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 fwmark_reflect - BOOLEAN
60 Controls the fwmark of kernel-generated IPv4 reply packets that are not
61 associated with a socket for example, TCP RSTs or ICMP echo replies).
62 If unset, these packets have a fwmark of zero. If set, they have the
63 fwmark of the packet they are replying to.
66 fib_multipath_use_neigh - BOOLEAN
67 Use status of existing neighbor entry when determining nexthop for
68 multipath routes. If disabled, neighbor information is not used and
69 packets could be directed to a failed nexthop. Only valid for kernels
70 built with CONFIG_IP_ROUTE_MULTIPATH enabled.
76 fib_multipath_hash_policy - INTEGER
77 Controls which hash policy to use for multipath routes. Only valid
78 for kernels built with CONFIG_IP_ROUTE_MULTIPATH enabled.
84 route/max_size - INTEGER
85 Maximum number of routes allowed in the kernel. Increase
86 this when using large numbers of interfaces and/or routes.
87 From linux kernel 3.6 onwards, this is deprecated for ipv4
88 as route cache is no longer used.
90 neigh/default/gc_thresh1 - INTEGER
91 Minimum number of entries to keep. Garbage collector will not
92 purge entries if there are fewer than this number.
95 neigh/default/gc_thresh2 - INTEGER
96 Threshold when garbage collector becomes more aggressive about
97 purging entries. Entries older than 5 seconds will be cleared
98 when over this number.
101 neigh/default/gc_thresh3 - INTEGER
102 Maximum number of neighbor entries allowed. Increase this
103 when using large numbers of interfaces and when communicating
104 with large numbers of directly-connected peers.
107 neigh/default/unres_qlen_bytes - INTEGER
108 The maximum number of bytes which may be used by packets
109 queued for each unresolved address by other network layers.
111 Setting negative value is meaningless and will return error.
112 Default: SK_WMEM_MAX, (same as net.core.wmem_default).
113 Exact value depends on architecture and kernel options,
114 but should be enough to allow queuing 256 packets
117 neigh/default/unres_qlen - INTEGER
118 The maximum number of packets which may be queued for each
119 unresolved address by other network layers.
120 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
121 Prior to linux 3.3, the default value is 3 which may cause
122 unexpected packet loss. The current default value is calculated
123 according to default value of unres_qlen_bytes and true size of
127 mtu_expires - INTEGER
128 Time, in seconds, that cached PMTU information is kept.
130 min_adv_mss - INTEGER
131 The advertised MSS depends on the first hop route MTU, but will
132 never be lower than this setting.
136 ipfrag_high_thresh - LONG INTEGER
137 Maximum memory used to reassemble IP fragments.
139 ipfrag_low_thresh - LONG INTEGER
140 (Obsolete since linux-4.17)
141 Maximum memory used to reassemble IP fragments before the kernel
142 begins to remove incomplete fragment queues to free up resources.
143 The kernel still accepts new fragments for defragmentation.
145 ipfrag_time - INTEGER
146 Time in seconds to keep an IP fragment in memory.
148 ipfrag_max_dist - INTEGER
149 ipfrag_max_dist is a non-negative integer value which defines the
150 maximum "disorder" which is allowed among fragments which share a
151 common IP source address. Note that reordering of packets is
152 not unusual, but if a large number of fragments arrive from a source
153 IP address while a particular fragment queue remains incomplete, it
154 probably indicates that one or more fragments belonging to that queue
155 have been lost. When ipfrag_max_dist is positive, an additional check
156 is done on fragments before they are added to a reassembly queue - if
157 ipfrag_max_dist (or more) fragments have arrived from a particular IP
158 address between additions to any IP fragment queue using that source
159 address, it's presumed that one or more fragments in the queue are
160 lost. The existing fragment queue will be dropped, and a new one
161 started. An ipfrag_max_dist value of zero disables this check.
163 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
164 result in unnecessarily dropping fragment queues when normal
165 reordering of packets occurs, which could lead to poor application
166 performance. Using a very large value, e.g. 50000, increases the
167 likelihood of incorrectly reassembling IP fragments that originate
168 from different IP datagrams, which could result in data corruption.
173 inet_peer_threshold - INTEGER
174 The approximate size of the storage. Starting from this threshold
175 entries will be thrown aggressively. This threshold also determines
176 entries' time-to-live and time intervals between garbage collection
177 passes. More entries, less time-to-live, less GC interval.
179 inet_peer_minttl - INTEGER
180 Minimum time-to-live of entries. Should be enough to cover fragment
181 time-to-live on the reassembling side. This minimum time-to-live is
182 guaranteed if the pool size is less than inet_peer_threshold.
185 inet_peer_maxttl - INTEGER
186 Maximum time-to-live of entries. Unused entries will expire after
187 this period of time if there is no memory pressure on the pool (i.e.
188 when the number of entries in the pool is very small).
194 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
195 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
198 tcp_abort_on_overflow - BOOLEAN
199 If listening service is too slow to accept new connections,
200 reset them. Default state is FALSE. It means that if overflow
201 occurred due to a burst, connection will recover. Enable this
202 option _only_ if you are really sure that listening daemon
203 cannot be tuned to accept connections faster. Enabling this
204 option can harm clients of your server.
206 tcp_adv_win_scale - INTEGER
207 Count buffering overhead as bytes/2^tcp_adv_win_scale
208 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
210 Possible values are [-31, 31], inclusive.
213 tcp_allowed_congestion_control - STRING
214 Show/set the congestion control choices available to non-privileged
215 processes. The list is a subset of those listed in
216 tcp_available_congestion_control.
217 Default is "reno" and the default setting (tcp_congestion_control).
219 tcp_app_win - INTEGER
220 Reserve max(window/2^tcp_app_win, mss) of window for application
221 buffer. Value 0 is special, it means that nothing is reserved.
224 tcp_autocorking - BOOLEAN
225 Enable TCP auto corking :
226 When applications do consecutive small write()/sendmsg() system calls,
227 we try to coalesce these small writes as much as possible, to lower
228 total amount of sent packets. This is done if at least one prior
229 packet for the flow is waiting in Qdisc queues or device transmit
230 queue. Applications can still use TCP_CORK for optimal behavior
231 when they know how/when to uncork their sockets.
234 tcp_available_congestion_control - STRING
235 Shows the available congestion control choices that are registered.
236 More congestion control algorithms may be available as modules,
239 tcp_base_mss - INTEGER
240 The initial value of search_low to be used by the packetization layer
241 Path MTU discovery (MTU probing). If MTU probing is enabled,
242 this is the initial MSS used by the connection.
244 tcp_min_snd_mss - INTEGER
245 TCP SYN and SYNACK messages usually advertise an ADVMSS option,
246 as described in RFC 1122 and RFC 6691.
247 If this ADVMSS option is smaller than tcp_min_snd_mss,
248 it is silently capped to tcp_min_snd_mss.
250 Default : 48 (at least 8 bytes of payload per segment)
252 tcp_congestion_control - STRING
253 Set the congestion control algorithm to be used for new
254 connections. The algorithm "reno" is always available, but
255 additional choices may be available based on kernel configuration.
256 Default is set as part of kernel configuration.
257 For passive connections, the listener congestion control choice
259 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
262 Allows TCP to send "duplicate" SACKs.
264 tcp_early_retrans - INTEGER
265 Tail loss probe (TLP) converts RTOs occurring due to tail
266 losses into fast recovery (draft-ietf-tcpm-rack). Note that
267 TLP requires RACK to function properly (see tcp_recovery below)
274 Control use of Explicit Congestion Notification (ECN) by TCP.
275 ECN is used only when both ends of the TCP connection indicate
276 support for it. This feature is useful in avoiding losses due
277 to congestion by allowing supporting routers to signal
278 congestion before having to drop packets.
280 0 Disable ECN. Neither initiate nor accept ECN.
281 1 Enable ECN when requested by incoming connections and
282 also request ECN on outgoing connection attempts.
283 2 Enable ECN when requested by incoming connections
284 but do not request ECN on outgoing connections.
287 tcp_ecn_fallback - BOOLEAN
288 If the kernel detects that ECN connection misbehaves, enable fall
289 back to non-ECN. Currently, this knob implements the fallback
290 from RFC3168, section 6.1.1.1., but we reserve that in future,
291 additional detection mechanisms could be implemented under this
292 knob. The value is not used, if tcp_ecn or per route (or congestion
293 control) ECN settings are disabled.
294 Default: 1 (fallback enabled)
297 Enable FACK congestion avoidance and fast retransmission.
298 The value is not used, if tcp_sack is not enabled.
300 tcp_fin_timeout - INTEGER
301 The length of time an orphaned (no longer referenced by any
302 application) connection will remain in the FIN_WAIT_2 state
303 before it is aborted at the local end. While a perfectly
304 valid "receive only" state for an un-orphaned connection, an
305 orphaned connection in FIN_WAIT_2 state could otherwise wait
306 forever for the remote to close its end of the connection.
