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 hardened 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 ip_forward_update_priority - INTEGER
85 Whether to update SKB priority from "TOS" field in IPv4 header after it
86 is forwarded. The new SKB priority is mapped from TOS field value
87 according to an rt_tos2priority table (see e.g. man tc-prio).
88 Default: 1 (Update priority.)
90 0 - Do not update priority.
93 route/max_size - INTEGER
94 Maximum number of routes allowed in the kernel. Increase
95 this when using large numbers of interfaces and/or routes.
96 From linux kernel 3.6 onwards, this is deprecated for ipv4
97 as route cache is no longer used.
99 neigh/default/gc_thresh1 - INTEGER
100 Minimum number of entries to keep. Garbage collector will not
101 purge entries if there are fewer than this number.
104 neigh/default/gc_thresh2 - INTEGER
105 Threshold when garbage collector becomes more aggressive about
106 purging entries. Entries older than 5 seconds will be cleared
107 when over this number.
110 neigh/default/gc_thresh3 - INTEGER
111 Maximum number of neighbor entries allowed. Increase this
112 when using large numbers of interfaces and when communicating
113 with large numbers of directly-connected peers.
116 neigh/default/unres_qlen_bytes - INTEGER
117 The maximum number of bytes which may be used by packets
118 queued for each unresolved address by other network layers.
120 Setting negative value is meaningless and will return error.
121 Default: SK_WMEM_MAX, (same as net.core.wmem_default).
122 Exact value depends on architecture and kernel options,
123 but should be enough to allow queuing 256 packets
126 neigh/default/unres_qlen - INTEGER
127 The maximum number of packets which may be queued for each
128 unresolved address by other network layers.
129 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
130 Prior to linux 3.3, the default value is 3 which may cause
131 unexpected packet loss. The current default value is calculated
132 according to default value of unres_qlen_bytes and true size of
136 mtu_expires - INTEGER
137 Time, in seconds, that cached PMTU information is kept.
139 min_adv_mss - INTEGER
140 The advertised MSS depends on the first hop route MTU, but will
141 never be lower than this setting.
145 ipfrag_high_thresh - LONG INTEGER
146 Maximum memory used to reassemble IP fragments.
148 ipfrag_low_thresh - LONG INTEGER
149 (Obsolete since linux-4.17)
150 Maximum memory used to reassemble IP fragments before the kernel
151 begins to remove incomplete fragment queues to free up resources.
152 The kernel still accepts new fragments for defragmentation.
154 ipfrag_time - INTEGER
155 Time in seconds to keep an IP fragment in memory.
157 ipfrag_max_dist - INTEGER
158 ipfrag_max_dist is a non-negative integer value which defines the
159 maximum "disorder" which is allowed among fragments which share a
160 common IP source address. Note that reordering of packets is
161 not unusual, but if a large number of fragments arrive from a source
162 IP address while a particular fragment queue remains incomplete, it
163 probably indicates that one or more fragments belonging to that queue
164 have been lost. When ipfrag_max_dist is positive, an additional check
165 is done on fragments before they are added to a reassembly queue - if
166 ipfrag_max_dist (or more) fragments have arrived from a particular IP
167 address between additions to any IP fragment queue using that source
168 address, it's presumed that one or more fragments in the queue are
169 lost. The existing fragment queue will be dropped, and a new one
170 started. An ipfrag_max_dist value of zero disables this check.
172 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
173 result in unnecessarily dropping fragment queues when normal
174 reordering of packets occurs, which could lead to poor application
175 performance. Using a very large value, e.g. 50000, increases the
176 likelihood of incorrectly reassembling IP fragments that originate
177 from different IP datagrams, which could result in data corruption.
182 inet_peer_threshold - INTEGER
183 The approximate size of the storage. Starting from this threshold
184 entries will be thrown aggressively. This threshold also determines
185 entries' time-to-live and time intervals between garbage collection
186 passes. More entries, less time-to-live, less GC interval.
188 inet_peer_minttl - INTEGER
189 Minimum time-to-live of entries. Should be enough to cover fragment
190 time-to-live on the reassembling side. This minimum time-to-live is
191 guaranteed if the pool size is less than inet_peer_threshold.
194 inet_peer_maxttl - INTEGER
195 Maximum time-to-live of entries. Unused entries will expire after
196 this period of time if there is no memory pressure on the pool (i.e.
197 when the number of entries in the pool is very small).
203 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
204 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
207 tcp_abort_on_overflow - BOOLEAN
208 If listening service is too slow to accept new connections,
209 reset them. Default state is FALSE. It means that if overflow
210 occurred due to a burst, connection will recover. Enable this
211 option _only_ if you are really sure that listening daemon
212 cannot be tuned to accept connections faster. Enabling this
213 option can harm clients of your server.
215 tcp_adv_win_scale - INTEGER
216 Count buffering overhead as bytes/2^tcp_adv_win_scale
217 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
219 Possible values are [-31, 31], inclusive.
222 tcp_allowed_congestion_control - STRING
223 Show/set the congestion control choices available to non-privileged
224 processes. The list is a subset of those listed in
225 tcp_available_congestion_control.
226 Default is "reno" and the default setting (tcp_congestion_control).
228 tcp_app_win - INTEGER
229 Reserve max(window/2^tcp_app_win, mss) of window for application
230 buffer. Value 0 is special, it means that nothing is reserved.
233 tcp_autocorking - BOOLEAN
234 Enable TCP auto corking :
235 When applications do consecutive small write()/sendmsg() system calls,
236 we try to coalesce these small writes as much as possible, to lower
237 total amount of sent packets. This is done if at least one prior
238 packet for the flow is waiting in Qdisc queues or device transmit
239 queue. Applications can still use TCP_CORK for optimal behavior
240 when they know how/when to uncork their sockets.
243 tcp_available_congestion_control - STRING
244 Shows the available congestion control choices that are registered.
245 More congestion control algorithms may be available as modules,
248 tcp_base_mss - INTEGER
249 The initial value of search_low to be used by the packetization layer
250 Path MTU discovery (MTU probing). If MTU probing is enabled,
251 this is the initial MSS used by the connection.
253 tcp_min_snd_mss - INTEGER
254 TCP SYN and SYNACK messages usually advertise an ADVMSS option,
255 as described in RFC 1122 and RFC 6691.
256 If this ADVMSS option is smaller than tcp_min_snd_mss,
257 it is silently capped to tcp_min_snd_mss.
259 Default : 48 (at least 8 bytes of payload per segment)
261 tcp_congestion_control - STRING
262 Set the congestion control algorithm to be used for new
263 connections. The algorithm "reno" is always available, but
264 additional choices may be available based on kernel configuration.
265 Default is set as part of kernel configuration.
266 For passive connections, the listener congestion control choice
268 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
271 Allows TCP to send "duplicate" SACKs.
273 tcp_early_retrans - INTEGER
274 Tail loss probe (TLP) converts RTOs occurring due to tail
275 losses into fast recovery (draft-ietf-tcpm-rack). Note that
276 TLP requires RACK to function properly (see tcp_recovery below)
283 Control use of Explicit Congestion Notification (ECN) by TCP.
284 ECN is used only when both ends of the TCP connection indicate
285 support for it. This feature is useful in avoiding losses due
286 to congestion by allowing supporting routers to signal
287 congestion before having to drop packets.
289 0 Disable ECN. Neither initiate nor accept ECN.
290 1 Enable ECN when requested by incoming connections and
291 also request ECN on outgoing connection attempts.
292 2 Enable ECN when requested by incoming connections
293 but do not request ECN on outgoing connections.
296 tcp_ecn_fallback - BOOLEAN
297 If the kernel detects that ECN connection misbehaves, enable fall
298 back to non-ECN. Currently, this knob implements the fallback
299 from RFC3168, section 6.1.1.1., but we reserve that in future,
300 additional detection mechanisms could be implemented under this
301 knob. The value is not used, if tcp_ecn or per route (or congestion
302 control) ECN settings are disabled.
303 Default: 1 (fallback enabled)
306 This is a legacy option, it has no effect anymore.
308 tcp_fin_timeout - INTEGER
309 The length of time an orphaned (no longer referenced by any
310 application) connection will remain in the FIN_WAIT_2 state
311 before it is aborted at the local end. While a perfectly
312 valid "receive only" state for an un-orphaned connection, an
313 orphaned connection in FIN_WAIT_2 state could otherwise wait
314 forever for the remote to close its end of the connection.
319 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
320 F-RTO is an enhanced recovery algorithm for TCP retransmission
321 timeouts. It is particularly beneficial in networks where the
322 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
323 modification. It does not require any support from the peer.
325 By default it's enabled with a non-zero value. 0 disables F-RTO.
327 tcp_invalid_ratelimit - INTEGER
328 Limit the maximal rate for sending duplicate acknowledgments
329 in response to incoming TCP packets that are for an existing
330 connection but that are invalid due to any of these reasons:
332 (a) out-of-window sequence number,
333 (b) out-of-window acknowledgment number, or
334 (c) PAWS (Protection Against Wrapped Sequence numbers) check failure
336 This can help mitigate simple "ack loop" DoS attacks, wherein
337 a buggy or malicious middlebox or man-in-the-middle can
338 rewrite TCP header fields in manner that causes each endpoint
339 to think that the other is sending invalid TCP segments, thus
340 causing each side to send an unterminating stream of duplicate
341 acknowledgments for invalid segments.
