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 fib_sync_mem - UNSIGNED INTEGER
85 Amount of dirty memory from fib entries that can be backlogged before
86 synchronize_rcu is forced.
87 Default: 512kB Minimum: 64kB Maximum: 64MB
89 ip_forward_update_priority - INTEGER
90 Whether to update SKB priority from "TOS" field in IPv4 header after it
91 is forwarded. The new SKB priority is mapped from TOS field value
92 according to an rt_tos2priority table (see e.g. man tc-prio).
93 Default: 1 (Update priority.)
95 0 - Do not update priority.
98 route/max_size - INTEGER
99 Maximum number of routes allowed in the kernel. Increase
100 this when using large numbers of interfaces and/or routes.
101 From linux kernel 3.6 onwards, this is deprecated for ipv4
102 as route cache is no longer used.
104 neigh/default/gc_thresh1 - INTEGER
105 Minimum number of entries to keep. Garbage collector will not
106 purge entries if there are fewer than this number.
109 neigh/default/gc_thresh2 - INTEGER
110 Threshold when garbage collector becomes more aggressive about
111 purging entries. Entries older than 5 seconds will be cleared
112 when over this number.
115 neigh/default/gc_thresh3 - INTEGER
116 Maximum number of non-PERMANENT neighbor entries allowed. Increase
117 this when using large numbers of interfaces and when communicating
118 with large numbers of directly-connected peers.
121 neigh/default/unres_qlen_bytes - INTEGER
122 The maximum number of bytes which may be used by packets
123 queued for each unresolved address by other network layers.
125 Setting negative value is meaningless and will return error.
126 Default: SK_WMEM_MAX, (same as net.core.wmem_default).
127 Exact value depends on architecture and kernel options,
128 but should be enough to allow queuing 256 packets
131 neigh/default/unres_qlen - INTEGER
132 The maximum number of packets which may be queued for each
133 unresolved address by other network layers.
134 (deprecated in linux 3.3) : use unres_qlen_bytes instead.
135 Prior to linux 3.3, the default value is 3 which may cause
136 unexpected packet loss. The current default value is calculated
137 according to default value of unres_qlen_bytes and true size of
141 mtu_expires - INTEGER
142 Time, in seconds, that cached PMTU information is kept.
144 min_adv_mss - INTEGER
145 The advertised MSS depends on the first hop route MTU, but will
146 never be lower than this setting.
150 ipfrag_high_thresh - LONG INTEGER
151 Maximum memory used to reassemble IP fragments.
153 ipfrag_low_thresh - LONG INTEGER
154 (Obsolete since linux-4.17)
155 Maximum memory used to reassemble IP fragments before the kernel
156 begins to remove incomplete fragment queues to free up resources.
157 The kernel still accepts new fragments for defragmentation.
159 ipfrag_time - INTEGER
160 Time in seconds to keep an IP fragment in memory.
162 ipfrag_max_dist - INTEGER
163 ipfrag_max_dist is a non-negative integer value which defines the
164 maximum "disorder" which is allowed among fragments which share a
165 common IP source address. Note that reordering of packets is
166 not unusual, but if a large number of fragments arrive from a source
167 IP address while a particular fragment queue remains incomplete, it
168 probably indicates that one or more fragments belonging to that queue
169 have been lost. When ipfrag_max_dist is positive, an additional check
170 is done on fragments before they are added to a reassembly queue - if
171 ipfrag_max_dist (or more) fragments have arrived from a particular IP
172 address between additions to any IP fragment queue using that source
173 address, it's presumed that one or more fragments in the queue are
174 lost. The existing fragment queue will be dropped, and a new one
175 started. An ipfrag_max_dist value of zero disables this check.
177 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
178 result in unnecessarily dropping fragment queues when normal
179 reordering of packets occurs, which could lead to poor application
180 performance. Using a very large value, e.g. 50000, increases the
181 likelihood of incorrectly reassembling IP fragments that originate
182 from different IP datagrams, which could result in data corruption.
187 inet_peer_threshold - INTEGER
188 The approximate size of the storage. Starting from this threshold
189 entries will be thrown aggressively. This threshold also determines
190 entries' time-to-live and time intervals between garbage collection
191 passes. More entries, less time-to-live, less GC interval.
193 inet_peer_minttl - INTEGER
194 Minimum time-to-live of entries. Should be enough to cover fragment
195 time-to-live on the reassembling side. This minimum time-to-live is
196 guaranteed if the pool size is less than inet_peer_threshold.
199 inet_peer_maxttl - INTEGER
200 Maximum time-to-live of entries. Unused entries will expire after
201 this period of time if there is no memory pressure on the pool (i.e.
202 when the number of entries in the pool is very small).
208 Limit of socket listen() backlog, known in userspace as SOMAXCONN.
209 Defaults to 128. See also tcp_max_syn_backlog for additional tuning
212 tcp_abort_on_overflow - BOOLEAN
213 If listening service is too slow to accept new connections,
214 reset them. Default state is FALSE. It means that if overflow
215 occurred due to a burst, connection will recover. Enable this
216 option _only_ if you are really sure that listening daemon
217 cannot be tuned to accept connections faster. Enabling this
218 option can harm clients of your server.
220 tcp_adv_win_scale - INTEGER
221 Count buffering overhead as bytes/2^tcp_adv_win_scale
222 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
224 Possible values are [-31, 31], inclusive.
227 tcp_allowed_congestion_control - STRING
228 Show/set the congestion control choices available to non-privileged
229 processes. The list is a subset of those listed in
230 tcp_available_congestion_control.
231 Default is "reno" and the default setting (tcp_congestion_control).
233 tcp_app_win - INTEGER
234 Reserve max(window/2^tcp_app_win, mss) of window for application
235 buffer. Value 0 is special, it means that nothing is reserved.
238 tcp_autocorking - BOOLEAN
239 Enable TCP auto corking :
240 When applications do consecutive small write()/sendmsg() system calls,
241 we try to coalesce these small writes as much as possible, to lower
242 total amount of sent packets. This is done if at least one prior
243 packet for the flow is waiting in Qdisc queues or device transmit
244 queue. Applications can still use TCP_CORK for optimal behavior
245 when they know how/when to uncork their sockets.