311 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
312 F-RTO is an enhanced recovery algorithm for TCP retransmission
313 timeouts. It is particularly beneficial in networks where the
314 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
315 modification. It does not require any support from the peer.
317 By default it's enabled with a non-zero value. 0 disables F-RTO.
319 tcp_invalid_ratelimit - INTEGER
320 Limit the maximal rate for sending duplicate acknowledgments
321 in response to incoming TCP packets that are for an existing
322 connection but that are invalid due to any of these reasons:
324 (a) out-of-window sequence number,
325 (b) out-of-window acknowledgment number, or
326 (c) PAWS (Protection Against Wrapped Sequence numbers) check failure
328 This can help mitigate simple "ack loop" DoS attacks, wherein
329 a buggy or malicious middlebox or man-in-the-middle can
330 rewrite TCP header fields in manner that causes each endpoint
331 to think that the other is sending invalid TCP segments, thus
332 causing each side to send an unterminating stream of duplicate
333 acknowledgments for invalid segments.
335 Using 0 disables rate-limiting of dupacks in response to
336 invalid segments; otherwise this value specifies the minimal
337 space between sending such dupacks, in milliseconds.
339 Default: 500 (milliseconds).
341 tcp_keepalive_time - INTEGER
342 How often TCP sends out keepalive messages when keepalive is enabled.
345 tcp_keepalive_probes - INTEGER
346 How many keepalive probes TCP sends out, until it decides that the
347 connection is broken. Default value: 9.
349 tcp_keepalive_intvl - INTEGER
350 How frequently the probes are send out. Multiplied by
351 tcp_keepalive_probes it is time to kill not responding connection,
352 after probes started. Default value: 75sec i.e. connection
353 will be aborted after ~11 minutes of retries.
355 tcp_l3mdev_accept - BOOLEAN
356 Enables child sockets to inherit the L3 master device index.
357 Enabling this option allows a "global" listen socket to work
358 across L3 master domains (e.g., VRFs) with connected sockets
359 derived from the listen socket to be bound to the L3 domain in
360 which the packets originated. Only valid when the kernel was
361 compiled with CONFIG_NET_L3_MASTER_DEV.
363 tcp_low_latency - BOOLEAN
364 This is a legacy option, it has no effect anymore.
366 tcp_max_orphans - INTEGER
367 Maximal number of TCP sockets not attached to any user file handle,
368 held by system. If this number is exceeded orphaned connections are
369 reset immediately and warning is printed. This limit exists
370 only to prevent simple DoS attacks, you _must_ not rely on this
371 or lower the limit artificially, but rather increase it
372 (probably, after increasing installed memory),
373 if network conditions require more than default value,
374 and tune network services to linger and kill such states
375 more aggressively. Let me to remind again: each orphan eats
376 up to ~64K of unswappable memory.
378 tcp_max_syn_backlog - INTEGER
379 Maximal number of remembered connection requests, which have not
380 received an acknowledgment from connecting client.
381 The minimal value is 128 for low memory machines, and it will
382 increase in proportion to the memory of machine.
383 If server suffers from overload, try increasing this number.
385 tcp_max_tw_buckets - INTEGER
386 Maximal number of timewait sockets held by system simultaneously.
387 If this number is exceeded time-wait socket is immediately destroyed
388 and warning is printed. This limit exists only to prevent
389 simple DoS attacks, you _must_ not lower the limit artificially,
390 but rather increase it (probably, after increasing installed memory),
391 if network conditions require more than default value.
393 tcp_mem - vector of 3 INTEGERs: min, pressure, max
394 min: below this number of pages TCP is not bothered about its
397 pressure: when amount of memory allocated by TCP exceeds this number
398 of pages, TCP moderates its memory consumption and enters memory
399 pressure mode, which is exited when memory consumption falls
402 max: number of pages allowed for queueing by all TCP sockets.
404 Defaults are calculated at boot time from amount of available
407 tcp_min_rtt_wlen - INTEGER
408 The window length of the windowed min filter to track the minimum RTT.
409 A shorter window lets a flow more quickly pick up new (higher)
410 minimum RTT when it is moved to a longer path (e.g., due to traffic
411 engineering). A longer window makes the filter more resistant to RTT
412 inflations such as transient congestion. The unit is seconds.
413 Possible values: 0 - 86400 (1 day)
416 tcp_moderate_rcvbuf - BOOLEAN
417 If set, TCP performs receive buffer auto-tuning, attempting to
418 automatically size the buffer (no greater than tcp_rmem[2]) to
419 match the size required by the path for full throughput. Enabled by
422 tcp_mtu_probing - INTEGER
423 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
426 1 - Disabled by default, enabled when an ICMP black hole detected
427 2 - Always enabled, use initial MSS of tcp_base_mss.
429 tcp_probe_interval - INTEGER
430 Controls how often to start TCP Packetization-Layer Path MTU
431 Discovery reprobe. The default is reprobing every 10 minutes as
434 tcp_probe_threshold - INTEGER
435 Controls when TCP Packetization-Layer Path MTU Discovery probing
436 will stop in respect to the width of search range in bytes. Default
439 tcp_no_metrics_save - BOOLEAN
440 By default, TCP saves various connection metrics in the route cache
441 when the connection closes, so that connections established in the
442 near future can use these to set initial conditions. Usually, this
443 increases overall performance, but may sometimes cause performance
444 degradation. If set, TCP will not cache metrics on closing
447 tcp_orphan_retries - INTEGER
448 This value influences the timeout of a locally closed TCP connection,
449 when RTO retransmissions remain unacknowledged.
450 See tcp_retries2 for more details.
452 The default value is 8.
453 If your machine is a loaded WEB server,
454 you should think about lowering this value, such sockets
455 may consume significant resources. Cf. tcp_max_orphans.
457 tcp_recovery - INTEGER
458 This value is a bitmap to enable various experimental loss recovery
461 RACK: 0x1 enables the RACK loss detection for fast detection of lost
462 retransmissions and tail drops.
466 tcp_reordering - INTEGER
467 Initial reordering level of packets in a TCP stream.
468 TCP stack can then dynamically adjust flow reordering level
469 between this initial value and tcp_max_reordering
472 tcp_max_reordering - INTEGER
473 Maximal reordering level of packets in a TCP stream.
474 300 is a fairly conservative value, but you might increase it
475 if paths are using per packet load balancing (like bonding rr mode)
478 tcp_retrans_collapse - BOOLEAN
479 Bug-to-bug compatibility with some broken printers.
480 On retransmit try to send bigger packets to work around bugs in
483 tcp_retries1 - INTEGER
484 This value influences the time, after which TCP decides, that
485 something is wrong due to unacknowledged RTO retransmissions,
486 and reports this suspicion to the network layer.
487 See tcp_retries2 for more details.
489 RFC 1122 recommends at least 3 retransmissions, which is the
492 tcp_retries2 - INTEGER
493 This value influences the timeout of an alive TCP connection,
494 when RTO retransmissions remain unacknowledged.
495 Given a value of N, a hypothetical TCP connection following
496 exponential backoff with an initial RTO of TCP_RTO_MIN would
497 retransmit N times before killing the connection at the (N+1)th RTO.
499 The default value of 15 yields a hypothetical timeout of 924.6
500 seconds and is a lower bound for the effective timeout.
501 TCP will effectively time out at the first RTO which exceeds the
502 hypothetical timeout.
504 RFC 1122 recommends at least 100 seconds for the timeout,
505 which corresponds to a value of at least 8.
507 tcp_rfc1337 - BOOLEAN
508 If set, the TCP stack behaves conforming to RFC1337. If unset,
509 we are not conforming to RFC, but prevent TCP TIME_WAIT
513 tcp_rmem - vector of 3 INTEGERs: min, default, max
514 min: Minimal size of receive buffer used by TCP sockets.
515 It is guaranteed to each TCP socket, even under moderate memory
519 default: initial size of receive buffer used by TCP sockets.
520 This value overrides net.core.rmem_default used by other protocols.
521 Default: 87380 bytes. This value results in window of 65535 with
522 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
523 less for default tcp_app_win. See below about these variables.
525 max: maximal size of receive buffer allowed for automatically
526 selected receiver buffers for TCP socket. This value does not override
527 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
528 automatic tuning of that socket's receive buffer size, in which
529 case this value is ignored.
530 Default: between 87380B and 6MB, depending on RAM size.
533 Enable select acknowledgments (SACKS).
535 tcp_slow_start_after_idle - BOOLEAN
536 If set, provide RFC2861 behavior and time out the congestion
537 window after an idle period. An idle period is defined at
538 the current RTO. If unset, the congestion window will not
539 be timed out after an idle period.