343 Using 0 disables rate-limiting of dupacks in response to
344 invalid segments; otherwise this value specifies the minimal
345 space between sending such dupacks, in milliseconds.
347 Default: 500 (milliseconds).
349 tcp_keepalive_time - INTEGER
350 How often TCP sends out keepalive messages when keepalive is enabled.
353 tcp_keepalive_probes - INTEGER
354 How many keepalive probes TCP sends out, until it decides that the
355 connection is broken. Default value: 9.
357 tcp_keepalive_intvl - INTEGER
358 How frequently the probes are send out. Multiplied by
359 tcp_keepalive_probes it is time to kill not responding connection,
360 after probes started. Default value: 75sec i.e. connection
361 will be aborted after ~11 minutes of retries.
363 tcp_l3mdev_accept - BOOLEAN
364 Enables child sockets to inherit the L3 master device index.
365 Enabling this option allows a "global" listen socket to work
366 across L3 master domains (e.g., VRFs) with connected sockets
367 derived from the listen socket to be bound to the L3 domain in
368 which the packets originated. Only valid when the kernel was
369 compiled with CONFIG_NET_L3_MASTER_DEV.
371 tcp_low_latency - BOOLEAN
372 This is a legacy option, it has no effect anymore.
374 tcp_max_orphans - INTEGER
375 Maximal number of TCP sockets not attached to any user file handle,
376 held by system. If this number is exceeded orphaned connections are
377 reset immediately and warning is printed. This limit exists
378 only to prevent simple DoS attacks, you _must_ not rely on this
379 or lower the limit artificially, but rather increase it
380 (probably, after increasing installed memory),
381 if network conditions require more than default value,
382 and tune network services to linger and kill such states
383 more aggressively. Let me to remind again: each orphan eats
384 up to ~64K of unswappable memory.
386 tcp_max_syn_backlog - INTEGER
387 Maximal number of remembered connection requests, which have not
388 received an acknowledgment from connecting client.
389 The minimal value is 128 for low memory machines, and it will
390 increase in proportion to the memory of machine.
391 If server suffers from overload, try increasing this number.
393 tcp_max_tw_buckets - INTEGER
394 Maximal number of timewait sockets held by system simultaneously.
395 If this number is exceeded time-wait socket is immediately destroyed
396 and warning is printed. This limit exists only to prevent
397 simple DoS attacks, you _must_ not lower the limit artificially,
398 but rather increase it (probably, after increasing installed memory),
399 if network conditions require more than default value.
401 tcp_mem - vector of 3 INTEGERs: min, pressure, max
402 min: below this number of pages TCP is not bothered about its
405 pressure: when amount of memory allocated by TCP exceeds this number
406 of pages, TCP moderates its memory consumption and enters memory
407 pressure mode, which is exited when memory consumption falls
410 max: number of pages allowed for queueing by all TCP sockets.
412 Defaults are calculated at boot time from amount of available
415 tcp_min_rtt_wlen - INTEGER
416 The window length of the windowed min filter to track the minimum RTT.
417 A shorter window lets a flow more quickly pick up new (higher)
418 minimum RTT when it is moved to a longer path (e.g., due to traffic
419 engineering). A longer window makes the filter more resistant to RTT
420 inflations such as transient congestion. The unit is seconds.
421 Possible values: 0 - 86400 (1 day)
424 tcp_moderate_rcvbuf - BOOLEAN
425 If set, TCP performs receive buffer auto-tuning, attempting to
426 automatically size the buffer (no greater than tcp_rmem[2]) to
427 match the size required by the path for full throughput. Enabled by
430 tcp_mtu_probing - INTEGER
431 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
434 1 - Disabled by default, enabled when an ICMP black hole detected
435 2 - Always enabled, use initial MSS of tcp_base_mss.
437 tcp_probe_interval - UNSIGNED INTEGER
438 Controls how often to start TCP Packetization-Layer Path MTU
439 Discovery reprobe. The default is reprobing every 10 minutes as
442 tcp_probe_threshold - INTEGER
443 Controls when TCP Packetization-Layer Path MTU Discovery probing
444 will stop in respect to the width of search range in bytes. Default
447 tcp_no_metrics_save - BOOLEAN
448 By default, TCP saves various connection metrics in the route cache
449 when the connection closes, so that connections established in the
450 near future can use these to set initial conditions. Usually, this
451 increases overall performance, but may sometimes cause performance
452 degradation. If set, TCP will not cache metrics on closing
455 tcp_orphan_retries - INTEGER
456 This value influences the timeout of a locally closed TCP connection,
457 when RTO retransmissions remain unacknowledged.
458 See tcp_retries2 for more details.
460 The default value is 8.
461 If your machine is a loaded WEB server,
462 you should think about lowering this value, such sockets
463 may consume significant resources. Cf. tcp_max_orphans.
465 tcp_recovery - INTEGER
466 This value is a bitmap to enable various experimental loss recovery
469 RACK: 0x1 enables the RACK loss detection for fast detection of lost
470 retransmissions and tail drops. It also subsumes and disables
471 RFC6675 recovery for SACK connections.
472 RACK: 0x2 makes RACK's reordering window static (min_rtt/4).
473 RACK: 0x4 disables RACK's DUPACK threshold heuristic
477 tcp_reordering - INTEGER
478 Initial reordering level of packets in a TCP stream.
479 TCP stack can then dynamically adjust flow reordering level
480 between this initial value and tcp_max_reordering
483 tcp_max_reordering - INTEGER
484 Maximal reordering level of packets in a TCP stream.
485 300 is a fairly conservative value, but you might increase it
486 if paths are using per packet load balancing (like bonding rr mode)
489 tcp_retrans_collapse - BOOLEAN
490 Bug-to-bug compatibility with some broken printers.
491 On retransmit try to send bigger packets to work around bugs in
494 tcp_retries1 - INTEGER
495 This value influences the time, after which TCP decides, that
496 something is wrong due to unacknowledged RTO retransmissions,
497 and reports this suspicion to the network layer.
498 See tcp_retries2 for more details.
500 RFC 1122 recommends at least 3 retransmissions, which is the
503 tcp_retries2 - INTEGER
504 This value influences the timeout of an alive TCP connection,
505 when RTO retransmissions remain unacknowledged.
506 Given a value of N, a hypothetical TCP connection following
507 exponential backoff with an initial RTO of TCP_RTO_MIN would
508 retransmit N times before killing the connection at the (N+1)th RTO.
510 The default value of 15 yields a hypothetical timeout of 924.6
511 seconds and is a lower bound for the effective timeout.
512 TCP will effectively time out at the first RTO which exceeds the
513 hypothetical timeout.
515 RFC 1122 recommends at least 100 seconds for the timeout,
516 which corresponds to a value of at least 8.
518 tcp_rfc1337 - BOOLEAN
519 If set, the TCP stack behaves conforming to RFC1337. If unset,
520 we are not conforming to RFC, but prevent TCP TIME_WAIT
524 tcp_rmem - vector of 3 INTEGERs: min, default, max
525 min: Minimal size of receive buffer used by TCP sockets.
526 It is guaranteed to each TCP socket, even under moderate memory
530 default: initial size of receive buffer used by TCP sockets.
531 This value overrides net.core.rmem_default used by other protocols.
532 Default: 87380 bytes. This value results in window of 65535 with
533 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
534 less for default tcp_app_win. See below about these variables.
536 max: maximal size of receive buffer allowed for automatically
537 selected receiver buffers for TCP socket. This value does not override
538 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
539 automatic tuning of that socket's receive buffer size, in which
540 case this value is ignored.
541 Default: between 87380B and 6MB, depending on RAM size.
544 Enable select acknowledgments (SACKS).
546 tcp_comp_sack_delay_ns - LONG INTEGER
547 TCP tries to reduce number of SACK sent, using a timer
548 based on 5% of SRTT, capped by this sysctl, in nano seconds.
549 The default is 1ms, based on TSO autosizing period.
551 Default : 1,000,000 ns (1 ms)
553 tcp_comp_sack_nr - INTEGER
554 Max numer of SACK that can be compressed.
555 Using 0 disables SACK compression.
559 tcp_slow_start_after_idle - BOOLEAN
560 If set, provide RFC2861 behavior and time out the congestion
561 window after an idle period. An idle period is defined at
562 the current RTO. If unset, the congestion window will not
563 be timed out after an idle period.
567 Use the Host requirements interpretation of the TCP urgent pointer field.
568 Most hosts use the older BSD interpretation, so if you turn this on
569 Linux might not communicate correctly with them.
572 tcp_synack_retries - INTEGER
573 Number of times SYNACKs for a passive TCP connection attempt will
574 be retransmitted. Should not be higher than 255. Default value
575 is 5, which corresponds to 31seconds till the last retransmission
576 with the current initial RTO of 1second. With this the final timeout
577 for a passive TCP connection will happen after 63seconds.