248 tcp_available_congestion_control - STRING
249 Shows the available congestion control choices that are registered.
250 More congestion control algorithms may be available as modules,
253 tcp_base_mss - INTEGER
254 The initial value of search_low to be used by the packetization layer
255 Path MTU discovery (MTU probing). If MTU probing is enabled,
256 this is the initial MSS used by the connection.
258 tcp_min_snd_mss - INTEGER
259 TCP SYN and SYNACK messages usually advertise an ADVMSS option,
260 as described in RFC 1122 and RFC 6691.
261 If this ADVMSS option is smaller than tcp_min_snd_mss,
262 it is silently capped to tcp_min_snd_mss.
264 Default : 48 (at least 8 bytes of payload per segment)
266 tcp_congestion_control - STRING
267 Set the congestion control algorithm to be used for new
268 connections. The algorithm "reno" is always available, but
269 additional choices may be available based on kernel configuration.
270 Default is set as part of kernel configuration.
271 For passive connections, the listener congestion control choice
273 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
276 Allows TCP to send "duplicate" SACKs.
278 tcp_early_retrans - INTEGER
279 Tail loss probe (TLP) converts RTOs occurring due to tail
280 losses into fast recovery (draft-ietf-tcpm-rack). Note that
281 TLP requires RACK to function properly (see tcp_recovery below)
288 Control use of Explicit Congestion Notification (ECN) by TCP.
289 ECN is used only when both ends of the TCP connection indicate
290 support for it. This feature is useful in avoiding losses due
291 to congestion by allowing supporting routers to signal
292 congestion before having to drop packets.
294 0 Disable ECN. Neither initiate nor accept ECN.
295 1 Enable ECN when requested by incoming connections and
296 also request ECN on outgoing connection attempts.
297 2 Enable ECN when requested by incoming connections
298 but do not request ECN on outgoing connections.
301 tcp_ecn_fallback - BOOLEAN
302 If the kernel detects that ECN connection misbehaves, enable fall
303 back to non-ECN. Currently, this knob implements the fallback
304 from RFC3168, section 6.1.1.1., but we reserve that in future,
305 additional detection mechanisms could be implemented under this
306 knob. The value is not used, if tcp_ecn or per route (or congestion
307 control) ECN settings are disabled.
308 Default: 1 (fallback enabled)
311 This is a legacy option, it has no effect anymore.
313 tcp_fin_timeout - INTEGER
314 The length of time an orphaned (no longer referenced by any
315 application) connection will remain in the FIN_WAIT_2 state
316 before it is aborted at the local end. While a perfectly
317 valid "receive only" state for an un-orphaned connection, an
318 orphaned connection in FIN_WAIT_2 state could otherwise wait
319 forever for the remote to close its end of the connection.
324 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.
325 F-RTO is an enhanced recovery algorithm for TCP retransmission
326 timeouts. It is particularly beneficial in networks where the
327 RTT fluctuates (e.g., wireless). F-RTO is sender-side only
328 modification. It does not require any support from the peer.
330 By default it's enabled with a non-zero value. 0 disables F-RTO.
332 tcp_fwmark_accept - BOOLEAN
333 If set, incoming connections to listening sockets that do not have a
334 socket mark will set the mark of the accepting socket to the fwmark of
335 the incoming SYN packet. This will cause all packets on that connection
336 (starting from the first SYNACK) to be sent with that fwmark. The
337 listening socket's mark is unchanged. Listening sockets that already
338 have a fwmark set via setsockopt(SOL_SOCKET, SO_MARK, ...) are
343 tcp_invalid_ratelimit - INTEGER
344 Limit the maximal rate for sending duplicate acknowledgments
345 in response to incoming TCP packets that are for an existing
346 connection but that are invalid due to any of these reasons:
348 (a) out-of-window sequence number,
349 (b) out-of-window acknowledgment number, or
350 (c) PAWS (Protection Against Wrapped Sequence numbers) check failure
352 This can help mitigate simple "ack loop" DoS attacks, wherein
353 a buggy or malicious middlebox or man-in-the-middle can
354 rewrite TCP header fields in manner that causes each endpoint
355 to think that the other is sending invalid TCP segments, thus
356 causing each side to send an unterminating stream of duplicate
357 acknowledgments for invalid segments.
359 Using 0 disables rate-limiting of dupacks in response to
360 invalid segments; otherwise this value specifies the minimal
361 space between sending such dupacks, in milliseconds.
363 Default: 500 (milliseconds).
365 tcp_keepalive_time - INTEGER
366 How often TCP sends out keepalive messages when keepalive is enabled.
369 tcp_keepalive_probes - INTEGER
370 How many keepalive probes TCP sends out, until it decides that the
371 connection is broken. Default value: 9.
373 tcp_keepalive_intvl - INTEGER
374 How frequently the probes are send out. Multiplied by
375 tcp_keepalive_probes it is time to kill not responding connection,
376 after probes started. Default value: 75sec i.e. connection
377 will be aborted after ~11 minutes of retries.
379 tcp_l3mdev_accept - BOOLEAN
380 Enables child sockets to inherit the L3 master device index.
381 Enabling this option allows a "global" listen socket to work
382 across L3 master domains (e.g., VRFs) with connected sockets
383 derived from the listen socket to be bound to the L3 domain in
384 which the packets originated. Only valid when the kernel was
385 compiled with CONFIG_NET_L3_MASTER_DEV.
386 Default: 0 (disabled)
388 tcp_low_latency - BOOLEAN
389 This is a legacy option, it has no effect anymore.
391 tcp_max_orphans - INTEGER
392 Maximal number of TCP sockets not attached to any user file handle,
393 held by system. If this number is exceeded orphaned connections are
394 reset immediately and warning is printed. This limit exists
395 only to prevent simple DoS attacks, you _must_ not rely on this
396 or lower the limit artificially, but rather increase it
397 (probably, after increasing installed memory),
398 if network conditions require more than default value,
399 and tune network services to linger and kill such states
400 more aggressively. Let me to remind again: each orphan eats
401 up to ~64K of unswappable memory.
403 tcp_max_syn_backlog - INTEGER
404 Maximal number of remembered connection requests, which have not
405 received an acknowledgment from connecting client.
406 The minimal value is 128 for low memory machines, and it will
407 increase in proportion to the memory of machine.