543 Use the Host requirements interpretation of the TCP urgent pointer field.
544 Most hosts use the older BSD interpretation, so if you turn this on
545 Linux might not communicate correctly with them.
548 tcp_synack_retries - INTEGER
549 Number of times SYNACKs for a passive TCP connection attempt will
550 be retransmitted. Should not be higher than 255. Default value
551 is 5, which corresponds to 31seconds till the last retransmission
552 with the current initial RTO of 1second. With this the final timeout
553 for a passive TCP connection will happen after 63seconds.
555 tcp_syncookies - BOOLEAN
556 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
557 Send out syncookies when the syn backlog queue of a socket
558 overflows. This is to prevent against the common 'SYN flood attack'
561 Note, that syncookies is fallback facility.
562 It MUST NOT be used to help highly loaded servers to stand
563 against legal connection rate. If you see SYN flood warnings
564 in your logs, but investigation shows that they occur
565 because of overload with legal connections, you should tune
566 another parameters until this warning disappear.
567 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
569 syncookies seriously violate TCP protocol, do not allow
570 to use TCP extensions, can result in serious degradation
571 of some services (f.e. SMTP relaying), visible not by you,
572 but your clients and relays, contacting you. While you see
573 SYN flood warnings in logs not being really flooded, your server
574 is seriously misconfigured.
576 If you want to test which effects syncookies have to your
577 network connections you can set this knob to 2 to enable
578 unconditionally generation of syncookies.
580 tcp_fastopen - INTEGER
581 Enable TCP Fast Open (RFC7413) to send and accept data in the opening
584 The client support is enabled by flag 0x1 (on by default). The client
585 then must use sendmsg() or sendto() with the MSG_FASTOPEN flag,
586 rather than connect() to send data in SYN.
588 The server support is enabled by flag 0x2 (off by default). Then
589 either enable for all listeners with another flag (0x400) or
590 enable individual listeners via TCP_FASTOPEN socket option with
591 the option value being the length of the syn-data backlog.
593 The values (bitmap) are
594 0x1: (client) enables sending data in the opening SYN on the client.
595 0x2: (server) enables the server support, i.e., allowing data in
596 a SYN packet to be accepted and passed to the
597 application before 3-way handshake finishes.
598 0x4: (client) send data in the opening SYN regardless of cookie
599 availability and without a cookie option.
600 0x200: (server) accept data-in-SYN w/o any cookie option present.
601 0x400: (server) enable all listeners to support Fast Open by
602 default without explicit TCP_FASTOPEN socket option.
606 Note that that additional client or server features are only
607 effective if the basic support (0x1 and 0x2) are enabled respectively.
609 tcp_fastopen_blackhole_timeout_sec - INTEGER
610 Initial time period in second to disable Fastopen on active TCP sockets
611 when a TFO firewall blackhole issue happens.
612 This time period will grow exponentially when more blackhole issues
613 get detected right after Fastopen is re-enabled and will reset to
614 initial value when the blackhole issue goes away.
615 By default, it is set to 1hr.
617 tcp_syn_retries - INTEGER
618 Number of times initial SYNs for an active TCP connection attempt
619 will be retransmitted. Should not be higher than 127. Default value
620 is 6, which corresponds to 63seconds till the last retransmission
621 with the current initial RTO of 1second. With this the final timeout
622 for an active TCP connection attempt will happen after 127seconds.
624 tcp_timestamps - INTEGER
625 Enable timestamps as defined in RFC1323.
627 1: Enable timestamps as defined in RFC1323 and use random offset for
628 each connection rather than only using the current time.
629 2: Like 1, but without random offsets.
632 tcp_min_tso_segs - INTEGER
633 Minimal number of segments per TSO frame.
634 Since linux-3.12, TCP does an automatic sizing of TSO frames,
635 depending on flow rate, instead of filling 64Kbytes packets.
636 For specific usages, it's possible to force TCP to build big
637 TSO frames. Note that TCP stack might split too big TSO packets
638 if available window is too small.
641 tcp_pacing_ss_ratio - INTEGER
642 sk->sk_pacing_rate is set by TCP stack using a ratio applied
643 to current rate. (current_rate = cwnd * mss / srtt)
644 If TCP is in slow start, tcp_pacing_ss_ratio is applied
645 to let TCP probe for bigger speeds, assuming cwnd can be
646 doubled every other RTT.
649 tcp_pacing_ca_ratio - INTEGER
650 sk->sk_pacing_rate is set by TCP stack using a ratio applied
651 to current rate. (current_rate = cwnd * mss / srtt)
652 If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio
653 is applied to conservatively probe for bigger throughput.
656 tcp_tso_win_divisor - INTEGER
657 This allows control over what percentage of the congestion window
658 can be consumed by a single TSO frame.
659 The setting of this parameter is a choice between burstiness and
660 building larger TSO frames.
663 tcp_tw_reuse - BOOLEAN
664 Allow to reuse TIME-WAIT sockets for new connections when it is
665 safe from protocol viewpoint. Default value is 0.
666 It should not be changed without advice/request of technical
669 tcp_window_scaling - BOOLEAN
670 Enable window scaling as defined in RFC1323.
672 tcp_wmem - vector of 3 INTEGERs: min, default, max
673 min: Amount of memory reserved for send buffers for TCP sockets.
674 Each TCP socket has rights to use it due to fact of its birth.
677 default: initial size of send buffer used by TCP sockets. This
678 value overrides net.core.wmem_default used by other protocols.
679 It is usually lower than net.core.wmem_default.
682 max: Maximal amount of memory allowed for automatically tuned
683 send buffers for TCP sockets. This value does not override
684 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
685 automatic tuning of that socket's send buffer size, in which case
686 this value is ignored.
687 Default: between 64K and 4MB, depending on RAM size.
689 tcp_notsent_lowat - UNSIGNED INTEGER
690 A TCP socket can control the amount of unsent bytes in its write queue,
691 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
692 reports POLLOUT events if the amount of unsent bytes is below a per
693 socket value, and if the write queue is not full. sendmsg() will
694 also not add new buffers if the limit is hit.
696 This global variable controls the amount of unsent data for
697 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
698 to the global variable has immediate effect.
700 Default: UINT_MAX (0xFFFFFFFF)
702 tcp_workaround_signed_windows - BOOLEAN
703 If set, assume no receipt of a window scaling option means the
704 remote TCP is broken and treats the window as a signed quantity.
705 If unset, assume the remote TCP is not broken even if we do
706 not receive a window scaling option from them.
709 tcp_thin_linear_timeouts - BOOLEAN
710 Enable dynamic triggering of linear timeouts for thin streams.
711 If set, a check is performed upon retransmission by timeout to
712 determine if the stream is thin (less than 4 packets in flight).
713 As long as the stream is found to be thin, up to 6 linear
714 timeouts may be performed before exponential backoff mode is
715 initiated. This improves retransmission latency for
716 non-aggressive thin streams, often found to be time-dependent.
717 For more information on thin streams, see
718 Documentation/networking/tcp-thin.txt
721 tcp_limit_output_bytes - INTEGER
722 Controls TCP Small Queue limit per tcp socket.
723 TCP bulk sender tends to increase packets in flight until it
724 gets losses notifications. With SNDBUF autotuning, this can
725 result in a large amount of packets queued in qdisc/device
726 on the local machine, hurting latency of other flows, for
727 typical pfifo_fast qdiscs.
728 tcp_limit_output_bytes limits the number of bytes on qdisc
729 or device to reduce artificial RTT/cwnd and reduce bufferbloat.
732 tcp_challenge_ack_limit - INTEGER
733 Limits number of Challenge ACK sent per second, as recommended
734 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
739 udp_l3mdev_accept - BOOLEAN
740 Enabling this option allows a "global" bound socket to work
741 across L3 master domains (e.g., VRFs) with packets capable of
742 being received regardless of the L3 domain in which they
743 originated. Only valid when the kernel was compiled with
744 CONFIG_NET_L3_MASTER_DEV.
746 udp_mem - vector of 3 INTEGERs: min, pressure, max
747 Number of pages allowed for queueing by all UDP sockets.
749 min: Below this number of pages UDP is not bothered about its
750 memory appetite. When amount of memory allocated by UDP exceeds
751 this number, UDP starts to moderate memory usage.
753 pressure: This value was introduced to follow format of tcp_mem.
755 max: Number of pages allowed for queueing by all UDP sockets.
