Privacy Extensions
1020 == 1 : enable Privacy Extensions, but prefer public
1021 addresses over temporary addresses.
1022 > 1 : enable Privacy Extensions and prefer temporary
1023 addresses over public addresses.
1024 Default: 0 (for most devices)
1025 -1 (for point-to-point devices and loopback devices)
1027 temp_valid_lft - INTEGER
1028 valid lifetime (in seconds) for temporary addresses.
1029 Default: 604800 (7 days)
1031 temp_prefered_lft - INTEGER
1032 Preferred lifetime (in seconds) for temporary addresses.
1033 Default: 86400 (1 day)
1035 max_desync_factor - INTEGER
1036 Maximum value for DESYNC_FACTOR, which is a random value
1037 that ensures that clients don't synchronize with each
1038 other and generate new addresses at exactly the same time.
1039 value is in seconds.
1040 Default: 600
1042 regen_max_retry - INTEGER
1043 Number of attempts before give up attempting to generate
1044 valid temporary addresses.
1045 Default: 5
1047 max_addresses - INTEGER
1048 Number of maximum addresses per interface. 0 disables limitation.
1049 It is recommended not set too large value (or 0) because it would
1050 be too easy way to crash kernel to allow to create too much of
1051 autoconfigured addresses.
1052 Default: 16
1054 disable_ipv6 - BOOLEAN
1055 Disable IPv6 operation. If accept_dad is set to 2, this value
1056 will be dynamically set to TRUE if DAD fails for the link-local
1057 address.
1058 Default: FALSE (enable IPv6 operation)
1060 When this value is changed from 1 to 0 (IPv6 is being enabled),
1061 it will dynamically create a link-local address on the given
1062 interface and start Duplicate Address Detection, if necessary.
1064 When this value is changed from 0 to 1 (IPv6 is being disabled),
1065 it will dynamically delete all address on the given interface.
1067 accept_dad - INTEGER
1068 Whether to accept DAD (Duplicate Address Detection).
1069 0: Disable DAD
1070 1: Enable DAD (default)
1071 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1072 link-local address has been found.
1074 icmp/*:
1075 ratelimit - INTEGER
1076 Limit the maximal rates for sending ICMPv6 packets.
1077 0 to disable any limiting,
1078 otherwise the minimal space between responses in milliseconds.
1079 Default: 1000
1082 IPv6 Update by:
1083 Pekka Savola <pekkas@netcore.fi>
1084 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1087 /proc/sys/net/bridge/* Variables:
1089 bridge-nf-call-arptables - BOOLEAN
1090 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1091 0 : disable this.
1092 Default: 1
1094 bridge-nf-call-iptables - BOOLEAN
1095 1 : pass bridged IPv4 traffic to iptables' chains.
1096 0 : disable this.
1097 Default: 1
1099 bridge-nf-call-ip6tables - BOOLEAN
1100 1 : pass bridged IPv6 traffic to ip6tables' chains.
1101 0 : disable this.
1102 Default: 1
1104 bridge-nf-filter-vlan-tagged - BOOLEAN
1105 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1106 0 : disable this.
1107 Default: 1
1109 bridge-nf-filter-pppoe-tagged - BOOLEAN
1110 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1111 0 : disable this.
1112 Default: 1
1115 proc/sys/net/sctp/* Variables:
1117 addip_enable - BOOLEAN
1118 Enable or disable extension of Dynamic Address Reconfiguration
1119 (ADD-IP) functionality specified in RFC5061. This extension provides
1120 the ability to dynamically add and remove new addresses for the SCTP
1121 associations.
1123 1: Enable extension.
1125 0: Disable extension.
1127 Default: 0
1129 addip_noauth_enable - BOOLEAN
1130 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1131 authentication to protect the operations of adding or removing new
1132 addresses. This requirement is mandated so that unauthorized hosts
1133 would not be able to hijack associations. However, older
1134 implementations may not have implemented this requirement while
1135 allowing the ADD-IP extension. For reasons of interoperability,
1136 we provide this variable to control the enforcement of the
1137 authentication requirement.
1139 1: Allow ADD-IP extension to be used without authentication. This
1140 should only be set in a closed environment for interoperability
1141 with older implementations.
1143 0: Enforce the authentication requirement
1145 Default: 0
1147 auth_enable - BOOLEAN
1148 Enable or disable Authenticated Chunks extension. This extension
1149 provides the ability to send and receive authenticated chunks and is
1150 required for secure operation of Dynamic Address Reconfiguration
1151 (ADD-IP) extension.
1153 1: Enable this extension.
1154 0: Disable this extension.
1156 Default: 0
1158 prsctp_enable - BOOLEAN
1159 Enable or disable the Partial Reliability extension (RFC3758) which
1160 is used to notify peers that a given DATA should no longer be expected.