408 If server suffers from overload, try increasing this number.
410 tcp_max_tw_buckets - INTEGER
411 Maximal number of timewait sockets held by system simultaneously.
412 If this number is exceeded time-wait socket is immediately destroyed
413 and warning is printed. This limit exists only to prevent
414 simple DoS attacks, you _must_ not lower the limit artificially,
415 but rather increase it (probably, after increasing installed memory),
416 if network conditions require more than default value.
418 tcp_mem - vector of 3 INTEGERs: min, pressure, max
419 min: below this number of pages TCP is not bothered about its
422 pressure: when amount of memory allocated by TCP exceeds this number
423 of pages, TCP moderates its memory consumption and enters memory
424 pressure mode, which is exited when memory consumption falls
427 max: number of pages allowed for queueing by all TCP sockets.
429 Defaults are calculated at boot time from amount of available
432 tcp_min_rtt_wlen - INTEGER
433 The window length of the windowed min filter to track the minimum RTT.
434 A shorter window lets a flow more quickly pick up new (higher)
435 minimum RTT when it is moved to a longer path (e.g., due to traffic
436 engineering). A longer window makes the filter more resistant to RTT
437 inflations such as transient congestion. The unit is seconds.
438 Possible values: 0 - 86400 (1 day)
441 tcp_moderate_rcvbuf - BOOLEAN
442 If set, TCP performs receive buffer auto-tuning, attempting to
443 automatically size the buffer (no greater than tcp_rmem[2]) to
444 match the size required by the path for full throughput. Enabled by
447 tcp_mtu_probing - INTEGER
448 Controls TCP Packetization-Layer Path MTU Discovery. Takes three
451 1 - Disabled by default, enabled when an ICMP black hole detected
452 2 - Always enabled, use initial MSS of tcp_base_mss.
454 tcp_probe_interval - UNSIGNED INTEGER
455 Controls how often to start TCP Packetization-Layer Path MTU
456 Discovery reprobe. The default is reprobing every 10 minutes as
459 tcp_probe_threshold - INTEGER
460 Controls when TCP Packetization-Layer Path MTU Discovery probing
461 will stop in respect to the width of search range in bytes. Default
464 tcp_no_metrics_save - BOOLEAN
465 By default, TCP saves various connection metrics in the route cache
466 when the connection closes, so that connections established in the
467 near future can use these to set initial conditions. Usually, this
468 increases overall performance, but may sometimes cause performance
469 degradation. If set, TCP will not cache metrics on closing
472 tcp_orphan_retries - INTEGER
473 This value influences the timeout of a locally closed TCP connection,
474 when RTO retransmissions remain unacknowledged.
475 See tcp_retries2 for more details.
477 The default value is 8.
478 If your machine is a loaded WEB server,
479 you should think about lowering this value, such sockets
480 may consume significant resources. Cf. tcp_max_orphans.
482 tcp_recovery - INTEGER
483 This value is a bitmap to enable various experimental loss recovery
486 RACK: 0x1 enables the RACK loss detection for fast detection of lost
487 retransmissions and tail drops. It also subsumes and disables
488 RFC6675 recovery for SACK connections.
489 RACK: 0x2 makes RACK's reordering window static (min_rtt/4).
490 RACK: 0x4 disables RACK's DUPACK threshold heuristic
494 tcp_reordering - INTEGER
495 Initial reordering level of packets in a TCP stream.
496 TCP stack can then dynamically adjust flow reordering level
497 between this initial value and tcp_max_reordering
500 tcp_max_reordering - INTEGER
501 Maximal reordering level of packets in a TCP stream.
502 300 is a fairly conservative value, but you might increase it
503 if paths are using per packet load balancing (like bonding rr mode)
506 tcp_retrans_collapse - BOOLEAN
507 Bug-to-bug compatibility with some broken printers.
508 On retransmit try to send bigger packets to work around bugs in
511 tcp_retries1 - INTEGER
512 This value influences the time, after which TCP decides, that
513 something is wrong due to unacknowledged RTO retransmissions,
514 and reports this suspicion to the network layer.
515 See tcp_retries2 for more details.
517 RFC 1122 recommends at least 3 retransmissions, which is the
520 tcp_retries2 - INTEGER
521 This value influences the timeout of an alive TCP connection,
522 when RTO retransmissions remain unacknowledged.
523 Given a value of N, a hypothetical TCP connection following
524 exponential backoff with an initial RTO of TCP_RTO_MIN would
525 retransmit N times before killing the connection at the (N+1)th RTO.
527 The default value of 15 yields a hypothetical timeout of 924.6
528 seconds and is a lower bound for the effective timeout.
529 TCP will effectively time out at the first RTO which exceeds the
530 hypothetical timeout.
532 RFC 1122 recommends at least 100 seconds for the timeout,
533 which corresponds to a value of at least 8.
535 tcp_rfc1337 - BOOLEAN
536 If set, the TCP stack behaves conforming to RFC1337. If unset,
537 we are not conforming to RFC, but prevent TCP TIME_WAIT
541 tcp_rmem - vector of 3 INTEGERs: min, default, max
542 min: Minimal size of receive buffer used by TCP sockets.
543 It is guaranteed to each TCP socket, even under moderate memory
547 default: initial size of receive buffer used by TCP sockets.
548 This value overrides net.core.rmem_default used by other protocols.
549 Default: 87380 bytes. This value results in window of 65535 with
550 default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
551 less for default tcp_app_win. See below about these variables.
553 max: maximal size of receive buffer allowed for automatically
554 selected receiver buffers for TCP socket. This value does not override
555 net.core.rmem_max. Calling setsockopt() with SO_RCVBUF disables
556 automatic tuning of that socket's receive buffer size, in which
557 case this value is ignored.
558 Default: between 87380B and 6MB, depending on RAM size.
561 Enable select acknowledgments (SACKS).
563 tcp_comp_sack_delay_ns - LONG INTEGER
564 TCP tries to reduce number of SACK sent, using a timer
565 based on 5% of SRTT, capped by this sysctl, in nano seconds.
566 The default is 1ms, based on TSO autosizing period.
568 Default : 1,000,000 ns (1 ms)
570 tcp_comp_sack_nr - INTEGER
571 Max number of SACK that can be compressed.
572 Using 0 disables SACK compression.
576 tcp_slow_start_after_idle - BOOLEAN
577 If set, provide RFC2861 behavior and time out the congestion
578 window after an idle period. An idle period is defined at
579 the current RTO. If unset, the congestion window will not
580 be timed out after an idle period.