757 Default is calculated at boot time from amount of available memory.
759 udp_rmem_min - INTEGER
760 Minimal size of receive buffer used by UDP sockets in moderation.
761 Each UDP socket is able to use the size for receiving data, even if
762 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
765 udp_wmem_min - INTEGER
766 Minimal size of send buffer used by UDP sockets in moderation.
767 Each UDP socket is able to use the size for sending data, even if
768 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
773 cipso_cache_enable - BOOLEAN
774 If set, enable additions to and lookups from the CIPSO label mapping
775 cache. If unset, additions are ignored and lookups always result in a
776 miss. However, regardless of the setting the cache is still
777 invalidated when required when means you can safely toggle this on and
778 off and the cache will always be "safe".
781 cipso_cache_bucket_size - INTEGER
782 The CIPSO label cache consists of a fixed size hash table with each
783 hash bucket containing a number of cache entries. This variable limits
784 the number of entries in each hash bucket; the larger the value is, the
785 more CIPSO label mappings that can be cached. When the number of
786 entries in a given hash bucket reaches this limit adding new entries
787 causes the oldest entry in the bucket to be removed to make room.
790 cipso_rbm_optfmt - BOOLEAN
791 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
792 the CIPSO draft specification (see Documentation/netlabel for details).
793 This means that when set the CIPSO tag will be padded with empty
794 categories in order to make the packet data 32-bit aligned.
797 cipso_rbm_structvalid - BOOLEAN
798 If set, do a very strict check of the CIPSO option when
799 ip_options_compile() is called. If unset, relax the checks done during
800 ip_options_compile(). Either way is "safe" as errors are caught else
801 where in the CIPSO processing code but setting this to 0 (False) should
802 result in less work (i.e. it should be faster) but could cause problems
803 with other implementations that require strict checking.
808 ip_local_port_range - 2 INTEGERS
809 Defines the local port range that is used by TCP and UDP to
810 choose the local port. The first number is the first, the
811 second the last local port number.
812 If possible, it is better these numbers have different parity.
813 (one even and one odd values)
814 The default values are 32768 and 60999 respectively.
816 ip_local_reserved_ports - list of comma separated ranges
817 Specify the ports which are reserved for known third-party
818 applications. These ports will not be used by automatic port
819 assignments (e.g. when calling connect() or bind() with port
820 number 0). Explicit port allocation behavior is unchanged.
822 The format used for both input and output is a comma separated
823 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
824 10). Writing to the file will clear all previously reserved
825 ports and update the current list with the one given in the
828 Note that ip_local_port_range and ip_local_reserved_ports
829 settings are independent and both are considered by the kernel
830 when determining which ports are available for automatic port
833 You can reserve ports which are not in the current
834 ip_local_port_range, e.g.:
836 $ cat /proc/sys/net/ipv4/ip_local_port_range
838 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
841 although this is redundant. However such a setting is useful
842 if later the port range is changed to a value that will
843 include the reserved ports.
847 ip_unprivileged_port_start - INTEGER
848 This is a per-namespace sysctl. It defines the first
849 unprivileged port in the network namespace. Privileged ports
850 require root or CAP_NET_BIND_SERVICE in order to bind to them.
851 To disable all privileged ports, set this to 0. It may not
852 overlap with the ip_local_reserved_ports range.
856 ip_nonlocal_bind - BOOLEAN
857 If set, allows processes to bind() to non-local IP addresses,
858 which can be quite useful - but may break some applications.
862 If set non-zero, enables support for dynamic addresses.
863 If set to a non-zero value larger than 1, a kernel log
864 message will be printed when dynamic address rewriting
868 ip_early_demux - BOOLEAN
869 Optimize input packet processing down to one demux for
870 certain kinds of local sockets. Currently we only do this
871 for established TCP and connected UDP sockets.
873 It may add an additional cost for pure routing workloads that
874 reduces overall throughput, in such case you should disable it.
877 tcp_early_demux - BOOLEAN
878 Enable early demux for established TCP sockets.
881 udp_early_demux - BOOLEAN
882 Enable early demux for connected UDP sockets. Disable this if
883 your system could experience more unconnected load.
886 icmp_echo_ignore_all - BOOLEAN
887 If set non-zero, then the kernel will ignore all ICMP ECHO
891 icmp_echo_ignore_broadcasts - BOOLEAN
892 If set non-zero, then the kernel will ignore all ICMP ECHO and
893 TIMESTAMP requests sent to it via broadcast/multicast.
896 icmp_ratelimit - INTEGER
897 Limit the maximal rates for sending ICMP packets whose type matches
898 icmp_ratemask (see below) to specific targets.
899 0 to disable any limiting,
900 otherwise the minimal space between responses in milliseconds.
901 Note that another sysctl, icmp_msgs_per_sec limits the number
902 of ICMP packets sent on all targets.
905 icmp_msgs_per_sec - INTEGER
906 Limit maximal number of ICMP packets sent per second from this host.
907 Only messages whose type matches icmp_ratemask (see below) are
908 controlled by this limit. For security reasons, the precise count
909 of messages per second is randomized.
912 icmp_msgs_burst - INTEGER
913 icmp_msgs_per_sec controls number of ICMP packets sent per second,
914 while icmp_msgs_burst controls the burst size of these packets.
915 For security reasons, the precise burst size is randomized.
918 icmp_ratemask - INTEGER
919 Mask made of ICMP types for which rates are being limited.
920 Significant bits: IHGFEDCBA9876543210
921 Default mask: 0000001100000011000 (6168)
923 Bit definitions (see include/linux/icmp.h):
925 3 Destination Unreachable *
930 C Parameter Problem *
935 H Address Mask Request
938 * These are rate limited by default (see default mask above)
940 icmp_ignore_bogus_error_responses - BOOLEAN
941 Some routers violate RFC1122 by sending bogus responses to broadcast
942 frames. Such violations are normally logged via a kernel warning.
943 If this is set to TRUE, the kernel will not give such warnings, which
944 will avoid log file clutter.
947 icmp_errors_use_inbound_ifaddr - BOOLEAN
949 If zero, icmp error messages are sent with the primary address of
950 the exiting interface.
952 If non-zero, the message will be sent with the primary address of
953 the interface that received the packet that caused the icmp error.
954 This is the behaviour network many administrators will expect from
955 a router. And it can make debugging complicated network layouts
958 Note that if no primary address exists for the interface selected,
959 then the primary address of the first non-loopback interface that
960 has one will be used regardless of this setting.
964 igmp_max_memberships - INTEGER
965 Change the maximum number of multicast groups we can subscribe to.
968 Theoretical maximum value is bounded by having to send a membership
969 report in a single datagram (i.e. the report can't span multiple
970 datagrams, or risk confusing the switch and leaving groups you don't
973 The number of supported groups 'M' is bounded by the number of group
974 report entries you can fit into a single datagram of 65535 bytes.
976 M = 65536-sizeof (ip header)/(sizeof(Group record))
978 Group records are variable length, with a minimum of 12 bytes.
979 So net.ipv4.igmp_max_memberships should not be set higher than:
981 (65536-24) / 12 = 5459
983 The value 5459 assumes no IP header options, so in practice
984 this number may be lower.
986 igmp_max_msf - INTEGER
987 Maximum number of addresses allowed in the source filter list for a
992 Controls the IGMP query robustness variable (see RFC2236 8.1).
993 Default: 2 (as specified by RFC2236 8.1)
994 Minimum: 1 (as specified by RFC6636 4.5)
996 force_igmp_version - INTEGER
997 0 - (default) No enforcement of a IGMP version, IGMPv1/v2 fallback
998 allowed. Will back to IGMPv3 mode again if all IGMPv1/v2 Querier
999 Present timer expires.
1000 1 - Enforce to use IGMP version 1. Will also reply IGMPv1 report if
1001 receive IGMPv2/v3 query.
1002 2 - Enforce to use IGMP version 2. Will fallback to IGMPv1 if receive
1003 IGMPv1 query message. Will reply report if receive IGMPv3 query.
1004 3 - Enforce to use IGMP version 3. The same react with default 0.
1006 Note: this is not the same with force_mld_version because IGMPv3 RFC3376
1007 Security Considerations does not have clear description that we could
1008 ignore other version messages completely as MLDv2 RFC3810. So make
1009 this value as default 0 is recommended.
1011 conf/interface/* changes special settings per interface (where
1012 "interface" is the name of your network interface)
1014 conf/all/* is special, changes the settings for all interfaces
1016 log_martians - BOOLEAN
1017 Log packets with impossible addresses to kernel log.
1018 log_martians for the interface will be enabled if at least one of
1019 conf/{all,interface}/log_martians is set to TRUE,
1020 it will be disabled otherwise
1022 accept_redirects - BOOLEAN
1023 Accept ICMP redirect messages.
1024 accept_redirects for the interface will be enabled if:
1025 - both conf/{all,interface}/accept_redirects are TRUE in the case
1026 forwarding for the interface is enabled
1028 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
1029 case forwarding for the interface is disabled
1030 accept_redirects for the interface will be disabled otherwise
1034 forwarding - BOOLEAN
1035 Enable IP forwarding on this interface. This controls whether packets
1036 received _on_ this interface can be forwarded.