1162 1: Enable extension
1163 0: Disable
1165 Default: 1
1167 max_burst - INTEGER
1168 The limit of the number of new packets that can be initially sent. It
1169 controls how bursty the generated traffic can be.
1171 Default: 4
1173 association_max_retrans - INTEGER
1174 Set the maximum number for retransmissions that an association can
1175 attempt deciding that the remote end is unreachable. If this value
1176 is exceeded, the association is terminated.
1178 Default: 10
1180 max_init_retransmits - INTEGER
1181 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1182 that an association will attempt before declaring the destination
1183 unreachable and terminating.
1185 Default: 8
1187 path_max_retrans - INTEGER
1188 The maximum number of retransmissions that will be attempted on a given
1189 path. Once this threshold is exceeded, the path is considered
1190 unreachable, and new traffic will use a different path when the
1191 association is multihomed.
1193 Default: 5
1195 rto_initial - INTEGER
1196 The initial round trip timeout value in milliseconds that will be used
1197 in calculating round trip times. This is the initial time interval
1198 for retransmissions.
1200 Default: 3000
1202 rto_max - INTEGER
1203 The maximum value (in milliseconds) of the round trip timeout. This
1204 is the largest time interval that can elapse between retransmissions.
1206 Default: 60000
1208 rto_min - INTEGER
1209 The minimum value (in milliseconds) of the round trip timeout. This
1210 is the smallest time interval the can elapse between retransmissions.
1212 Default: 1000
1214 hb_interval - INTEGER
1215 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
1216 are sent at the specified interval on idle paths to probe the state of
1217 a given path between 2 associations.
1219 Default: 30000
1221 sack_timeout - INTEGER
1222 The amount of time (in milliseconds) that the implementation will wait
1223 to send a SACK.
1225 Default: 200
1227 valid_cookie_life - INTEGER
1228 The default lifetime of the SCTP cookie (in milliseconds). The cookie
1229 is used during association establishment.
1231 Default: 60000
1233 cookie_preserve_enable - BOOLEAN
1234 Enable or disable the ability to extend the lifetime of the SCTP cookie
1235 that is used during the establishment phase of SCTP association
1237 1: Enable cookie lifetime extension.
1238 0: Disable
1240 Default: 1
1242 rcvbuf_policy - INTEGER
1243 Determines if the receive buffer is attributed to the socket or to
1244 association. SCTP supports the capability to create multiple
1245 associations on a single socket. When using this capability, it is
1246 possible that a single stalled association that's buffering a lot
1247 of data may block other associations from delivering their data by
1248 consuming all of the receive buffer space. To work around this,
1249 the rcvbuf_policy could be set to attribute the receiver buffer space
1250 to each association instead of the socket. This prevents the described
1251 blocking.
1253 1: rcvbuf space is per association
1254 0: recbuf space is per socket
1256 Default: 0
1258 sndbuf_policy - INTEGER
1259 Similar to rcvbuf_policy above, this applies to send buffer space.
1261 1: Send buffer is tracked per association
1262 0: Send buffer is tracked per socket.
1264 Default: 0
1266 sctp_mem - vector of 3 INTEGERs: min, pressure, max
1267 Number of pages allowed for queueing by all SCTP sockets.
1269 min: Below this number of pages SCTP is not bothered about its
1270 memory appetite. When amount of memory allocated by SCTP exceeds
1271 this number, SCTP starts to moderate memory usage.
1273 pressure: This value was introduced to follow format of tcp_mem.
1275 max: Number of pages allowed for queueing by all SCTP sockets.
1277 Default is calculated at boot time from amount of available memory.
1279 sctp_rmem - vector of 3 INTEGERs: min, default, max
1280 See tcp_rmem for a description.
1282 sctp_wmem - vector of 3 INTEGERs: min, default, max
1283 See tcp_wmem for a description.
1286 /proc/sys/net/core/*
1287 dev_weight - INTEGER
1288 The maximum number of packets that kernel can handle on a NAPI
1289 interrupt, it's a Per-CPU variable.
1291 Default: 64
1293 /proc/sys/net/unix/*
1294 max_dgram_qlen - INTEGER
1295 The maximum length of dgram socket receive queue
1297 Default: 10
1300 UNDOCUMENTED:
1302 /proc/sys/net/irda/*
1303 fast_poll_increase FIXME
1304 warn_noreply_time FIXME
1305 discovery_slots FIXME
1306 slot_timeout FIXME
1307 max_baud_rate FIXME
1308 discovery_timeout FIXME
1309 lap_keepalive_time FIXME
1310 max_noreply_time FIXME
1311 max_tx_data_size FIXME
1312 max_tx_window FIXME
1313 min_tx_turn_time FIXME