584 Use the Host requirements interpretation of the TCP urgent pointer field.
585 Most hosts use the older BSD interpretation, so if you turn this on
586 Linux might not communicate correctly with them.
589 tcp_synack_retries - INTEGER
590 Number of times SYNACKs for a passive TCP connection attempt will
591 be retransmitted. Should not be higher than 255. Default value
592 is 5, which corresponds to 31seconds till the last retransmission
593 with the current initial RTO of 1second. With this the final timeout
594 for a passive TCP connection will happen after 63seconds.
596 tcp_syncookies - BOOLEAN
597 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES
598 Send out syncookies when the syn backlog queue of a socket
599 overflows. This is to prevent against the common 'SYN flood attack'
602 Note, that syncookies is fallback facility.
603 It MUST NOT be used to help highly loaded servers to stand
604 against legal connection rate. If you see SYN flood warnings
605 in your logs, but investigation shows that they occur
606 because of overload with legal connections, you should tune
607 another parameters until this warning disappear.
608 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
610 syncookies seriously violate TCP protocol, do not allow
611 to use TCP extensions, can result in serious degradation
612 of some services (f.e. SMTP relaying), visible not by you,
613 but your clients and relays, contacting you. While you see
614 SYN flood warnings in logs not being really flooded, your server
615 is seriously misconfigured.
617 If you want to test which effects syncookies have to your
618 network connections you can set this knob to 2 to enable
619 unconditionally generation of syncookies.
621 tcp_fastopen - INTEGER
622 Enable TCP Fast Open (RFC7413) to send and accept data in the opening
625 The client support is enabled by flag 0x1 (on by default). The client
626 then must use sendmsg() or sendto() with the MSG_FASTOPEN flag,
627 rather than connect() to send data in SYN.
629 The server support is enabled by flag 0x2 (off by default). Then
630 either enable for all listeners with another flag (0x400) or
631 enable individual listeners via TCP_FASTOPEN socket option with
632 the option value being the length of the syn-data backlog.
634 The values (bitmap) are
635 0x1: (client) enables sending data in the opening SYN on the client.
636 0x2: (server) enables the server support, i.e., allowing data in
637 a SYN packet to be accepted and passed to the
638 application before 3-way handshake finishes.
639 0x4: (client) send data in the opening SYN regardless of cookie
640 availability and without a cookie option.
641 0x200: (server) accept data-in-SYN w/o any cookie option present.
642 0x400: (server) enable all listeners to support Fast Open by
643 default without explicit TCP_FASTOPEN socket option.
647 Note that that additional client or server features are only
648 effective if the basic support (0x1 and 0x2) are enabled respectively.
650 tcp_fastopen_blackhole_timeout_sec - INTEGER
651 Initial time period in second to disable Fastopen on active TCP sockets
652 when a TFO firewall blackhole issue happens.
653 This time period will grow exponentially when more blackhole issues
654 get detected right after Fastopen is re-enabled and will reset to
655 initial value when the blackhole issue goes away.
656 0 to disable the blackhole detection.
657 By default, it is set to 1hr.
659 tcp_syn_retries - INTEGER
660 Number of times initial SYNs for an active TCP connection attempt
661 will be retransmitted. Should not be higher than 127. Default value
662 is 6, which corresponds to 63seconds till the last retransmission
663 with the current initial RTO of 1second. With this the final timeout
664 for an active TCP connection attempt will happen after 127seconds.
666 tcp_timestamps - INTEGER
667 Enable timestamps as defined in RFC1323.
669 1: Enable timestamps as defined in RFC1323 and use random offset for
670 each connection rather than only using the current time.
671 2: Like 1, but without random offsets.
674 tcp_min_tso_segs - INTEGER
675 Minimal number of segments per TSO frame.
676 Since linux-3.12, TCP does an automatic sizing of TSO frames,
677 depending on flow rate, instead of filling 64Kbytes packets.
678 For specific usages, it's possible to force TCP to build big
679 TSO frames. Note that TCP stack might split too big TSO packets
680 if available window is too small.
683 tcp_pacing_ss_ratio - INTEGER
684 sk->sk_pacing_rate is set by TCP stack using a ratio applied
685 to current rate. (current_rate = cwnd * mss / srtt)
686 If TCP is in slow start, tcp_pacing_ss_ratio is applied
687 to let TCP probe for bigger speeds, assuming cwnd can be
688 doubled every other RTT.
691 tcp_pacing_ca_ratio - INTEGER
692 sk->sk_pacing_rate is set by TCP stack using a ratio applied
693 to current rate. (current_rate = cwnd * mss / srtt)
694 If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio
695 is applied to conservatively probe for bigger throughput.
698 tcp_tso_win_divisor - INTEGER
699 This allows control over what percentage of the congestion window
700 can be consumed by a single TSO frame.
701 The setting of this parameter is a choice between burstiness and
702 building larger TSO frames.
705 tcp_tw_reuse - INTEGER
706 Enable reuse of TIME-WAIT sockets for new connections when it is
707 safe from protocol viewpoint.
710 2 - enable for loopback traffic only
711 It should not be changed without advice/request of technical
715 tcp_window_scaling - BOOLEAN
716 Enable window scaling as defined in RFC1323.
718 tcp_wmem - vector of 3 INTEGERs: min, default, max
719 min: Amount of memory reserved for send buffers for TCP sockets.
720 Each TCP socket has rights to use it due to fact of its birth.
723 default: initial size of send buffer used by TCP sockets. This
724 value overrides net.core.wmem_default used by other protocols.
725 It is usually lower than net.core.wmem_default.
728 max: Maximal amount of memory allowed for automatically tuned
729 send buffers for TCP sockets. This value does not override
730 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables
731 automatic tuning of that socket's send buffer size, in which case
732 this value is ignored.
733 Default: between 64K and 4MB, depending on RAM size.
735 tcp_notsent_lowat - UNSIGNED INTEGER
736 A TCP socket can control the amount of unsent bytes in its write queue,
737 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()
738 reports POLLOUT events if the amount of unsent bytes is below a per
739 socket value, and if the write queue is not full. sendmsg() will
740 also not add new buffers if the limit is hit.