1038 mc_forwarding - BOOLEAN
1039 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
1040 and a multicast routing daemon is required.
1041 conf/all/mc_forwarding must also be set to TRUE to enable multicast
1042 routing for the interface
1045 Integer value used to differentiate the devices by the medium they
1046 are attached to. Two devices can have different id values when
1047 the broadcast packets are received only on one of them.
1048 The default value 0 means that the device is the only interface
1049 to its medium, value of -1 means that medium is not known.
1051 Currently, it is used to change the proxy_arp behavior:
1052 the proxy_arp feature is enabled for packets forwarded between
1053 two devices attached to different media.
1057 proxy_arp for the interface will be enabled if at least one of
1058 conf/{all,interface}/proxy_arp is set to TRUE,
1059 it will be disabled otherwise
1061 proxy_arp_pvlan - BOOLEAN
1062 Private VLAN proxy arp.
1063 Basically allow proxy arp replies back to the same interface
1064 (from which the ARP request/solicitation was received).
1066 This is done to support (ethernet) switch features, like RFC
1067 3069, where the individual ports are NOT allowed to
1068 communicate with each other, but they are allowed to talk to
1069 the upstream router. As described in RFC 3069, it is possible
1070 to allow these hosts to communicate through the upstream
1071 router by proxy_arp'ing. Don't need to be used together with
1074 This technology is known by different names:
1075 In RFC 3069 it is called VLAN Aggregation.
1076 Cisco and Allied Telesyn call it Private VLAN.
1077 Hewlett-Packard call it Source-Port filtering or port-isolation.
1078 Ericsson call it MAC-Forced Forwarding (RFC Draft).
1080 shared_media - BOOLEAN
1081 Send(router) or accept(host) RFC1620 shared media redirects.
1082 Overrides secure_redirects.
1083 shared_media for the interface will be enabled if at least one of
1084 conf/{all,interface}/shared_media is set to TRUE,
1085 it will be disabled otherwise
1088 secure_redirects - BOOLEAN
1089 Accept ICMP redirect messages only to gateways listed in the
1090 interface's current gateway list. Even if disabled, RFC1122 redirect
1092 Overridden by shared_media.
1093 secure_redirects for the interface will be enabled if at least one of
1094 conf/{all,interface}/secure_redirects is set to TRUE,
1095 it will be disabled otherwise
1098 send_redirects - BOOLEAN
1099 Send redirects, if router.
1100 send_redirects for the interface will be enabled if at least one of
1101 conf/{all,interface}/send_redirects is set to TRUE,
1102 it will be disabled otherwise
1105 bootp_relay - BOOLEAN
1106 Accept packets with source address 0.b.c.d destined
1107 not to this host as local ones. It is supposed, that
1108 BOOTP relay daemon will catch and forward such packets.
1109 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
1112 Not Implemented Yet.
1114 accept_source_route - BOOLEAN
1115 Accept packets with SRR option.
1116 conf/all/accept_source_route must also be set to TRUE to accept packets
1117 with SRR option on the interface
1118 default TRUE (router)
1121 accept_local - BOOLEAN
1122 Accept packets with local source addresses. In combination with
1123 suitable routing, this can be used to direct packets between two
1124 local interfaces over the wire and have them accepted properly.
1127 route_localnet - BOOLEAN
1128 Do not consider loopback addresses as martian source or destination
1129 while routing. This enables the use of 127/8 for local routing purposes.
1133 0 - No source validation.
1134 1 - Strict mode as defined in RFC3704 Strict Reverse Path
1135 Each incoming packet is tested against the FIB and if the interface
1136 is not the best reverse path the packet check will fail.
1137 By default failed packets are discarded.
1138 2 - Loose mode as defined in RFC3704 Loose Reverse Path
1139 Each incoming packet's source address is also tested against the FIB
1140 and if the source address is not reachable via any interface
1141 the packet check will fail.
1143 Current recommended practice in RFC3704 is to enable strict mode
1144 to prevent IP spoofing from DDos attacks. If using asymmetric routing
1145 or other complicated routing, then loose mode is recommended.
1147 The max value from conf/{all,interface}/rp_filter is used
1148 when doing source validation on the {interface}.
1150 Default value is 0. Note that some distributions enable it
1153 arp_filter - BOOLEAN
1154 1 - Allows you to have multiple network interfaces on the same
1155 subnet, and have the ARPs for each interface be answered
1156 based on whether or not the kernel would route a packet from
1157 the ARP'd IP out that interface (therefore you must use source
1158 based routing for this to work). In other words it allows control
1159 of which cards (usually 1) will respond to an arp request.
1161 0 - (default) The kernel can respond to arp requests with addresses
1162 from other interfaces. This may seem wrong but it usually makes
1163 sense, because it increases the chance of successful communication.
1164 IP addresses are owned by the complete host on Linux, not by
1165 particular interfaces. Only for more complex setups like load-
1166 balancing, does this behaviour cause problems.
1168 arp_filter for the interface will be enabled if at least one of
1169 conf/{all,interface}/arp_filter is set to TRUE,
1170 it will be disabled otherwise
1172 arp_announce - INTEGER
1173 Define different restriction levels for announcing the local
1174 source IP address from IP packets in ARP requests sent on
1176 0 - (default) Use any local address, configured on any interface
1177 1 - Try to avoid local addresses that are not in the target's
1178 subnet for this interface. This mode is useful when target
1179 hosts reachable via this interface require the source IP
1180 address in ARP requests to be part of their logical network
1181 configured on the receiving interface. When we generate the
1182 request we will check all our subnets that include the
1183 target IP and will preserve the source address if it is from
1184 such subnet. If there is no such subnet we select source
1185 address according to the rules for level 2.
1186 2 - Always use the best local address for this target.
1187 In this mode we ignore the source address in the IP packet
1188 and try to select local address that we prefer for talks with
1189 the target host. Such local address is selected by looking
1190 for primary IP addresses on all our subnets on the outgoing
1191 interface that include the target IP address. If no suitable
1192 local address is found we select the first local address
1193 we have on the outgoing interface or on all other interfaces,
1194 with the hope we will receive reply for our request and
1195 even sometimes no matter the source IP address we announce.
1197 The max value from conf/{all,interface}/arp_announce is used.
1199 Increasing the restriction level gives more chance for
1200 receiving answer from the resolved target while decreasing
1201 the level announces more valid sender's information.
1203 arp_ignore - INTEGER
1204 Define different modes for sending replies in response to
1205 received ARP requests that resolve local target IP addresses:
1206 0 - (default): reply for any local target IP address, configured
1208 1 - reply only if the target IP address is local address
1209 configured on the incoming interface
1210 2 - reply only if the target IP address is local address
1211 configured on the incoming interface and both with the
1212 sender's IP address are part from same subnet on this interface
1213 3 - do not reply for local addresses configured with scope host,
1214 only resolutions for global and link addresses are replied
1216 8 - do not reply for all local addresses
1218 The max value from conf/{all,interface}/arp_ignore is used
1219 when ARP request is received on the {interface}
1221 arp_notify - BOOLEAN
1222 Define mode for notification of address and device changes.
1223 0 - (default): do nothing
1224 1 - Generate gratuitous arp requests when device is brought up
1225 or hardware address changes.
1227 arp_accept - BOOLEAN
1228 Define behavior for gratuitous ARP frames who's IP is not
1229 already present in the ARP table:
1230 0 - don't create new entries in the ARP table
1231 1 - create new entries in the ARP table
1233 Both replies and requests type gratuitous arp will trigger the
1234 ARP table to be updated, if this setting is on.
1236 If the ARP table already contains the IP address of the
1237 gratuitous arp frame, the arp table will be updated regardless
1238 if this setting is on or off.
1240 mcast_solicit - INTEGER
1241 The maximum number of multicast probes in INCOMPLETE state,
1242 when the associated hardware address is unknown. Defaults
1245 ucast_solicit - INTEGER
1246 The maximum number of unicast probes in PROBE state, when
1247 the hardware address is being reconfirmed. Defaults to 3.
1249 app_solicit - INTEGER
1250 The maximum number of probes to send to the user space ARP daemon
1251 via netlink before dropping back to multicast probes (see
1252 mcast_resolicit). Defaults to 0.
1254 mcast_resolicit - INTEGER
1255 The maximum number of multicast probes after unicast and
1256 app probes in PROBE state. Defaults to 0.