579 tcp_syncookies - BOOLEAN
580 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
581 Send out syncookies when the syn backlog queue of a socket
582 overflows. This is to prevent against the common 'SYN flood attack'
585 Note, that syncookies is fallback facility.
586 It MUST NOT be used to help highly loaded servers to stand
587 against legal connection rate. If you see SYN flood warnings
588 in your logs, but investigation shows that they occur
589 because of overload with legal connections, you should tune
590 another parameters until this warning disappear.
591 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
593 syncookies seriously violate TCP protocol, do not allow
594 to use TCP extensions, can result in serious degradation
595 of some services (f.e. SMTP relaying), visible not by you,
596 but your clients and relays, contacting you. While you see
597 SYN flood warnings in logs not being really flooded, your server
598 is seriously misconfigured.
600 If you want to test which effects syncookies have to your
601 network connections you can set this knob to 2 to enable
602 unconditionally generation of syncookies.
604 tcp_fastopen - INTEGER
605 Enable TCP Fast Open (RFC7413) to send and accept data in the opening
608 The client support is enabled by flag 0x1 (on by default). The client
609 then must use sendmsg() or sendto() with the MSG_FASTOPEN flag,
610 rather than connect() to send data in SYN.
612 The server support is enabled by flag 0x2 (off by default). Then
613 either enable for all listeners with another flag (0x400) or
614 enable individual listeners via TCP_FASTOPEN socket option with
615 the option value being the length of the syn-data backlog.
617 The values (bitmap) are
618 0x1: (client) enables sending data in the opening SYN on the client.
619 0x2: (server) enables the server support, i.e., allowing data in
620 a SYN packet to be accepted and passed to the
621 application before 3-way handshake finishes.
622 0x4: (client) send data in the opening SYN regardless of cookie
623 availability and without a cookie option.
624 0x200: (server) accept data-in-SYN w/o any cookie option present.
625 0x400: (server) enable all listeners to support Fast Open by
626 default without explicit TCP_FASTOPEN socket option.
630 Note that that additional client or server features are only
631 effective if the basic support (0x1 and 0x2) are enabled respectively.
633 tcp_fastopen_blackhole_timeout_sec - INTEGER
634 Initial time period in second to disable Fastopen on active TCP sockets
635 when a TFO firewall blackhole issue happens.
636 This time period will grow exponentially when more blackhole issues
637 get detected right after Fastopen is re-enabled and will reset to
638 initial value when the blackhole issue goes away.
639 0 to disable the blackhole detection.
640 By default, it is set to 1hr.
642 tcp_syn_retries - INTEGER
643 Number of times initial SYNs for an active TCP connection attempt
644 will be retransmitted. Should not be higher than 127. Default value
645 is 6, which corresponds to 63seconds till the last retransmission
646 with the current initial RTO of 1second. With this the final timeout
647 for an active TCP connection attempt will happen after 127seconds.
649 tcp_timestamps - INTEGER
650 Enable timestamps as defined in RFC1323.
652 1: Enable timestamps as defined in RFC1323 and use random offset for
653 each connection rather than only using the current time.
654 2: Like 1, but without random offsets.
657 tcp_min_tso_segs - INTEGER
658 Minimal number of segments per TSO frame.
659 Since linux-3.12, TCP does an automatic sizing of TSO frames,
660 depending on flow rate, instead of filling 64Kbytes packets.
661 For specific usages, it's possible to force TCP to build big
662 TSO frames. Note that TCP stack might split too big TSO packets
663 if available window is too small.
666 tcp_pacing_ss_ratio - INTEGER
667 sk->sk_pacing_rate is set by TCP stack using a ratio applied
668 to current rate. (current_rate = cwnd * mss / srtt)
669 If TCP is in slow start, tcp_pacing_ss_ratio is applied
670 to let TCP probe for bigger speeds, assuming cwnd can be
671 doubled every other RTT.
674 tcp_pacing_ca_ratio - INTEGER
675 sk->sk_pacing_rate is set by TCP stack using a ratio applied
676 to current rate. (current_rate = cwnd * mss / srtt)
677 If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio
678 is applied to conservatively probe for bigger throughput.
681 tcp_tso_win_divisor - INTEGER
682 This allows control over what percentage of the congestion window
683 can be consumed by a single TSO frame.
684 The setting of this parameter is a choice between burstiness and
685 building larger TSO frames.
688 tcp_tw_reuse - INTEGER
689 Enable reuse of TIME-WAIT sockets for new connections when it is
690 safe from protocol viewpoint.
693 2 - enable for loopback traffic only
694 It should not be changed without advice/request of technical
698 tcp_window_scaling - BOOLEAN
699 Enable window scaling as defined in RFC1323.
701 tcp_wmem - vector of 3 INTEGERs: min, default, max
702 min: Amount of memory reserved for send buffers for TCP sockets.
703 Each TCP socket has rights to use it due to fact of its birth.
706 default: initial size of send buffer used by TCP sockets. This
707 value overrides net.core.wmem_default used by other protocols.
708 It is usually lower than net.core.wmem_default.
711 max: Maximal amount of memory allowed for automatically tuned
712 send buffers for TCP sockets. This value does not override
713 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
714 automatic tuning of that socket's send buffer size, in which case
715 this value is ignored.
716 Default: between 64K and 4MB, depending on RAM size.
718 tcp_notsent_lowat - UNSIGNED INTEGER
719 A TCP socket can control the amount of unsent bytes in its write queue,
720 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
721 reports POLLOUT events if the amount of unsent bytes is below a per
722 socket value, and if the write queue is not full. sendmsg() will
723 also not add new buffers if the limit is hit.
725 This global variable controls the amount of unsent data for
726 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
727 to the global variable has immediate effect.
729 Default: UINT_MAX (0xFFFFFFFF)
731 tcp_workaround_signed_windows - BOOLEAN
732 If set, assume no receipt of a window scaling option means the
733 remote TCP is broken and treats the window as a signed quantity.
734 If unset, assume the remote TCP is not broken even if we do
735 not receive a window scaling option from them.
738 tcp_thin_linear_timeouts - BOOLEAN
739 Enable dynamic triggering of linear timeouts for thin streams.
740 If set, a check is performed upon retransmission by timeout to
741 determine if the stream is thin (less than 4 packets in flight).
742 As long as the stream is found to be thin, up to 6 linear
743 timeouts may be performed before exponential backoff mode is
744 initiated. This improves retransmission latency for
745 non-aggressive thin streams, often found to be time-dependent.
746 For more information on thin streams, see
747 Documentation/networking/tcp-thin.txt
750 tcp_limit_output_bytes - INTEGER
751 Controls TCP Small Queue limit per tcp socket.
752 TCP bulk sender tends to increase packets in flight until it
753 gets losses notifications. With SNDBUF autotuning, this can
754 result in a large amount of packets queued on the local machine
755 (e.g.: qdiscs, CPU backlog, or device) hurting latency of other
756 flows, for typical pfifo_fast qdiscs. tcp_limit_output_bytes
757 limits the number of bytes on qdisc or device to reduce artificial
758 RTT/cwnd and reduce bufferbloat.
761 tcp_challenge_ack_limit - INTEGER
762 Limits number of Challenge ACK sent per second, as recommended
763 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
768 udp_l3mdev_accept - BOOLEAN
769 Enabling this option allows a "global" bound socket to work
770 across L3 master domains (e.g., VRFs) with packets capable of
771 being received regardless of the L3 domain in which they
772 originated. Only valid when the kernel was compiled with
773 CONFIG_NET_L3_MASTER_DEV.
775 udp_mem - vector of 3 INTEGERs: min, pressure, max
776 Number of pages allowed for queueing by all UDP sockets.
778 min: Below this number of pages UDP is not bothered about its
779 memory appetite. When amount of memory allocated by UDP exceeds
780 this number, UDP starts to moderate memory usage.
782 pressure: This value was introduced to follow format of tcp_mem.
784 max: Number of pages allowed for queueing by all UDP sockets.
786 Default is calculated at boot time from amount of available memory.
788 udp_rmem_min - INTEGER
789 Minimal size of receive buffer used by UDP sockets in moderation.
790 Each UDP socket is able to use the size for receiving data, even if
791 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
794 udp_wmem_min - INTEGER
795 Minimal size of send buffer used by UDP sockets in moderation.
796 Each UDP socket is able to use the size for sending data, even if
797 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
802 cipso_cache_enable - BOOLEAN
803 If set, enable additions to and lookups from the CIPSO label mapping
804 cache. If unset, additions are ignored and lookups always result in a
805 miss. However, regardless of the setting the cache is still
806 invalidated when required when means you can safely toggle this on and
807 off and the cache will always be "safe".
810 cipso_cache_bucket_size - INTEGER
811 The CIPSO label cache consists of a fixed size hash table with each
812 hash bucket containing a number of cache entries. This variable limits
813 the number of entries in each hash bucket; the larger the value is, the
814 more CIPSO label mappings that can be cached. When the number of
815 entries in a given hash bucket reaches this limit adding new entries
816 causes the oldest entry in the bucket to be removed to make room.