742 This global variable controls the amount of unsent data for
743 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change
744 to the global variable has immediate effect.
746 Default: UINT_MAX (0xFFFFFFFF)
748 tcp_workaround_signed_windows - BOOLEAN
749 If set, assume no receipt of a window scaling option means the
750 remote TCP is broken and treats the window as a signed quantity.
751 If unset, assume the remote TCP is not broken even if we do
752 not receive a window scaling option from them.
755 tcp_thin_linear_timeouts - BOOLEAN
756 Enable dynamic triggering of linear timeouts for thin streams.
757 If set, a check is performed upon retransmission by timeout to
758 determine if the stream is thin (less than 4 packets in flight).
759 As long as the stream is found to be thin, up to 6 linear
760 timeouts may be performed before exponential backoff mode is
761 initiated. This improves retransmission latency for
762 non-aggressive thin streams, often found to be time-dependent.
763 For more information on thin streams, see
764 Documentation/networking/tcp-thin.txt
767 tcp_limit_output_bytes - INTEGER
768 Controls TCP Small Queue limit per tcp socket.
769 TCP bulk sender tends to increase packets in flight until it
770 gets losses notifications. With SNDBUF autotuning, this can
771 result in a large amount of packets queued on the local machine
772 (e.g.: qdiscs, CPU backlog, or device) hurting latency of other
773 flows, for typical pfifo_fast qdiscs. tcp_limit_output_bytes
774 limits the number of bytes on qdisc or device to reduce artificial
775 RTT/cwnd and reduce bufferbloat.
776 Default: 1048576 (16 * 65536)
778 tcp_challenge_ack_limit - INTEGER
779 Limits number of Challenge ACK sent per second, as recommended
780 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
783 tcp_rx_skb_cache - BOOLEAN
784 Controls a per TCP socket cache of one skb, that might help
785 performance of some workloads. This might be dangerous
786 on systems with a lot of TCP sockets, since it increases
789 Default: 0 (disabled)
793 udp_l3mdev_accept - BOOLEAN
794 Enabling this option allows a "global" bound socket to work
795 across L3 master domains (e.g., VRFs) with packets capable of
796 being received regardless of the L3 domain in which they
797 originated. Only valid when the kernel was compiled with
798 CONFIG_NET_L3_MASTER_DEV.
799 Default: 0 (disabled)
801 udp_mem - vector of 3 INTEGERs: min, pressure, max
802 Number of pages allowed for queueing by all UDP sockets.
804 min: Below this number of pages UDP is not bothered about its
805 memory appetite. When amount of memory allocated by UDP exceeds
806 this number, UDP starts to moderate memory usage.
808 pressure: This value was introduced to follow format of tcp_mem.
810 max: Number of pages allowed for queueing by all UDP sockets.
812 Default is calculated at boot time from amount of available memory.
814 udp_rmem_min - INTEGER
815 Minimal size of receive buffer used by UDP sockets in moderation.
816 Each UDP socket is able to use the size for receiving data, even if
817 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
820 udp_wmem_min - INTEGER
821 Minimal size of send buffer used by UDP sockets in moderation.
822 Each UDP socket is able to use the size for sending data, even if
823 total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
828 raw_l3mdev_accept - BOOLEAN
829 Enabling this option allows a "global" bound socket to work
830 across L3 master domains (e.g., VRFs) with packets capable of
831 being received regardless of the L3 domain in which they
832 originated. Only valid when the kernel was compiled with
833 CONFIG_NET_L3_MASTER_DEV.
838 cipso_cache_enable - BOOLEAN
839 If set, enable additions to and lookups from the CIPSO label mapping
840 cache. If unset, additions are ignored and lookups always result in a
841 miss. However, regardless of the setting the cache is still
842 invalidated when required when means you can safely toggle this on and
843 off and the cache will always be "safe".
846 cipso_cache_bucket_size - INTEGER
847 The CIPSO label cache consists of a fixed size hash table with each
848 hash bucket containing a number of cache entries. This variable limits
849 the number of entries in each hash bucket; the larger the value the
850 more CIPSO label mappings that can be cached. When the number of
851 entries in a given hash bucket reaches this limit adding new entries
852 causes the oldest entry in the bucket to be removed to make room.
855 cipso_rbm_optfmt - BOOLEAN
856 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
857 the CIPSO draft specification (see Documentation/netlabel for details).
858 This means that when set the CIPSO tag will be padded with empty
859 categories in order to make the packet data 32-bit aligned.
862 cipso_rbm_structvalid - BOOLEAN
863 If set, do a very strict check of the CIPSO option when
864 ip_options_compile() is called. If unset, relax the checks done during
865 ip_options_compile(). Either way is "safe" as errors are caught else
866 where in the CIPSO processing code but setting this to 0 (False) should
867 result in less work (i.e. it should be faster) but could cause problems
868 with other implementations that require strict checking.
873 ip_local_port_range - 2 INTEGERS
874 Defines the local port range that is used by TCP and UDP to
875 choose the local port. The first number is the first, the
876 second the last local port number.
877 If possible, it is better these numbers have different parity.
878 (one even and one odd values)
879 The default values are 32768 and 60999 respectively.
881 ip_local_reserved_ports - list of comma separated ranges
882 Specify the ports which are reserved for known third-party
883 applications. These ports will not be used by automatic port
884 assignments (e.g. when calling connect() or bind() with port
885 number 0). Explicit port allocation behavior is unchanged.
887 The format used for both input and output is a comma separated
888 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
889 10). Writing to the file will clear all previously reserved
890 ports and update the current list with the one given in the
893 Note that ip_local_port_range and ip_local_reserved_ports
894 settings are independent and both are considered by the kernel
895 when determining which ports are available for automatic port
898 You can reserve ports which are not in the current
899 ip_local_port_range, e.g.:
901 $ cat /proc/sys/net/ipv4/ip_local_port_range
903 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
906 although this is redundant. However such a setting is useful
907 if later the port range is changed to a value that will
908 include the reserved ports.
912 ip_unprivileged_port_start - INTEGER
913 This is a per-namespace sysctl. It defines the first
914 unprivileged port in the network namespace. Privileged ports
915 require root or CAP_NET_BIND_SERVICE in order to bind to them.
916 To disable all privileged ports, set this to 0. It may not
917 overlap with the ip_local_reserved_ports range.