1258 disable_policy - BOOLEAN
1259 Disable IPSEC policy (SPD) for this interface
1261 disable_xfrm - BOOLEAN
1262 Disable IPSEC encryption on this interface, whatever the policy
1264 igmpv2_unsolicited_report_interval - INTEGER
1265 The interval in milliseconds in which the next unsolicited
1266 IGMPv1 or IGMPv2 report retransmit will take place.
1267 Default: 10000 (10 seconds)
1269 igmpv3_unsolicited_report_interval - INTEGER
1270 The interval in milliseconds in which the next unsolicited
1271 IGMPv3 report retransmit will take place.
1272 Default: 1000 (1 seconds)
1274 promote_secondaries - BOOLEAN
1275 When a primary IP address is removed from this interface
1276 promote a corresponding secondary IP address instead of
1277 removing all the corresponding secondary IP addresses.
1279 drop_unicast_in_l2_multicast - BOOLEAN
1280 Drop any unicast IP packets that are received in link-layer
1281 multicast (or broadcast) frames.
1282 This behavior (for multicast) is actually a SHOULD in RFC
1283 1122, but is disabled by default for compatibility reasons.
1286 drop_gratuitous_arp - BOOLEAN
1287 Drop all gratuitous ARP frames, for example if there's a known
1288 good ARP proxy on the network and such frames need not be used
1289 (or in the case of 802.11, must not be used to prevent attacks.)
1294 Allows you to write a number, which can be used as required.
1297 xfrm4_gc_thresh - INTEGER
1298 The threshold at which we will start garbage collecting for IPv4
1299 destination cache entries. At twice this value the system will
1300 refuse new allocations.
1302 igmp_link_local_mcast_reports - BOOLEAN
1303 Enable IGMP reports for link local multicast groups in the
1308 kuznet@ms2.inr.ac.ru
1314 delon.nicolas@wanadoo.fr
1319 /proc/sys/net/ipv6/* Variables:
1321 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1322 apply to IPv6 [XXX?].
1324 bindv6only - BOOLEAN
1325 Default value for IPV6_V6ONLY socket option,
1326 which restricts use of the IPv6 socket to IPv6 communication
1328 TRUE: disable IPv4-mapped address feature
1329 FALSE: enable IPv4-mapped address feature
1331 Default: FALSE (as specified in RFC3493)
1333 flowlabel_consistency - BOOLEAN
1334 Protect the consistency (and unicity) of flow label.
1335 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1341 auto_flowlabels - INTEGER
1342 Automatically generate flow labels based on a flow hash of the
1343 packet. This allows intermediate devices, such as routers, to
1344 identify packet flows for mechanisms like Equal Cost Multipath
1345 Routing (see RFC 6438).
1346 0: automatic flow labels are completely disabled
1347 1: automatic flow labels are enabled by default, they can be
1348 disabled on a per socket basis using the IPV6_AUTOFLOWLABEL
1350 2: automatic flow labels are allowed, they may be enabled on a
1351 per socket basis using the IPV6_AUTOFLOWLABEL socket option
1352 3: automatic flow labels are enabled and enforced, they cannot
1353 be disabled by the socket option
1356 flowlabel_state_ranges - BOOLEAN
1357 Split the flow label number space into two ranges. 0-0x7FFFF is
1358 reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF
1359 is reserved for stateless flow labels as described in RFC6437.
1364 flowlabel_reflect - BOOLEAN
1365 Automatically reflect the flow label. Needed for Path MTU
1366 Discovery to work with Equal Cost Multipath Routing in anycast
1367 environments. See RFC 7690 and:
1368 https://tools.ietf.org/html/draft-wang-6man-flow-label-reflection-01
1373 anycast_src_echo_reply - BOOLEAN
1374 Controls the use of anycast addresses as source addresses for ICMPv6
1380 idgen_delay - INTEGER
1381 Controls the delay in seconds after which time to retry
1382 privacy stable address generation if a DAD conflict is
1384 Default: 1 (as specified in RFC7217)
1386 idgen_retries - INTEGER
1387 Controls the number of retries to generate a stable privacy
1388 address if a DAD conflict is detected.
1389 Default: 3 (as specified in RFC7217)
1392 Controls the MLD query robustness variable (see RFC3810 9.1).
1393 Default: 2 (as specified by RFC3810 9.1)
1394 Minimum: 1 (as specified by RFC6636 4.5)
1398 ip6frag_high_thresh - INTEGER
1399 Maximum memory used to reassemble IPv6 fragments. When
1400 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1401 the fragment handler will toss packets until ip6frag_low_thresh
1404 ip6frag_low_thresh - INTEGER
1405 See ip6frag_high_thresh
1407 ip6frag_time - INTEGER
1408 Time in seconds to keep an IPv6 fragment in memory.
1411 Change the interface-specific default settings.
1415 Change all the interface-specific settings.
1417 [XXX: Other special features than forwarding?]
1419 conf/all/forwarding - BOOLEAN
1420 Enable global IPv6 forwarding between all interfaces.
1422 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1423 to control which interfaces may forward packets and which not.
1425 This also sets all interfaces' Host/Router setting
1426 'forwarding' to the specified value. See below for details.
1428 This referred to as global forwarding.
1433 fwmark_reflect - BOOLEAN
1434 Controls the fwmark of kernel-generated IPv6 reply packets that are not
1435 associated with a socket for example, TCP RSTs or ICMPv6 echo replies).
1436 If unset, these packets have a fwmark of zero. If set, they have the
1437 fwmark of the packet they are replying to.
1441 Change special settings per interface.
1443 The functional behaviour for certain settings is different
1444 depending on whether local forwarding is enabled or not.
1447 Accept Router Advertisements; autoconfigure using them.
1449 It also determines whether or not to transmit Router
1450 Solicitations. If and only if the functional setting is to
1451 accept Router Advertisements, Router Solicitations will be
1454 Possible values are:
1455 0 Do not accept Router Advertisements.
1456 1 Accept Router Advertisements if forwarding is disabled.
1457 2 Overrule forwarding behaviour. Accept Router Advertisements
1458 even if forwarding is enabled.
1460 Functional default: enabled if local forwarding is disabled.
1461 disabled if local forwarding is enabled.
1463 accept_ra_defrtr - BOOLEAN
1464 Learn default router in Router Advertisement.
1466 Functional default: enabled if accept_ra is enabled.
1467 disabled if accept_ra is disabled.
1469 accept_ra_from_local - BOOLEAN
1470 Accept RA with source-address that is found on local machine
1471 if the RA is otherwise proper and able to be accepted.
1472 Default is to NOT accept these as it may be an un-intended
1476 enabled if accept_ra_from_local is enabled
1477 on a specific interface.
1478 disabled if accept_ra_from_local is disabled
1479 on a specific interface.
1481 accept_ra_min_hop_limit - INTEGER
1482 Minimum hop limit Information in Router Advertisement.
1484 Hop limit Information in Router Advertisement less than this
1485 variable shall be ignored.
1489 accept_ra_pinfo - BOOLEAN
1490 Learn Prefix Information in Router Advertisement.
1492 Functional default: enabled if accept_ra is enabled.
1493 disabled if accept_ra is disabled.
1495 accept_ra_rt_info_min_plen - INTEGER
1496 Minimum prefix length of Route Information in RA.
1498 Route Information w/ prefix smaller than this variable shall
1501 Functional default: 0 if accept_ra_rtr_pref is enabled.
1502 -1 if accept_ra_rtr_pref is disabled.
1504 accept_ra_rt_info_max_plen - INTEGER
1505 Maximum prefix length of Route Information in RA.
1507 Route Information w/ prefix larger than this variable shall
1510 Functional default: 0 if accept_ra_rtr_pref is enabled.
1511 -1 if accept_ra_rtr_pref is disabled.
1513 accept_ra_rtr_pref - BOOLEAN
1514 Accept Router Preference in RA.
1516 Functional default: enabled if accept_ra is enabled.
1517 disabled if accept_ra is disabled.
1519 accept_ra_mtu - BOOLEAN
1520 Apply the MTU value specified in RA option 5 (RFC4861). If
1521 disabled, the MTU specified in the RA will be ignored.
1523 Functional default: enabled if accept_ra is enabled.
1524 disabled if accept_ra is disabled.
1526 accept_redirects - BOOLEAN
1529 Functional default: enabled if local forwarding is disabled.
1530 disabled if local forwarding is enabled.
1532 accept_source_route - INTEGER
1533 Accept source routing (routing extension header).
1535 >= 0: Accept only routing header type 2.
1536 < 0: Do not accept routing header.
1541 Autoconfigure addresses using Prefix Information in Router
1544 Functional default: enabled if accept_ra_pinfo is enabled.
1545 disabled if accept_ra_pinfo is disabled.