819 cipso_rbm_optfmt - BOOLEAN
820 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
821 the CIPSO draft specification (see Documentation/netlabel for details).
822 This means that when set the CIPSO tag will be padded with empty
823 categories in order to make the packet data 32-bit aligned.
826 cipso_rbm_structvalid - BOOLEAN
827 If set, do a very strict check of the CIPSO option when
828 ip_options_compile() is called. If unset, relax the checks done during
829 ip_options_compile(). Either way is "safe" as errors are caught else
830 where in the CIPSO processing code but setting this to 0 (False) should
831 result in less work (i.e. it should be faster) but could cause problems
832 with other implementations that require strict checking.
837 ip_local_port_range - 2 INTEGERS
838 Defines the local port range that is used by TCP and UDP to
839 choose the local port. The first number is the first, the
840 second the last local port number.
841 If possible, it is better these numbers have different parity.
842 (one even and one odd values)
843 The default values are 32768 and 60999 respectively.
845 ip_local_reserved_ports - list of comma separated ranges
846 Specify the ports which are reserved for known third-party
847 applications. These ports will not be used by automatic port
848 assignments (e.g. when calling connect() or bind() with port
849 number 0). Explicit port allocation behavior is unchanged.
851 The format used for both input and output is a comma separated
852 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
853 10). Writing to the file will clear all previously reserved
854 ports and update the current list with the one given in the
857 Note that ip_local_port_range and ip_local_reserved_ports
858 settings are independent and both are considered by the kernel
859 when determining which ports are available for automatic port
862 You can reserve ports which are not in the current
863 ip_local_port_range, e.g.:
865 $ cat /proc/sys/net/ipv4/ip_local_port_range
867 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
870 although this is redundant. However such a setting is useful
871 if later the port range is changed to a value that will
872 include the reserved ports.
876 ip_unprivileged_port_start - INTEGER
877 This is a per-namespace sysctl. It defines the first
878 unprivileged port in the network namespace. Privileged ports
879 require root or CAP_NET_BIND_SERVICE in order to bind to them.
880 To disable all privileged ports, set this to 0. It may not
881 overlap with the ip_local_reserved_ports range.
885 ip_nonlocal_bind - BOOLEAN
886 If set, allows processes to bind() to non-local IP addresses,
887 which can be quite useful - but may break some applications.
891 If set non-zero, enables support for dynamic addresses.
892 If set to a non-zero value larger than 1, a kernel log
893 message will be printed when dynamic address rewriting
897 ip_early_demux - BOOLEAN
898 Optimize input packet processing down to one demux for
899 certain kinds of local sockets. Currently we only do this
900 for established TCP and connected UDP sockets.
902 It may add an additional cost for pure routing workloads that
903 reduces overall throughput, in such case you should disable it.
906 tcp_early_demux - BOOLEAN
907 Enable early demux for established TCP sockets.
910 udp_early_demux - BOOLEAN
911 Enable early demux for connected UDP sockets. Disable this if
912 your system could experience more unconnected load.
915 icmp_echo_ignore_all - BOOLEAN
916 If set non-zero, then the kernel will ignore all ICMP ECHO
920 icmp_echo_ignore_broadcasts - BOOLEAN
921 If set non-zero, then the kernel will ignore all ICMP ECHO and
922 TIMESTAMP requests sent to it via broadcast/multicast.
925 icmp_ratelimit - INTEGER
926 Limit the maximal rates for sending ICMP packets whose type matches
927 icmp_ratemask (see below) to specific targets.
928 0 to disable any limiting,
929 otherwise the minimal space between responses in milliseconds.
930 Note that another sysctl, icmp_msgs_per_sec limits the number
931 of ICMP packets sent on all targets.
934 icmp_msgs_per_sec - INTEGER
935 Limit maximal number of ICMP packets sent per second from this host.
936 Only messages whose type matches icmp_ratemask (see below) are
937 controlled by this limit. For security reasons, the precise count
938 of messages per second is randomized.
941 icmp_msgs_burst - INTEGER
942 icmp_msgs_per_sec controls number of ICMP packets sent per second,
943 while icmp_msgs_burst controls the burst size of these packets.
944 For security reasons, the precise burst size is randomized.
947 icmp_ratemask - INTEGER
948 Mask made of ICMP types for which rates are being limited.
949 Significant bits: IHGFEDCBA9876543210
950 Default mask: 0000001100000011000 (6168)
952 Bit definitions (see include/linux/icmp.h):
954 3 Destination Unreachable *
959 C Parameter Problem *
964 H Address Mask Request
967 * These are rate limited by default (see default mask above)
969 icmp_ignore_bogus_error_responses - BOOLEAN
970 Some routers violate RFC1122 by sending bogus responses to broadcast
971 frames. Such violations are normally logged via a kernel warning.
972 If this is set to TRUE, the kernel will not give such warnings, which
973 will avoid log file clutter.
976 icmp_errors_use_inbound_ifaddr - BOOLEAN
978 If zero, icmp error messages are sent with the primary address of
979 the exiting interface.
981 If non-zero, the message will be sent with the primary address of
982 the interface that received the packet that caused the icmp error.
983 This is the behaviour network many administrators will expect from
984 a router. And it can make debugging complicated network layouts
987 Note that if no primary address exists for the interface selected,
988 then the primary address of the first non-loopback interface that
989 has one will be used regardless of this setting.
993 igmp_max_memberships - INTEGER
994 Change the maximum number of multicast groups we can subscribe to.
997 Theoretical maximum value is bounded by having to send a membership
998 report in a single datagram (i.e. the report can't span multiple
999 datagrams, or risk confusing the switch and leaving groups you don't
1002 The number of supported groups 'M' is bounded by the number of group
1003 report entries you can fit into a single datagram of 65535 bytes.
1005 M = 65536-sizeof (ip header)/(sizeof(Group record))
1007 Group records are variable length, with a minimum of 12 bytes.
1008 So net.ipv4.igmp_max_memberships should not be set higher than:
1010 (65536-24) / 12 = 5459
1012 The value 5459 assumes no IP header options, so in practice
1013 this number may be lower.
1015 igmp_max_msf - INTEGER
1016 Maximum number of addresses allowed in the source filter list for a
1021 Controls the IGMP query robustness variable (see RFC2236 8.1).
1022 Default: 2 (as specified by RFC2236 8.1)
1023 Minimum: 1 (as specified by RFC6636 4.5)
1025 force_igmp_version - INTEGER
1026 0 - (default) No enforcement of a IGMP version, IGMPv1/v2 fallback
1027 allowed. Will back to IGMPv3 mode again if all IGMPv1/v2 Querier
1028 Present timer expires.
1029 1 - Enforce to use IGMP version 1. Will also reply IGMPv1 report if
1030 receive IGMPv2/v3 query.
1031 2 - Enforce to use IGMP version 2. Will fallback to IGMPv1 if receive
1032 IGMPv1 query message. Will reply report if receive IGMPv3 query.
1033 3 - Enforce to use IGMP version 3. The same react with default 0.
1035 Note: this is not the same with force_mld_version because IGMPv3 RFC3376
1036 Security Considerations does not have clear description that we could
1037 ignore other version messages completely as MLDv2 RFC3810. So make
1038 this value as default 0 is recommended.
1040 conf/interface/* changes special settings per interface (where
1041 "interface" is the name of your network interface)
1043 conf/all/* is special, changes the settings for all interfaces
1045 log_martians - BOOLEAN
1046 Log packets with impossible addresses to kernel log.
1047 log_martians for the interface will be enabled if at least one of
1048 conf/{all,interface}/log_martians is set to TRUE,
1049 it will be disabled otherwise
1051 accept_redirects - BOOLEAN
1052 Accept ICMP redirect messages.
1053 accept_redirects for the interface will be enabled if:
1054 - both conf/{all,interface}/accept_redirects are TRUE in the case
1055 forwarding for the interface is enabled
1057 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
1058 case forwarding for the interface is disabled
1059 accept_redirects for the interface will be disabled otherwise
1063 forwarding - BOOLEAN
1064 Enable IP forwarding on this interface. This controls whether packets
1065 received _on_ this interface can be forwarded.
1067 mc_forwarding - BOOLEAN
1068 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
1069 and a multicast routing daemon is required.
1070 conf/all/mc_forwarding must also be set to TRUE to enable multicast
1071 routing for the interface
1074 Integer value used to differentiate the devices by the medium they
1075 are attached to. Two devices can have different id values when
1076 the broadcast packets are received only on one of them.
1077 The default value 0 means that the device is the only interface
1078 to its medium, value of -1 means that medium is not known.
1080 Currently, it is used to change the proxy_arp behavior:
1081 the proxy_arp feature is enabled for packets forwarded between
1082 two devices attached to different media.
1086 proxy_arp for the interface will be enabled if at least one of
1087 conf/{all,interface}/proxy_arp is set to TRUE,
1088 it will be disabled otherwise
1090 proxy_arp_pvlan - BOOLEAN
1091 Private VLAN proxy arp.
1092 Basically allow proxy arp replies back to the same interface
1093 (from which the ARP request/solicitation was received).