921 ip_nonlocal_bind - BOOLEAN
922 If set, allows processes to bind() to non-local IP addresses,
923 which can be quite useful - but may break some applications.
927 If set non-zero, enables support for dynamic addresses.
928 If set to a non-zero value larger than 1, a kernel log
929 message will be printed when dynamic address rewriting
933 ip_early_demux - BOOLEAN
934 Optimize input packet processing down to one demux for
935 certain kinds of local sockets. Currently we only do this
936 for established TCP and connected UDP sockets.
938 It may add an additional cost for pure routing workloads that
939 reduces overall throughput, in such case you should disable it.
942 tcp_early_demux - BOOLEAN
943 Enable early demux for established TCP sockets.
946 udp_early_demux - BOOLEAN
947 Enable early demux for connected UDP sockets. Disable this if
948 your system could experience more unconnected load.
951 icmp_echo_ignore_all - BOOLEAN
952 If set non-zero, then the kernel will ignore all ICMP ECHO
956 icmp_echo_ignore_broadcasts - BOOLEAN
957 If set non-zero, then the kernel will ignore all ICMP ECHO and
958 TIMESTAMP requests sent to it via broadcast/multicast.
961 icmp_ratelimit - INTEGER
962 Limit the maximal rates for sending ICMP packets whose type matches
963 icmp_ratemask (see below) to specific targets.
964 0 to disable any limiting,
965 otherwise the minimal space between responses in milliseconds.
966 Note that another sysctl, icmp_msgs_per_sec limits the number
967 of ICMP packets sent on all targets.
970 icmp_msgs_per_sec - INTEGER
971 Limit maximal number of ICMP packets sent per second from this host.
972 Only messages whose type matches icmp_ratemask (see below) are
973 controlled by this limit.
976 icmp_msgs_burst - INTEGER
977 icmp_msgs_per_sec controls number of ICMP packets sent per second,
978 while icmp_msgs_burst controls the burst size of these packets.
981 icmp_ratemask - INTEGER
982 Mask made of ICMP types for which rates are being limited.
983 Significant bits: IHGFEDCBA9876543210
984 Default mask: 0000001100000011000 (6168)
986 Bit definitions (see include/linux/icmp.h):
988 3 Destination Unreachable *
993 C Parameter Problem *
998 H Address Mask Request
1001 * These are rate limited by default (see default mask above)
1003 icmp_ignore_bogus_error_responses - BOOLEAN
1004 Some routers violate RFC1122 by sending bogus responses to broadcast
1005 frames. Such violations are normally logged via a kernel warning.
1006 If this is set to TRUE, the kernel will not give such warnings, which
1007 will avoid log file clutter.
1010 icmp_errors_use_inbound_ifaddr - BOOLEAN
1012 If zero, icmp error messages are sent with the primary address of
1013 the exiting interface.
1015 If non-zero, the message will be sent with the primary address of
1016 the interface that received the packet that caused the icmp error.
1017 This is the behaviour network many administrators will expect from
1018 a router. And it can make debugging complicated network layouts
1021 Note that if no primary address exists for the interface selected,
1022 then the primary address of the first non-loopback interface that
1023 has one will be used regardless of this setting.
1027 igmp_max_memberships - INTEGER
1028 Change the maximum number of multicast groups we can subscribe to.
1031 Theoretical maximum value is bounded by having to send a membership
1032 report in a single datagram (i.e. the report can't span multiple
1033 datagrams, or risk confusing the switch and leaving groups you don't
1036 The number of supported groups 'M' is bounded by the number of group
1037 report entries you can fit into a single datagram of 65535 bytes.
1039 M = 65536-sizeof (ip header)/(sizeof(Group record))
1041 Group records are variable length, with a minimum of 12 bytes.
1042 So net.ipv4.igmp_max_memberships should not be set higher than:
1044 (65536-24) / 12 = 5459
1046 The value 5459 assumes no IP header options, so in practice
1047 this number may be lower.
1049 igmp_max_msf - INTEGER
1050 Maximum number of addresses allowed in the source filter list for a
1055 Controls the IGMP query robustness variable (see RFC2236 8.1).
1056 Default: 2 (as specified by RFC2236 8.1)
1057 Minimum: 1 (as specified by RFC6636 4.5)
1059 force_igmp_version - INTEGER
1060 0 - (default) No enforcement of a IGMP version, IGMPv1/v2 fallback
1061 allowed. Will back to IGMPv3 mode again if all IGMPv1/v2 Querier
1062 Present timer expires.
1063 1 - Enforce to use IGMP version 1. Will also reply IGMPv1 report if
1064 receive IGMPv2/v3 query.
1065 2 - Enforce to use IGMP version 2. Will fallback to IGMPv1 if receive
1066 IGMPv1 query message. Will reply report if receive IGMPv3 query.
1067 3 - Enforce to use IGMP version 3. The same react with default 0.
1069 Note: this is not the same with force_mld_version because IGMPv3 RFC3376
1070 Security Considerations does not have clear description that we could
1071 ignore other version messages completely as MLDv2 RFC3810. So make
1072 this value as default 0 is recommended.
1074 conf/interface/* changes special settings per interface (where
1075 "interface" is the name of your network interface)
1077 conf/all/* is special, changes the settings for all interfaces
1079 log_martians - BOOLEAN
1080 Log packets with impossible addresses to kernel log.
1081 log_martians for the interface will be enabled if at least one of
1082 conf/{all,interface}/log_martians is set to TRUE,
1083 it will be disabled otherwise
1085 accept_redirects - BOOLEAN
1086 Accept ICMP redirect messages.
1087 accept_redirects for the interface will be enabled if:
1088 - both conf/{all,interface}/accept_redirects are TRUE in the case
1089 forwarding for the interface is enabled
1091 - at least one of conf/{all,interface}/accept_redirects is TRUE in the
1092 case forwarding for the interface is disabled
1093 accept_redirects for the interface will be disabled otherwise
1097 forwarding - BOOLEAN
1098 Enable IP forwarding on this interface. This controls whether packets
1099 received _on_ this interface can be forwarded.
1101 mc_forwarding - BOOLEAN
1102 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
1103 and a multicast routing daemon is required.