1547 dad_transmits - INTEGER
1548 The amount of Duplicate Address Detection probes to send.
1551 forwarding - INTEGER
1552 Configure interface-specific Host/Router behaviour.
1554 Note: It is recommended to have the same setting on all
1555 interfaces; mixed router/host scenarios are rather uncommon.
1557 Possible values are:
1558 0 Forwarding disabled
1559 1 Forwarding enabled
1563 By default, Host behaviour is assumed. This means:
1565 1. IsRouter flag is not set in Neighbour Advertisements.
1566 2. If accept_ra is TRUE (default), transmit Router
1568 3. If accept_ra is TRUE (default), accept Router
1569 Advertisements (and do autoconfiguration).
1570 4. If accept_redirects is TRUE (default), accept Redirects.
1574 If local forwarding is enabled, Router behaviour is assumed.
1575 This means exactly the reverse from the above:
1577 1. IsRouter flag is set in Neighbour Advertisements.
1578 2. Router Solicitations are not sent unless accept_ra is 2.
1579 3. Router Advertisements are ignored unless accept_ra is 2.
1580 4. Redirects are ignored.
1582 Default: 0 (disabled) if global forwarding is disabled (default),
1583 otherwise 1 (enabled).
1586 Default Hop Limit to set.
1590 Default Maximum Transfer Unit
1591 Default: 1280 (IPv6 required minimum)
1593 ip_nonlocal_bind - BOOLEAN
1594 If set, allows processes to bind() to non-local IPv6 addresses,
1595 which can be quite useful - but may break some applications.
1598 router_probe_interval - INTEGER
1599 Minimum interval (in seconds) between Router Probing described
1604 router_solicitation_delay - INTEGER
1605 Number of seconds to wait after interface is brought up
1606 before sending Router Solicitations.
1609 router_solicitation_interval - INTEGER
1610 Number of seconds to wait between Router Solicitations.
1613 router_solicitations - INTEGER
1614 Number of Router Solicitations to send until assuming no
1615 routers are present.
1618 use_oif_addrs_only - BOOLEAN
1619 When enabled, the candidate source addresses for destinations
1620 routed via this interface are restricted to the set of addresses
1621 configured on this interface (vis. RFC 6724, section 4).
1625 use_tempaddr - INTEGER
1626 Preference for Privacy Extensions (RFC3041).
1627 <= 0 : disable Privacy Extensions
1628 == 1 : enable Privacy Extensions, but prefer public
1629 addresses over temporary addresses.
1630 > 1 : enable Privacy Extensions and prefer temporary
1631 addresses over public addresses.
1632 Default: 0 (for most devices)
1633 -1 (for point-to-point devices and loopback devices)
1635 temp_valid_lft - INTEGER
1636 valid lifetime (in seconds) for temporary addresses.
1637 Default: 604800 (7 days)
1639 temp_prefered_lft - INTEGER
1640 Preferred lifetime (in seconds) for temporary addresses.
1641 Default: 86400 (1 day)
1643 keep_addr_on_down - INTEGER
1644 Keep all IPv6 addresses on an interface down event. If set static
1645 global addresses with no expiration time are not flushed.
1650 Default: 0 (addresses are removed)
1652 max_desync_factor - INTEGER
1653 Maximum value for DESYNC_FACTOR, which is a random value
1654 that ensures that clients don't synchronize with each
1655 other and generate new addresses at exactly the same time.
1656 value is in seconds.
1659 regen_max_retry - INTEGER
1660 Number of attempts before give up attempting to generate
1661 valid temporary addresses.
1664 max_addresses - INTEGER
1665 Maximum number of autoconfigured addresses per interface. Setting
1666 to zero disables the limitation. It is not recommended to set this
1667 value too large (or to zero) because it would be an easy way to
1668 crash the kernel by allowing too many addresses to be created.
1671 disable_ipv6 - BOOLEAN
1672 Disable IPv6 operation. If accept_dad is set to 2, this value
1673 will be dynamically set to TRUE if DAD fails for the link-local
1675 Default: FALSE (enable IPv6 operation)
1677 When this value is changed from 1 to 0 (IPv6 is being enabled),
1678 it will dynamically create a link-local address on the given
1679 interface and start Duplicate Address Detection, if necessary.
1681 When this value is changed from 0 to 1 (IPv6 is being disabled),
1682 it will dynamically delete all address on the given interface.
1684 accept_dad - INTEGER
1685 Whether to accept DAD (Duplicate Address Detection).
1687 1: Enable DAD (default)
1688 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1689 link-local address has been found.
1691 DAD operation and mode on a given interface will be selected according
1692 to the maximum value of conf/{all,interface}/accept_dad.
1694 force_tllao - BOOLEAN
1695 Enable sending the target link-layer address option even when
1696 responding to a unicast neighbor solicitation.
1699 Quoting from RFC 2461, section 4.4, Target link-layer address:
1701 "The option MUST be included for multicast solicitations in order to
1702 avoid infinite Neighbor Solicitation "recursion" when the peer node
1703 does not have a cache entry to return a Neighbor Advertisements
1704 message. When responding to unicast solicitations, the option can be
1705 omitted since the sender of the solicitation has the correct link-
1706 layer address; otherwise it would not have be able to send the unicast
1707 solicitation in the first place. However, including the link-layer
1708 address in this case adds little overhead and eliminates a potential
1709 race condition where the sender deletes the cached link-layer address
1710 prior to receiving a response to a previous solicitation."
1712 ndisc_notify - BOOLEAN
1713 Define mode for notification of address and device changes.
1714 0 - (default): do nothing
1715 1 - Generate unsolicited neighbour advertisements when device is brought
1716 up or hardware address changes.
1718 mldv1_unsolicited_report_interval - INTEGER
1719 The interval in milliseconds in which the next unsolicited
1720 MLDv1 report retransmit will take place.
1721 Default: 10000 (10 seconds)
1723 mldv2_unsolicited_report_interval - INTEGER
1724 The interval in milliseconds in which the next unsolicited
1725 MLDv2 report retransmit will take place.
1726 Default: 1000 (1 second)
1728 force_mld_version - INTEGER
1729 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
1730 1 - Enforce to use MLD version 1
1731 2 - Enforce to use MLD version 2
1733 suppress_frag_ndisc - INTEGER
1734 Control RFC 6980 (Security Implications of IPv6 Fragmentation
1735 with IPv6 Neighbor Discovery) behavior:
1736 1 - (default) discard fragmented neighbor discovery packets
1737 0 - allow fragmented neighbor discovery packets
1739 optimistic_dad - BOOLEAN
1740 Whether to perform Optimistic Duplicate Address Detection (RFC 4429).
1741 0: disabled (default)
1744 Optimistic Duplicate Address Detection for the interface will be enabled
1745 if at least one of conf/{all,interface}/optimistic_dad is set to 1,
1746 it will be disabled otherwise.
1748 use_optimistic - BOOLEAN
1749 If enabled, do not classify optimistic addresses as deprecated during
1750 source address selection. Preferred addresses will still be chosen
1751 before optimistic addresses, subject to other ranking in the source
1752 address selection algorithm.
1753 0: disabled (default)
1756 This will be enabled if at least one of
1757 conf/{all,interface}/use_optimistic is set to 1, disabled otherwise.
1759 stable_secret - IPv6 address
1760 This IPv6 address will be used as a secret to generate IPv6
1761 addresses for link-local addresses and autoconfigured
1762 ones. All addresses generated after setting this secret will
1763 be stable privacy ones by default. This can be changed via the
1764 addrgenmode ip-link. conf/default/stable_secret is used as the
1765 secret for the namespace, the interface specific ones can
1766 overwrite that. Writes to conf/all/stable_secret are refused.
1768 It is recommended to generate this secret during installation
1769 of a system and keep it stable after that.
1771 By default the stable secret is unset.
1773 drop_unicast_in_l2_multicast - BOOLEAN
1774 Drop any unicast IPv6 packets that are received in link-layer
1775 multicast (or broadcast) frames.
1777 By default this is turned off.
1779 drop_unsolicited_na - BOOLEAN
1780 Drop all unsolicited neighbor advertisements, for example if there's
1781 a known good NA proxy on the network and such frames need not be used
1782 (or in the case of 802.11, must not be used to prevent attacks.)
1784 By default this is turned off.
1786 enhanced_dad - BOOLEAN
1787 Include a nonce option in the IPv6 neighbor solicitation messages used for
1788 duplicate address detection per RFC7527. A received DAD NS will only signal
1789 a duplicate address if the nonce is different. This avoids any false
1790 detection of duplicates due to loopback of the NS messages that we send.
1791 The nonce option will be sent on an interface unless both of
1792 conf/{all,interface}/enhanced_dad are set to FALSE.