1095 This is done to support (ethernet) switch features, like RFC
1096 3069, where the individual ports are NOT allowed to
1097 communicate with each other, but they are allowed to talk to
1098 the upstream router. As described in RFC 3069, it is possible
1099 to allow these hosts to communicate through the upstream
1100 router by proxy_arp'ing. Don't need to be used together with
1103 This technology is known by different names:
1104 In RFC 3069 it is called VLAN Aggregation.
1105 Cisco and Allied Telesyn call it Private VLAN.
1106 Hewlett-Packard call it Source-Port filtering or port-isolation.
1107 Ericsson call it MAC-Forced Forwarding (RFC Draft).
1109 shared_media - BOOLEAN
1110 Send(router) or accept(host) RFC1620 shared media redirects.
1111 Overrides secure_redirects.
1112 shared_media for the interface will be enabled if at least one of
1113 conf/{all,interface}/shared_media is set to TRUE,
1114 it will be disabled otherwise
1117 secure_redirects - BOOLEAN
1118 Accept ICMP redirect messages only to gateways listed in the
1119 interface's current gateway list. Even if disabled, RFC1122 redirect
1121 Overridden by shared_media.
1122 secure_redirects for the interface will be enabled if at least one of
1123 conf/{all,interface}/secure_redirects is set to TRUE,
1124 it will be disabled otherwise
1127 send_redirects - BOOLEAN
1128 Send redirects, if router.
1129 send_redirects for the interface will be enabled if at least one of
1130 conf/{all,interface}/send_redirects is set to TRUE,
1131 it will be disabled otherwise
1134 bootp_relay - BOOLEAN
1135 Accept packets with source address 0.b.c.d destined
1136 not to this host as local ones. It is supposed, that
1137 BOOTP relay daemon will catch and forward such packets.
1138 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
1141 Not Implemented Yet.
1143 accept_source_route - BOOLEAN
1144 Accept packets with SRR option.
1145 conf/all/accept_source_route must also be set to TRUE to accept packets
1146 with SRR option on the interface
1147 default TRUE (router)
1150 accept_local - BOOLEAN
1151 Accept packets with local source addresses. In combination with
1152 suitable routing, this can be used to direct packets between two
1153 local interfaces over the wire and have them accepted properly.
1156 route_localnet - BOOLEAN
1157 Do not consider loopback addresses as martian source or destination
1158 while routing. This enables the use of 127/8 for local routing purposes.
1162 0 - No source validation.
1163 1 - Strict mode as defined in RFC3704 Strict Reverse Path
1164 Each incoming packet is tested against the FIB and if the interface
1165 is not the best reverse path the packet check will fail.
1166 By default failed packets are discarded.
1167 2 - Loose mode as defined in RFC3704 Loose Reverse Path
1168 Each incoming packet's source address is also tested against the FIB
1169 and if the source address is not reachable via any interface
1170 the packet check will fail.
1172 Current recommended practice in RFC3704 is to enable strict mode
1173 to prevent IP spoofing from DDos attacks. If using asymmetric routing
1174 or other complicated routing, then loose mode is recommended.
1176 The max value from conf/{all,interface}/rp_filter is used
1177 when doing source validation on the {interface}.
1179 Default value is 0. Note that some distributions enable it
1182 arp_filter - BOOLEAN
1183 1 - Allows you to have multiple network interfaces on the same
1184 subnet, and have the ARPs for each interface be answered
1185 based on whether or not the kernel would route a packet from
1186 the ARP'd IP out that interface (therefore you must use source
1187 based routing for this to work). In other words it allows control
1188 of which cards (usually 1) will respond to an arp request.
1190 0 - (default) The kernel can respond to arp requests with addresses
1191 from other interfaces. This may seem wrong but it usually makes
1192 sense, because it increases the chance of successful communication.
1193 IP addresses are owned by the complete host on Linux, not by
1194 particular interfaces. Only for more complex setups like load-
1195 balancing, does this behaviour cause problems.
1197 arp_filter for the interface will be enabled if at least one of
1198 conf/{all,interface}/arp_filter is set to TRUE,
1199 it will be disabled otherwise
1201 arp_announce - INTEGER
1202 Define different restriction levels for announcing the local
1203 source IP address from IP packets in ARP requests sent on
1205 0 - (default) Use any local address, configured on any interface
1206 1 - Try to avoid local addresses that are not in the target's
1207 subnet for this interface. This mode is useful when target
1208 hosts reachable via this interface require the source IP
1209 address in ARP requests to be part of their logical network
1210 configured on the receiving interface. When we generate the
1211 request we will check all our subnets that include the
1212 target IP and will preserve the source address if it is from
1213 such subnet. If there is no such subnet we select source
1214 address according to the rules for level 2.
1215 2 - Always use the best local address for this target.
1216 In this mode we ignore the source address in the IP packet
1217 and try to select local address that we prefer for talks with
1218 the target host. Such local address is selected by looking
1219 for primary IP addresses on all our subnets on the outgoing
1220 interface that include the target IP address. If no suitable
1221 local address is found we select the first local address
1222 we have on the outgoing interface or on all other interfaces,
1223 with the hope we will receive reply for our request and
1224 even sometimes no matter the source IP address we announce.
1226 The max value from conf/{all,interface}/arp_announce is used.
1228 Increasing the restriction level gives more chance for
1229 receiving answer from the resolved target while decreasing
1230 the level announces more valid sender's information.
1232 arp_ignore - INTEGER
1233 Define different modes for sending replies in response to
1234 received ARP requests that resolve local target IP addresses:
1235 0 - (default): reply for any local target IP address, configured
1237 1 - reply only if the target IP address is local address
1238 configured on the incoming interface
1239 2 - reply only if the target IP address is local address
1240 configured on the incoming interface and both with the
1241 sender's IP address are part from same subnet on this interface
1242 3 - do not reply for local addresses configured with scope host,
1243 only resolutions for global and link addresses are replied
1245 8 - do not reply for all local addresses
1247 The max value from conf/{all,interface}/arp_ignore is used
1248 when ARP request is received on the {interface}
1250 arp_notify - BOOLEAN
1251 Define mode for notification of address and device changes.
1252 0 - (default): do nothing
1253 1 - Generate gratuitous arp requests when device is brought up
1254 or hardware address changes.
1256 arp_accept - BOOLEAN
1257 Define behavior for gratuitous ARP frames who's IP is not
1258 already present in the ARP table:
1259 0 - don't create new entries in the ARP table
1260 1 - create new entries in the ARP table
1262 Both replies and requests type gratuitous arp will trigger the
1263 ARP table to be updated, if this setting is on.
1265 If the ARP table already contains the IP address of the
1266 gratuitous arp frame, the arp table will be updated regardless
1267 if this setting is on or off.
1269 mcast_solicit - INTEGER
1270 The maximum number of multicast probes in INCOMPLETE state,
1271 when the associated hardware address is unknown. Defaults
1274 ucast_solicit - INTEGER
1275 The maximum number of unicast probes in PROBE state, when
1276 the hardware address is being reconfirmed. Defaults to 3.
1278 app_solicit - INTEGER
1279 The maximum number of probes to send to the user space ARP daemon
1280 via netlink before dropping back to multicast probes (see
1281 mcast_resolicit). Defaults to 0.
1283 mcast_resolicit - INTEGER
1284 The maximum number of multicast probes after unicast and
1285 app probes in PROBE state. Defaults to 0.
1287 disable_policy - BOOLEAN
1288 Disable IPSEC policy (SPD) for this interface
1290 disable_xfrm - BOOLEAN
1291 Disable IPSEC encryption on this interface, whatever the policy
1293 igmpv2_unsolicited_report_interval - INTEGER
1294 The interval in milliseconds in which the next unsolicited
1295 IGMPv1 or IGMPv2 report retransmit will take place.
1296 Default: 10000 (10 seconds)
1298 igmpv3_unsolicited_report_interval - INTEGER
1299 The interval in milliseconds in which the next unsolicited
1300 IGMPv3 report retransmit will take place.
1301 Default: 1000 (1 seconds)
1303 promote_secondaries - BOOLEAN
1304 When a primary IP address is removed from this interface
1305 promote a corresponding secondary IP address instead of
1306 removing all the corresponding secondary IP addresses.
1308 drop_unicast_in_l2_multicast - BOOLEAN
1309 Drop any unicast IP packets that are received in link-layer
1310 multicast (or broadcast) frames.
1311 This behavior (for multicast) is actually a SHOULD in RFC
1312 1122, but is disabled by default for compatibility reasons.
1315 drop_gratuitous_arp - BOOLEAN
1316 Drop all gratuitous ARP frames, for example if there's a known
1317 good ARP proxy on the network and such frames need not be used
1318 (or in the case of 802.11, must not be used to prevent attacks.)
1323 Allows you to write a number, which can be used as required.
1326 xfrm4_gc_thresh - INTEGER
1327 The threshold at which we will start garbage collecting for IPv4
1328 destination cache entries. At twice this value the system will
1329 refuse new allocations.