1104 conf/all/mc_forwarding must also be set to TRUE to enable multicast
1105 routing for the interface
1108 Integer value used to differentiate the devices by the medium they
1109 are attached to. Two devices can have different id values when
1110 the broadcast packets are received only on one of them.
1111 The default value 0 means that the device is the only interface
1112 to its medium, value of -1 means that medium is not known.
1114 Currently, it is used to change the proxy_arp behavior:
1115 the proxy_arp feature is enabled for packets forwarded between
1116 two devices attached to different media.
1120 proxy_arp for the interface will be enabled if at least one of
1121 conf/{all,interface}/proxy_arp is set to TRUE,
1122 it will be disabled otherwise
1124 proxy_arp_pvlan - BOOLEAN
1125 Private VLAN proxy arp.
1126 Basically allow proxy arp replies back to the same interface
1127 (from which the ARP request/solicitation was received).
1129 This is done to support (ethernet) switch features, like RFC
1130 3069, where the individual ports are NOT allowed to
1131 communicate with each other, but they are allowed to talk to
1132 the upstream router. As described in RFC 3069, it is possible
1133 to allow these hosts to communicate through the upstream
1134 router by proxy_arp'ing. Don't need to be used together with
1137 This technology is known by different names:
1138 In RFC 3069 it is called VLAN Aggregation.
1139 Cisco and Allied Telesyn call it Private VLAN.
1140 Hewlett-Packard call it Source-Port filtering or port-isolation.
1141 Ericsson call it MAC-Forced Forwarding (RFC Draft).
1143 shared_media - BOOLEAN
1144 Send(router) or accept(host) RFC1620 shared media redirects.
1145 Overrides secure_redirects.
1146 shared_media for the interface will be enabled if at least one of
1147 conf/{all,interface}/shared_media is set to TRUE,
1148 it will be disabled otherwise
1151 secure_redirects - BOOLEAN
1152 Accept ICMP redirect messages only to gateways listed in the
1153 interface's current gateway list. Even if disabled, RFC1122 redirect
1155 Overridden by shared_media.
1156 secure_redirects for the interface will be enabled if at least one of
1157 conf/{all,interface}/secure_redirects is set to TRUE,
1158 it will be disabled otherwise
1161 send_redirects - BOOLEAN
1162 Send redirects, if router.
1163 send_redirects for the interface will be enabled if at least one of
1164 conf/{all,interface}/send_redirects is set to TRUE,
1165 it will be disabled otherwise
1168 bootp_relay - BOOLEAN
1169 Accept packets with source address 0.b.c.d destined
1170 not to this host as local ones. It is supposed, that
1171 BOOTP relay daemon will catch and forward such packets.
1172 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
1175 Not Implemented Yet.
1177 accept_source_route - BOOLEAN
1178 Accept packets with SRR option.
1179 conf/all/accept_source_route must also be set to TRUE to accept packets
1180 with SRR option on the interface
1181 default TRUE (router)
1184 accept_local - BOOLEAN
1185 Accept packets with local source addresses. In combination with
1186 suitable routing, this can be used to direct packets between two
1187 local interfaces over the wire and have them accepted properly.
1190 route_localnet - BOOLEAN
1191 Do not consider loopback addresses as martian source or destination
1192 while routing. This enables the use of 127/8 for local routing purposes.
1196 0 - No source validation.
1197 1 - Strict mode as defined in RFC3704 Strict Reverse Path
1198 Each incoming packet is tested against the FIB and if the interface
1199 is not the best reverse path the packet check will fail.
1200 By default failed packets are discarded.
1201 2 - Loose mode as defined in RFC3704 Loose Reverse Path
1202 Each incoming packet's source address is also tested against the FIB
1203 and if the source address is not reachable via any interface
1204 the packet check will fail.
1206 Current recommended practice in RFC3704 is to enable strict mode
1207 to prevent IP spoofing from DDos attacks. If using asymmetric routing
1208 or other complicated routing, then loose mode is recommended.
1210 The max value from conf/{all,interface}/rp_filter is used
1211 when doing source validation on the {interface}.
1213 Default value is 0. Note that some distributions enable it
1216 arp_filter - BOOLEAN
1217 1 - Allows you to have multiple network interfaces on the same
1218 subnet, and have the ARPs for each interface be answered
1219 based on whether or not the kernel would route a packet from
1220 the ARP'd IP out that interface (therefore you must use source
1221 based routing for this to work). In other words it allows control
1222 of which cards (usually 1) will respond to an arp request.
1224 0 - (default) The kernel can respond to arp requests with addresses
1225 from other interfaces. This may seem wrong but it usually makes
1226 sense, because it increases the chance of successful communication.
1227 IP addresses are owned by the complete host on Linux, not by
1228 particular interfaces. Only for more complex setups like load-
1229 balancing, does this behaviour cause problems.
1231 arp_filter for the interface will be enabled if at least one of
1232 conf/{all,interface}/arp_filter is set to TRUE,
1233 it will be disabled otherwise
1235 arp_announce - INTEGER
1236 Define different restriction levels for announcing the local
1237 source IP address from IP packets in ARP requests sent on
1239 0 - (default) Use any local address, configured on any interface
1240 1 - Try to avoid local addresses that are not in the target's
1241 subnet for this interface. This mode is useful when target
1242 hosts reachable via this interface require the source IP
1243 address in ARP requests to be part of their logical network
1244 configured on the receiving interface. When we generate the
1245 request we will check all our subnets that include the
1246 target IP and will preserve the source address if it is from
1247 such subnet. If there is no such subnet we select source
1248 address according to the rules for level 2.
1249 2 - Always use the best local address for this target.
1250 In this mode we ignore the source address in the IP packet
1251 and try to select local address that we prefer for talks with
1252 the target host. Such local address is selected by looking
1253 for primary IP addresses on all our subnets on the outgoing
1254 interface that include the target IP address. If no suitable
1255 local address is found we select the first local address
1256 we have on the outgoing interface or on all other interfaces,
1257 with the hope we will receive reply for our request and
1258 even sometimes no matter the source IP address we announce.
1260 The max value from conf/{all,interface}/arp_announce is used.
1262 Increasing the restriction level gives more chance for
1263 receiving answer from the resolved target while decreasing
1264 the level announces more valid sender's information.