1797 Limit the maximal rates for sending ICMPv6 packets.
1798 0 to disable any limiting,
1799 otherwise the minimal space between responses in milliseconds.
1802 xfrm6_gc_thresh - INTEGER
1803 The threshold at which we will start garbage collecting for IPv6
1804 destination cache entries. At twice this value the system will
1805 refuse new allocations.
1809 Pekka Savola <pekkas@netcore.fi>
1810 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1813 /proc/sys/net/bridge/* Variables:
1815 bridge-nf-call-arptables - BOOLEAN
1816 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1820 bridge-nf-call-iptables - BOOLEAN
1821 1 : pass bridged IPv4 traffic to iptables' chains.
1825 bridge-nf-call-ip6tables - BOOLEAN
1826 1 : pass bridged IPv6 traffic to ip6tables' chains.
1830 bridge-nf-filter-vlan-tagged - BOOLEAN
1831 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1835 bridge-nf-filter-pppoe-tagged - BOOLEAN
1836 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1840 bridge-nf-pass-vlan-input-dev - BOOLEAN
1841 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
1842 interface on the bridge and set the netfilter input device to the vlan.
1843 This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
1844 target work with vlan-on-top-of-bridge interfaces. When no matching
1845 vlan interface is found, or this switch is off, the input device is
1846 set to the bridge interface.
1847 0: disable bridge netfilter vlan interface lookup.
1850 proc/sys/net/sctp/* Variables:
1852 addip_enable - BOOLEAN
1853 Enable or disable extension of Dynamic Address Reconfiguration
1854 (ADD-IP) functionality specified in RFC5061. This extension provides
1855 the ability to dynamically add and remove new addresses for the SCTP
1858 1: Enable extension.
1860 0: Disable extension.
1865 Enable or disable pf (pf is short for potentially failed) state. A value
1866 of pf_retrans > path_max_retrans also disables pf state. That is, one of
1867 both pf_enable and pf_retrans > path_max_retrans can disable pf state.
1868 Since pf_retrans and path_max_retrans can be changed by userspace
1869 application, sometimes user expects to disable pf state by the value of
1870 pf_retrans > path_max_retrans, but occasionally the value of pf_retrans
1871 or path_max_retrans is changed by the user application, this pf state is
1872 enabled. As such, it is necessary to add this to dynamically enable
1873 and disable pf state. See:
1874 https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for
1883 addip_noauth_enable - BOOLEAN
1884 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1885 authentication to protect the operations of adding or removing new
1886 addresses. This requirement is mandated so that unauthorized hosts
1887 would not be able to hijack associations. However, older
1888 implementations may not have implemented this requirement while
1889 allowing the ADD-IP extension. For reasons of interoperability,
1890 we provide this variable to control the enforcement of the
1891 authentication requirement.
1893 1: Allow ADD-IP extension to be used without authentication. This
1894 should only be set in a closed environment for interoperability
1895 with older implementations.
1897 0: Enforce the authentication requirement
1901 auth_enable - BOOLEAN
1902 Enable or disable Authenticated Chunks extension. This extension
1903 provides the ability to send and receive authenticated chunks and is
1904 required for secure operation of Dynamic Address Reconfiguration
1907 1: Enable this extension.
1908 0: Disable this extension.
1912 prsctp_enable - BOOLEAN
1913 Enable or disable the Partial Reliability extension (RFC3758) which
1914 is used to notify peers that a given DATA should no longer be expected.
1922 The limit of the number of new packets that can be initially sent. It
1923 controls how bursty the generated traffic can be.
1927 association_max_retrans - INTEGER
1928 Set the maximum number for retransmissions that an association can
1929 attempt deciding that the remote end is unreachable. If this value
1930 is exceeded, the association is terminated.
1934 max_init_retransmits - INTEGER
1935 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1936 that an association will attempt before declaring the destination
1937 unreachable and terminating.
1941 path_max_retrans - INTEGER
1942 The maximum number of retransmissions that will be attempted on a given
1943 path. Once this threshold is exceeded, the path is considered
1944 unreachable, and new traffic will use a different path when the
1945 association is multihomed.
1949 pf_retrans - INTEGER
1950 The number of retransmissions that will be attempted on a given path
1951 before traffic is redirected to an alternate transport (should one
1952 exist). Note this is distinct from path_max_retrans, as a path that
1953 passes the pf_retrans threshold can still be used. Its only
1954 deprioritized when a transmission path is selected by the stack. This
1955 setting is primarily used to enable fast failover mechanisms without
1956 having to reduce path_max_retrans to a very low value. See:
1957 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
1958 for details. Note also that a value of pf_retrans > path_max_retrans
1959 disables this feature. Since both pf_retrans and path_max_retrans can
1960 be changed by userspace application, a variable pf_enable is used to
1965 rto_initial - INTEGER
1966 The initial round trip timeout value in milliseconds that will be used
1967 in calculating round trip times. This is the initial time interval
1968 for retransmissions.
1973 The maximum value (in milliseconds) of the round trip timeout. This
1974 is the largest time interval that can elapse between retransmissions.
1979 The minimum value (in milliseconds) of the round trip timeout. This
1980 is the smallest time interval the can elapse between retransmissions.
1984 hb_interval - INTEGER
1985 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
1986 are sent at the specified interval on idle paths to probe the state of
1987 a given path between 2 associations.
1991 sack_timeout - INTEGER
1992 The amount of time (in milliseconds) that the implementation will wait
1997 valid_cookie_life - INTEGER
1998 The default lifetime of the SCTP cookie (in milliseconds). The cookie
1999 is used during association establishment.
2003 cookie_preserve_enable - BOOLEAN
2004 Enable or disable the ability to extend the lifetime of the SCTP cookie
2005 that is used during the establishment phase of SCTP association
2007 1: Enable cookie lifetime extension.
2012 cookie_hmac_alg - STRING
2013 Select the hmac algorithm used when generating the cookie value sent by
2014 a listening sctp socket to a connecting client in the INIT-ACK chunk.
2019 Ability to assign md5 or sha1 as the selected alg is predicated on the
2020 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
2021 CONFIG_CRYPTO_SHA1).
2023 Default: Dependent on configuration. MD5 if available, else SHA1 if
2024 available, else none.
2026 rcvbuf_policy - INTEGER
2027 Determines if the receive buffer is attributed to the socket or to
2028 association. SCTP supports the capability to create multiple
2029 associations on a single socket. When using this capability, it is
2030 possible that a single stalled association that's buffering a lot
2031 of data may block other associations from delivering their data by
2032 consuming all of the receive buffer space. To work around this,
2033 the rcvbuf_policy could be set to attribute the receiver buffer space
2034 to each association instead of the socket. This prevents the described
2037 1: rcvbuf space is per association
2038 0: rcvbuf space is per socket
2042 sndbuf_policy - INTEGER
2043 Similar to rcvbuf_policy above, this applies to send buffer space.
2045 1: Send buffer is tracked per association
2046 0: Send buffer is tracked per socket.
2050 sctp_mem - vector of 3 INTEGERs: min, pressure, max
2051 Number of pages allowed for queueing by all SCTP sockets.
2053 min: Below this number of pages SCTP is not bothered about its
2054 memory appetite. When amount of memory allocated by SCTP exceeds
2055 this number, SCTP starts to moderate memory usage.
2057 pressure: This value was introduced to follow format of tcp_mem.
2059 max: Number of pages allowed for queueing by all SCTP sockets.
2061 Default is calculated at boot time from amount of available memory.
2063 sctp_rmem - vector of 3 INTEGERs: min, default, max
2064 Only the first value ("min") is used, "default" and "max" are
2067 min: Minimal size of receive buffer used by SCTP socket.
2068 It is guaranteed to each SCTP socket (but not association) even
2069 under moderate memory pressure.
2073 sctp_wmem - vector of 3 INTEGERs: min, default, max
2074 Currently this tunable has no effect.
2076 addr_scope_policy - INTEGER
2077 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
2079 0 - Disable IPv4 address scoping
2080 1 - Enable IPv4 address scoping
2081 2 - Follow draft but allow IPv4 private addresses
2082 3 - Follow draft but allow IPv4 link local addresses
2087 /proc/sys/net/core/*
2088 Please see: Documentation/sysctl/net.txt for descriptions of these entries.
2091 /proc/sys/net/unix/*
2092 max_dgram_qlen - INTEGER
2093 The maximum length of dgram socket receive queue
2100 /proc/sys/net/irda/*
2101 fast_poll_increase FIXME
2102 warn_noreply_time FIXME
2103 discovery_slots FIXME
2106 discovery_timeout FIXME
2107 lap_keepalive_time FIXME
2108 max_noreply_time FIXME
2109 max_tx_data_size FIXME
2111 min_tx_turn_time FIXME