1331 igmp_link_local_mcast_reports - BOOLEAN
1332 Enable IGMP reports for link local multicast groups in the
1337 kuznet@ms2.inr.ac.ru
1343 delon.nicolas@wanadoo.fr
1348 /proc/sys/net/ipv6/* Variables:
1350 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1351 apply to IPv6 [XXX?].
1353 bindv6only - BOOLEAN
1354 Default value for IPV6_V6ONLY socket option,
1355 which restricts use of the IPv6 socket to IPv6 communication
1357 TRUE: disable IPv4-mapped address feature
1358 FALSE: enable IPv4-mapped address feature
1360 Default: FALSE (as specified in RFC3493)
1362 flowlabel_consistency - BOOLEAN
1363 Protect the consistency (and unicity) of flow label.
1364 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1370 auto_flowlabels - INTEGER
1371 Automatically generate flow labels based on a flow hash of the
1372 packet. This allows intermediate devices, such as routers, to
1373 identify packet flows for mechanisms like Equal Cost Multipath
1374 Routing (see RFC 6438).
1375 0: automatic flow labels are completely disabled
1376 1: automatic flow labels are enabled by default, they can be
1377 disabled on a per socket basis using the IPV6_AUTOFLOWLABEL
1379 2: automatic flow labels are allowed, they may be enabled on a
1380 per socket basis using the IPV6_AUTOFLOWLABEL socket option
1381 3: automatic flow labels are enabled and enforced, they cannot
1382 be disabled by the socket option
1385 flowlabel_state_ranges - BOOLEAN
1386 Split the flow label number space into two ranges. 0-0x7FFFF is
1387 reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF
1388 is reserved for stateless flow labels as described in RFC6437.
1393 flowlabel_reflect - BOOLEAN
1394 Automatically reflect the flow label. Needed for Path MTU
1395 Discovery to work with Equal Cost Multipath Routing in anycast
1396 environments. See RFC 7690 and:
1397 https://tools.ietf.org/html/draft-wang-6man-flow-label-reflection-01
1402 fib_multipath_hash_policy - INTEGER
1403 Controls which hash policy to use for multipath routes.
1404 Default: 0 (Layer 3)
1406 0 - Layer 3 (source and destination addresses plus flow label)
1407 1 - Layer 4 (standard 5-tuple)
1409 anycast_src_echo_reply - BOOLEAN
1410 Controls the use of anycast addresses as source addresses for ICMPv6
1416 idgen_delay - INTEGER
1417 Controls the delay in seconds after which time to retry
1418 privacy stable address generation if a DAD conflict is
1420 Default: 1 (as specified in RFC7217)
1422 idgen_retries - INTEGER
1423 Controls the number of retries to generate a stable privacy
1424 address if a DAD conflict is detected.
1425 Default: 3 (as specified in RFC7217)
1428 Controls the MLD query robustness variable (see RFC3810 9.1).
1429 Default: 2 (as specified by RFC3810 9.1)
1430 Minimum: 1 (as specified by RFC6636 4.5)
1432 max_dst_opts_number - INTEGER
1433 Maximum number of non-padding TLVs allowed in a Destination
1434 options extension header. If this value is less than zero
1435 then unknown options are disallowed and the number of known
1436 TLVs allowed is the absolute value of this number.
1439 max_hbh_opts_number - INTEGER
1440 Maximum number of non-padding TLVs allowed in a Hop-by-Hop
1441 options extension header. If this value is less than zero
1442 then unknown options are disallowed and the number of known
1443 TLVs allowed is the absolute value of this number.
1446 max_dst_opts_length - INTEGER
1447 Maximum length allowed for a Destination options extension
1449 Default: INT_MAX (unlimited)
1451 max_hbh_length - INTEGER
1452 Maximum length allowed for a Hop-by-Hop options extension
1454 Default: INT_MAX (unlimited)
1458 ip6frag_high_thresh - INTEGER
1459 Maximum memory used to reassemble IPv6 fragments. When
1460 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1461 the fragment handler will toss packets until ip6frag_low_thresh
1464 ip6frag_low_thresh - INTEGER
1465 See ip6frag_high_thresh
1467 ip6frag_time - INTEGER
1468 Time in seconds to keep an IPv6 fragment in memory.
1470 IPv6 Segment Routing:
1472 seg6_flowlabel - INTEGER
1473 Controls the behaviour of computing the flowlabel of outer
1474 IPv6 header in case of SR T.encaps
1476 -1 set flowlabel to zero.
1477 0 copy flowlabel from Inner packet in case of Inner IPv6
1478 (Set flowlabel to 0 in case IPv4/L2)
1479 1 Compute the flowlabel using seg6_make_flowlabel()
1484 Change the interface-specific default settings.
1488 Change all the interface-specific settings.
1490 [XXX: Other special features than forwarding?]
1492 conf/all/forwarding - BOOLEAN
1493 Enable global IPv6 forwarding between all interfaces.
1495 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1496 to control which interfaces may forward packets and which not.
1498 This also sets all interfaces' Host/Router setting
1499 'forwarding' to the specified value. See below for details.
1501 This referred to as global forwarding.
1506 fwmark_reflect - BOOLEAN
1507 Controls the fwmark of kernel-generated IPv6 reply packets that are not
1508 associated with a socket for example, TCP RSTs or ICMPv6 echo replies).
1509 If unset, these packets have a fwmark of zero. If set, they have the
1510 fwmark of the packet they are replying to.
1514 Change special settings per interface.
1516 The functional behaviour for certain settings is different
1517 depending on whether local forwarding is enabled or not.
1520 Accept Router Advertisements; autoconfigure using them.
1522 It also determines whether or not to transmit Router
1523 Solicitations. If and only if the functional setting is to
1524 accept Router Advertisements, Router Solicitations will be
1527 Possible values are:
1528 0 Do not accept Router Advertisements.
1529 1 Accept Router Advertisements if forwarding is disabled.
1530 2 Overrule forwarding behaviour. Accept Router Advertisements
1531 even if forwarding is enabled.
1533 Functional default: enabled if local forwarding is disabled.
1534 disabled if local forwarding is enabled.
1536 accept_ra_defrtr - BOOLEAN
1537 Learn default router in Router Advertisement.
1539 Functional default: enabled if accept_ra is enabled.
1540 disabled if accept_ra is disabled.
1542 accept_ra_from_local - BOOLEAN
1543 Accept RA with source-address that is found on local machine
1544 if the RA is otherwise proper and able to be accepted.
1545 Default is to NOT accept these as it may be an un-intended
1549 enabled if accept_ra_from_local is enabled
1550 on a specific interface.
1551 disabled if accept_ra_from_local is disabled
1552 on a specific interface.
1554 accept_ra_min_hop_limit - INTEGER
1555 Minimum hop limit Information in Router Advertisement.
1557 Hop limit Information in Router Advertisement less than this
1558 variable shall be ignored.
1562 accept_ra_pinfo - BOOLEAN
1563 Learn Prefix Information in Router Advertisement.
1565 Functional default: enabled if accept_ra is enabled.
1566 disabled if accept_ra is disabled.
1568 accept_ra_rt_info_min_plen - INTEGER
1569 Minimum prefix length of Route Information in RA.
1571 Route Information w/ prefix smaller than this variable shall
1574 Functional default: 0 if accept_ra_rtr_pref is enabled.
1575 -1 if accept_ra_rtr_pref is disabled.
1577 accept_ra_rt_info_max_plen - INTEGER
1578 Maximum prefix length of Route Information in RA.
1580 Route Information w/ prefix larger than this variable shall
1583 Functional default: 0 if accept_ra_rtr_pref is enabled.
1584 -1 if accept_ra_rtr_pref is disabled.
1586 accept_ra_rtr_pref - BOOLEAN
1587 Accept Router Preference in RA.
1589 Functional default: enabled if accept_ra is enabled.
1590 disabled if accept_ra is disabled.
1592 accept_ra_mtu - BOOLEAN
1593 Apply the MTU value specified in RA option 5 (RFC4861). If
1594 disabled, the MTU specified in the RA will be ignored.
1596 Functional default: enabled if accept_ra is enabled.
1597 disabled if accept_ra is disabled.
1599 accept_redirects - BOOLEAN
1602 Functional default: enabled if local forwarding is disabled.
1603 disabled if local forwarding is enabled.
1605 accept_source_route - INTEGER
1606 Accept source routing (routing extension header).
1608 >= 0: Accept only routing header type 2.
1609 < 0: Do not accept routing header.
1614 Autoconfigure addresses using Prefix Information in Router
1617 Functional default: enabled if accept_ra_pinfo is enabled.
1618 disabled if accept_ra_pinfo is disabled.
1620 dad_transmits - INTEGER
1621 The amount of Duplicate Address Detection probes to send.
1624 forwarding - INTEGER
1625 Configure interface-specific Host/Router behaviour.
1627 Note: It is recommended to have the same setting on all
1628 interfaces; mixed router/host scenarios are rather uncommon.
1630 Possible values are:
1631 0 Forwarding disabled
1632 1 Forwarding enabled
1636 By default, Host behaviour is assumed. This means:
1638 1. IsRouter flag is not set in Neighbour Advertisements.
1639 2. If accept_ra is TRUE (default), transmit Router
1641 3. If accept_ra is TRUE (default), accept Router
1642 Advertisements (and do autoconfiguration).
1643 4. If accept_redirects is TRUE (default), accept Redirects.
1647 If local forwarding is enabled, Router behaviour is assumed.
1648 This means exactly the reverse from the above:
1650 1. IsRouter flag is set in Neighbour Advertisements.
1651 2. Router Solicitations are not sent unless accept_ra is 2.
1652 3. Router Advertisements are ignored unless accept_ra is 2.
1653 4. Redirects are ignored.
1655 Default: 0 (disabled) if global forwarding is disabled (default),
1656 otherwise 1 (enabled).
1659 Default Hop Limit to set.
1663 Default Maximum Transfer Unit
1664 Default: 1280 (IPv6 required minimum)
1666 ip_nonlocal_bind - BOOLEAN
1667 If set, allows processes to bind() to non-local IPv6 addresses,
1668 which can be quite useful - but may break some applications.
1671 router_probe_interval - INTEGER
1672 Minimum interval (in seconds) between Router Probing described
1677 router_solicitation_delay - INTEGER
1678 Number of seconds to wait after interface is brought up
1679 before sending Router Solicitations.
1682 router_solicitation_interval - INTEGER
1683 Number of seconds to wait between Router Solicitations.
1686 router_solicitations - INTEGER
1687 Number of Router Solicitations to send until assuming no
1688 routers are present.
1691 use_oif_addrs_only - BOOLEAN
1692 When enabled, the candidate source addresses for destinations
1693 routed via this interface are restricted to the set of addresses
1694 configured on this interface (vis. RFC 6724, section 4).
1698 use_tempaddr - INTEGER
1699 Preference for Privacy Extensions (RFC3041).
1700 <= 0 : disable Privacy Extensions
1701 == 1 : enable Privacy Extensions, but prefer public
1702 addresses over temporary addresses.
1703 > 1 : enable Privacy Extensions and prefer temporary
1704 addresses over public addresses.
1705 Default: 0 (for most devices)
1706 -1 (for point-to-point devices and loopback devices)
1708 temp_valid_lft - INTEGER
1709 valid lifetime (in seconds) for temporary addresses.
1710 Default: 604800 (7 days)
1712 temp_prefered_lft - INTEGER
1713 Preferred lifetime (in seconds) for temporary addresses.
1714 Default: 86400 (1 day)
1716 keep_addr_on_down - INTEGER
1717 Keep all IPv6 addresses on an interface down event. If set static
1718 global addresses with no expiration time are not flushed.
1723 Default: 0 (addresses are removed)
1725 max_desync_factor - INTEGER
1726 Maximum value for DESYNC_FACTOR, which is a random value
1727 that ensures that clients don't synchronize with each
1728 other and generate new addresses at exactly the same time.
1729 value is in seconds.
1732 regen_max_retry - INTEGER
1733 Number of attempts before give up attempting to generate
1734 valid temporary addresses.
1737 max_addresses - INTEGER
1738 Maximum number of autoconfigured addresses per interface. Setting
1739 to zero disables the limitation. It is not recommended to set this
1740 value too large (or to zero) because it would be an easy way to
1741 crash the kernel by allowing too many addresses to be created.
1744 disable_ipv6 - BOOLEAN
1745 Disable IPv6 operation. If accept_dad is set to 2, this value
1746 will be dynamically set to TRUE if DAD fails for the link-local
1748 Default: FALSE (enable IPv6 operation)
1750 When this value is changed from 1 to 0 (IPv6 is being enabled),
1751 it will dynamically create a link-local address on the given
1752 interface and start Duplicate Address Detection, if necessary.
1754 When this value is changed from 0 to 1 (IPv6 is being disabled),
1755 it will dynamically delete all addresses and routes on the given
1756 interface. From now on it will not possible to add addresses/routes
1757 to the selected interface.
1759 accept_dad - INTEGER
1760 Whether to accept DAD (Duplicate Address Detection).
1762 1: Enable DAD (default)
1763 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1764 link-local address has been found.
1766 DAD operation and mode on a given interface will be selected according
1767 to the maximum value of conf/{all,interface}/accept_dad.
1769 force_tllao - BOOLEAN
1770 Enable sending the target link-layer address option even when
1771 responding to a unicast neighbor solicitation.
1774 Quoting from RFC 2461, section 4.4, Target link-layer address:
1776 "The option MUST be included for multicast solicitations in order to
1777 avoid infinite Neighbor Solicitation "recursion" when the peer node
1778 does not have a cache entry to return a Neighbor Advertisements
1779 message. When responding to unicast solicitations, the option can be
1780 omitted since the sender of the solicitation has the correct link-
1781 layer address; otherwise it would not have be able to send the unicast
1782 solicitation in the first place. However, including the link-layer
1783 address in this case adds little overhead and eliminates a potential
1784 race condition where the sender deletes the cached link-layer address
1785 prior to receiving a response to a previous solicitation."
1787 ndisc_notify - BOOLEAN
1788 Define mode for notification of address and device changes.
1789 0 - (default): do nothing
1790 1 - Generate unsolicited neighbour advertisements when device is brought
1791 up or hardware address changes.
1793 ndisc_tclass - INTEGER
1794 The IPv6 Traffic Class to use by default when sending IPv6 Neighbor
1795 Discovery (Router Solicitation, Router Advertisement, Neighbor
1796 Solicitation, Neighbor Advertisement, Redirect) messages.
1797 These 8 bits can be interpreted as 6 high order bits holding the DSCP
1798 value and 2 low order bits representing ECN (which you probably want
1802 mldv1_unsolicited_report_interval - INTEGER
1803 The interval in milliseconds in which the next unsolicited
1804 MLDv1 report retransmit will take place.
1805 Default: 10000 (10 seconds)
1807 mldv2_unsolicited_report_interval - INTEGER
1808 The interval in milliseconds in which the next unsolicited
1809 MLDv2 report retransmit will take place.
1810 Default: 1000 (1 second)
1812 force_mld_version - INTEGER
1813 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
1814 1 - Enforce to use MLD version 1
1815 2 - Enforce to use MLD version 2
1817 suppress_frag_ndisc - INTEGER
1818 Control RFC 6980 (Security Implications of IPv6 Fragmentation
1819 with IPv6 Neighbor Discovery) behavior:
1820 1 - (default) discard fragmented neighbor discovery packets
1821 0 - allow fragmented neighbor discovery packets
1823 optimistic_dad - BOOLEAN
1824 Whether to perform Optimistic Duplicate Address Detection (RFC 4429).
1825 0: disabled (default)
1828 Optimistic Duplicate Address Detection for the interface will be enabled
1829 if at least one of conf/{all,interface}/optimistic_dad is set to 1,
1830 it will be disabled otherwise.
1832 use_optimistic - BOOLEAN
1833 If enabled, do not classify optimistic addresses as deprecated during
1834 source address selection. Preferred addresses will still be chosen
1835 before optimistic addresses, subject to other ranking in the source
1836 address selection algorithm.
1837 0: disabled (default)
1840 This will be enabled if at least one of
1841 conf/{all,interface}/use_optimistic is set to 1, disabled otherwise.
1843 stable_secret - IPv6 address
1844 This IPv6 address will be used as a secret to generate IPv6
1845 addresses for link-local addresses and autoconfigured
1846 ones. All addresses generated after setting this secret will
1847 be stable privacy ones by default. This can be changed via the
1848 addrgenmode ip-link. conf/default/stable_secret is used as the
1849 secret for the namespace, the interface specific ones can
1850 overwrite that. Writes to conf/all/stable_secret are refused.
1852 It is recommended to generate this secret during installation
1853 of a system and keep it stable after that.
1855 By default the stable secret is unset.
1857 addr_gen_mode - INTEGER
1858 Defines how link-local and autoconf addresses are generated.
1860 0: generate address based on EUI64 (default)
1861 1: do no generate a link-local address, use EUI64 for addresses generated
1863 2: generate stable privacy addresses, using the secret from
1864 stable_secret (RFC7217)
1865 3: generate stable privacy addresses, using a random secret if unset
1867 drop_unicast_in_l2_multicast - BOOLEAN
1868 Drop any unicast IPv6 packets that are received in link-layer
1869 multicast (or broadcast) frames.
1871 By default this is turned off.
1873 drop_unsolicited_na - BOOLEAN
1874 Drop all unsolicited neighbor advertisements, for example if there's
1875 a known good NA proxy on the network and such frames need not be used
1876 (or in the case of 802.11, must not be used to prevent attacks.)
1878 By default this is turned off.
1880 enhanced_dad - BOOLEAN
1881 Include a nonce option in the IPv6 neighbor solicitation messages used for
1882 duplicate address detection per RFC7527. A received DAD NS will only signal
1883 a duplicate address if the nonce is different. This avoids any false
1884 detection of duplicates due to loopback of the NS messages that we send.
1885 The nonce option will be sent on an interface unless both of
1886 conf/{all,interface}/enhanced_dad are set to FALSE.
1891 Limit the maximal rates for sending ICMPv6 packets.
1892 0 to disable any limiting,
1893 otherwise the minimal space between responses in milliseconds.
1896 echo_ignore_all - BOOLEAN
1897 If set non-zero, then the kernel will ignore all ICMP ECHO
1898 requests sent to it over the IPv6 protocol.
1901 xfrm6_gc_thresh - INTEGER
1902 The threshold at which we will start garbage collecting for IPv6
1903 destination cache entries. At twice this value the system will
1904 refuse new allocations.
1908 Pekka Savola <pekkas@netcore.fi>
1909 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1912 /proc/sys/net/bridge/* Variables:
1914 bridge-nf-call-arptables - BOOLEAN
1915 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1919 bridge-nf-call-iptables - BOOLEAN
1920 1 : pass bridged IPv4 traffic to iptables' chains.
1924 bridge-nf-call-ip6tables - BOOLEAN
1925 1 : pass bridged IPv6 traffic to ip6tables' chains.
1929 bridge-nf-filter-vlan-tagged - BOOLEAN
1930 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1934 bridge-nf-filter-pppoe-tagged - BOOLEAN
1935 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1939 bridge-nf-pass-vlan-input-dev - BOOLEAN
1940 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
1941 interface on the bridge and set the netfilter input device to the vlan.
1942 This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
1943 target work with vlan-on-top-of-bridge interfaces. When no matching
1944 vlan interface is found, or this switch is off, the input device is
1945 set to the bridge interface.
1946 0: disable bridge netfilter vlan interface lookup.
1949 proc/sys/net/sctp/* Variables:
1951 addip_enable - BOOLEAN
1952 Enable or disable extension of Dynamic Address Reconfiguration
1953 (ADD-IP) functionality specified in RFC5061. This extension provides
1954 the ability to dynamically add and remove new addresses for the SCTP
1957 1: Enable extension.
1959 0: Disable extension.
1964 Enable or disable pf (pf is short for potentially failed) state. A value
1965 of pf_retrans > path_max_retrans also disables pf state. That is, one of
1966 both pf_enable and pf_retrans > path_max_retrans can disable pf state.
1967 Since pf_retrans and path_max_retrans can be changed by userspace
1968 application, sometimes user expects to disable pf state by the value of
1969 pf_retrans > path_max_retrans, but occasionally the value of pf_retrans
1970 or path_max_retrans is changed by the user application, this pf state is
1971 enabled. As such, it is necessary to add this to dynamically enable
1972 and disable pf state. See:
1973 https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for
1982 addip_noauth_enable - BOOLEAN
1983 Dynamic Address Reconfiguration (ADD-IP) requires the use of
1984 authentication to protect the operations of adding or removing new
1985 addresses. This requirement is mandated so that unauthorized hosts
1986 would not be able to hijack associations. However, older
1987 implementations may not have implemented this requirement while
1988 allowing the ADD-IP extension. For reasons of interoperability,
1989 we provide this variable to control the enforcement of the
1990 authentication requirement.
1992 1: Allow ADD-IP extension to be used without authentication. This
1993 should only be set in a closed environment for interoperability
1994 with older implementations.
1996 0: Enforce the authentication requirement
2000 auth_enable - BOOLEAN
2001 Enable or disable Authenticated Chunks extension. This extension
2002 provides the ability to send and receive authenticated chunks and is
2003 required for secure operation of Dynamic Address Reconfiguration
2006 1: Enable this extension.
2007 0: Disable this extension.
2011 prsctp_enable - BOOLEAN
2012 Enable or disable the Partial Reliability extension (RFC3758) which
2013 is used to notify peers that a given DATA should no longer be expected.
2021 The limit of the number of new packets that can be initially sent. It
2022 controls how bursty the generated traffic can be.
2026 association_max_retrans - INTEGER
2027 Set the maximum number for retransmissions that an association can
2028 attempt deciding that the remote end is unreachable. If this value
2029 is exceeded, the association is terminated.
2033 max_init_retransmits - INTEGER
2034 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
2035 that an association will attempt before declaring the destination
2036 unreachable and terminating.
2040 path_max_retrans - INTEGER
2041 The maximum number of retransmissions that will be attempted on a given
2042 path. Once this threshold is exceeded, the path is considered
2043 unreachable, and new traffic will use a different path when the
2044 association is multihomed.
2048 pf_retrans - INTEGER
2049 The number of retransmissions that will be attempted on a given path
2050 before traffic is redirected to an alternate transport (should one
2051 exist). Note this is distinct from path_max_retrans, as a path that
2052 passes the pf_retrans threshold can still be used. Its only
2053 deprioritized when a transmission path is selected by the stack. This
2054 setting is primarily used to enable fast failover mechanisms without
2055 having to reduce path_max_retrans to a very low value. See:
2056 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
2057 for details. Note also that a value of pf_retrans > path_max_retrans
2058 disables this feature. Since both pf_retrans and path_max_retrans can
2059 be changed by userspace application, a variable pf_enable is used to
2064 rto_initial - INTEGER
2065 The initial round trip timeout value in milliseconds that will be used
2066 in calculating round trip times. This is the initial time interval
2067 for retransmissions.
2072 The maximum value (in milliseconds) of the round trip timeout. This
2073 is the largest time interval that can elapse between retransmissions.
2078 The minimum value (in milliseconds) of the round trip timeout. This
2079 is the smallest time interval the can elapse between retransmissions.
2083 hb_interval - INTEGER
2084 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
2085 are sent at the specified interval on idle paths to probe the state of
2086 a given path between 2 associations.
2090 sack_timeout - INTEGER
2091 The amount of time (in milliseconds) that the implementation will wait
2096 valid_cookie_life - INTEGER
2097 The default lifetime of the SCTP cookie (in milliseconds). The cookie
2098 is used during association establishment.
2102 cookie_preserve_enable - BOOLEAN
2103 Enable or disable the ability to extend the lifetime of the SCTP cookie
2104 that is used during the establishment phase of SCTP association
2106 1: Enable cookie lifetime extension.
2111 cookie_hmac_alg - STRING
2112 Select the hmac algorithm used when generating the cookie value sent by
2113 a listening sctp socket to a connecting client in the INIT-ACK chunk.
2118 Ability to assign md5 or sha1 as the selected alg is predicated on the
2119 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
2120 CONFIG_CRYPTO_SHA1).
2122 Default: Dependent on configuration. MD5 if available, else SHA1 if
2123 available, else none.
2125 rcvbuf_policy - INTEGER
2126 Determines if the receive buffer is attributed to the socket or to
2127 association. SCTP supports the capability to create multiple
2128 associations on a single socket. When using this capability, it is
2129 possible that a single stalled association that's buffering a lot
2130 of data may block other associations from delivering their data by
2131 consuming all of the receive buffer space. To work around this,
2132 the rcvbuf_policy could be set to attribute the receiver buffer space
2133 to each association instead of the socket. This prevents the described
2136 1: rcvbuf space is per association
2137 0: rcvbuf space is per socket
2141 sndbuf_policy - INTEGER
2142 Similar to rcvbuf_policy above, this applies to send buffer space.
2144 1: Send buffer is tracked per association
2145 0: Send buffer is tracked per socket.
2149 sctp_mem - vector of 3 INTEGERs: min, pressure, max
2150 Number of pages allowed for queueing by all SCTP sockets.
2152 min: Below this number of pages SCTP is not bothered about its
2153 memory appetite. When amount of memory allocated by SCTP exceeds
2154 this number, SCTP starts to moderate memory usage.
2156 pressure: This value was introduced to follow format of tcp_mem.
2158 max: Number of pages allowed for queueing by all SCTP sockets.
2160 Default is calculated at boot time from amount of available memory.
2162 sctp_rmem - vector of 3 INTEGERs: min, default, max
2163 Only the first value ("min") is used, "default" and "max" are
2166 min: Minimal size of receive buffer used by SCTP socket.
2167 It is guaranteed to each SCTP socket (but not association) even
2168 under moderate memory pressure.
2172 sctp_wmem - vector of 3 INTEGERs: min, default, max
2173 Only the first value ("min") is used, "default" and "max" are
2176 min: Minimum size of send buffer that can be used by SCTP sockets.
2177 It is guaranteed to each SCTP socket (but not association) even
2178 under moderate memory pressure.
2182 addr_scope_policy - INTEGER
2183 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
2185 0 - Disable IPv4 address scoping
2186 1 - Enable IPv4 address scoping
2187 2 - Follow draft but allow IPv4 private addresses
2188 3 - Follow draft but allow IPv4 link local addresses
2193 /proc/sys/net/core/*
2194 Please see: Documentation/sysctl/net.txt for descriptions of these entries.
2197 /proc/sys/net/unix/*
2198 max_dgram_qlen - INTEGER
2199 The maximum length of dgram socket receive queue