1266 arp_ignore - INTEGER
1267 Define different modes for sending replies in response to
1268 received ARP requests that resolve local target IP addresses:
1269 0 - (default): reply for any local target IP address, configured
1271 1 - reply only if the target IP address is local address
1272 configured on the incoming interface
1273 2 - reply only if the target IP address is local address
1274 configured on the incoming interface and both with the
1275 sender's IP address are part from same subnet on this interface
1276 3 - do not reply for local addresses configured with scope host,
1277 only resolutions for global and link addresses are replied
1279 8 - do not reply for all local addresses
1281 The max value from conf/{all,interface}/arp_ignore is used
1282 when ARP request is received on the {interface}
1284 arp_notify - BOOLEAN
1285 Define mode for notification of address and device changes.
1286 0 - (default): do nothing
1287 1 - Generate gratuitous arp requests when device is brought up
1288 or hardware address changes.
1290 arp_accept - BOOLEAN
1291 Define behavior for gratuitous ARP frames who's IP is not
1292 already present in the ARP table:
1293 0 - don't create new entries in the ARP table
1294 1 - create new entries in the ARP table
1296 Both replies and requests type gratuitous arp will trigger the
1297 ARP table to be updated, if this setting is on.
1299 If the ARP table already contains the IP address of the
1300 gratuitous arp frame, the arp table will be updated regardless
1301 if this setting is on or off.
1303 mcast_solicit - INTEGER
1304 The maximum number of multicast probes in INCOMPLETE state,
1305 when the associated hardware address is unknown. Defaults
1308 ucast_solicit - INTEGER
1309 The maximum number of unicast probes in PROBE state, when
1310 the hardware address is being reconfirmed. Defaults to 3.
1312 app_solicit - INTEGER
1313 The maximum number of probes to send to the user space ARP daemon
1314 via netlink before dropping back to multicast probes (see
1315 mcast_resolicit). Defaults to 0.
1317 mcast_resolicit - INTEGER
1318 The maximum number of multicast probes after unicast and
1319 app probes in PROBE state. Defaults to 0.
1321 disable_policy - BOOLEAN
1322 Disable IPSEC policy (SPD) for this interface
1324 disable_xfrm - BOOLEAN
1325 Disable IPSEC encryption on this interface, whatever the policy
1327 igmpv2_unsolicited_report_interval - INTEGER
1328 The interval in milliseconds in which the next unsolicited
1329 IGMPv1 or IGMPv2 report retransmit will take place.
1330 Default: 10000 (10 seconds)
1332 igmpv3_unsolicited_report_interval - INTEGER
1333 The interval in milliseconds in which the next unsolicited
1334 IGMPv3 report retransmit will take place.
1335 Default: 1000 (1 seconds)
1337 promote_secondaries - BOOLEAN
1338 When a primary IP address is removed from this interface
1339 promote a corresponding secondary IP address instead of
1340 removing all the corresponding secondary IP addresses.
1342 drop_unicast_in_l2_multicast - BOOLEAN
1343 Drop any unicast IP packets that are received in link-layer
1344 multicast (or broadcast) frames.
1345 This behavior (for multicast) is actually a SHOULD in RFC
1346 1122, but is disabled by default for compatibility reasons.
1349 drop_gratuitous_arp - BOOLEAN
1350 Drop all gratuitous ARP frames, for example if there's a known
1351 good ARP proxy on the network and such frames need not be used
1352 (or in the case of 802.11, must not be used to prevent attacks.)
1357 Allows you to write a number, which can be used as required.
1360 xfrm4_gc_thresh - INTEGER
1361 (Obsolete since linux-4.14)
1362 The threshold at which we will start garbage collecting for IPv4
1363 destination cache entries. At twice this value the system will
1364 refuse new allocations.
1366 igmp_link_local_mcast_reports - BOOLEAN
1367 Enable IGMP reports for link local multicast groups in the
1372 kuznet@ms2.inr.ac.ru
1378 delon.nicolas@wanadoo.fr
1383 /proc/sys/net/ipv6/* Variables:
1385 IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
1386 apply to IPv6 [XXX?].
1388 bindv6only - BOOLEAN
1389 Default value for IPV6_V6ONLY socket option,
1390 which restricts use of the IPv6 socket to IPv6 communication
1392 TRUE: disable IPv4-mapped address feature
1393 FALSE: enable IPv4-mapped address feature
1395 Default: FALSE (as specified in RFC3493)
1397 flowlabel_consistency - BOOLEAN
1398 Protect the consistency (and unicity) of flow label.
1399 You have to disable it to use IPV6_FL_F_REFLECT flag on the
1405 auto_flowlabels - INTEGER
1406 Automatically generate flow labels based on a flow hash of the
1407 packet. This allows intermediate devices, such as routers, to
1408 identify packet flows for mechanisms like Equal Cost Multipath
1409 Routing (see RFC 6438).
1410 0: automatic flow labels are completely disabled
1411 1: automatic flow labels are enabled by default, they can be
1412 disabled on a per socket basis using the IPV6_AUTOFLOWLABEL
1414 2: automatic flow labels are allowed, they may be enabled on a
1415 per socket basis using the IPV6_AUTOFLOWLABEL socket option
1416 3: automatic flow labels are enabled and enforced, they cannot
1417 be disabled by the socket option
1420 flowlabel_state_ranges - BOOLEAN
1421 Split the flow label number space into two ranges. 0-0x7FFFF is
1422 reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF
1423 is reserved for stateless flow labels as described in RFC6437.
1428 flowlabel_reflect - BOOLEAN
1429 Automatically reflect the flow label. Needed for Path MTU
1430 Discovery to work with Equal Cost Multipath Routing in anycast
1431 environments. See RFC 7690 and:
1432 https://tools.ietf.org/html/draft-wang-6man-flow-label-reflection-01
1437 fib_multipath_hash_policy - INTEGER
1438 Controls which hash policy to use for multipath routes.
1439 Default: 0 (Layer 3)
1441 0 - Layer 3 (source and destination addresses plus flow label)
1442 1 - Layer 4 (standard 5-tuple)
1444 anycast_src_echo_reply - BOOLEAN
1445 Controls the use of anycast addresses as source addresses for ICMPv6
1451 idgen_delay - INTEGER
1452 Controls the delay in seconds after which time to retry
1453 privacy stable address generation if a DAD conflict is
1455 Default: 1 (as specified in RFC7217)
1457 idgen_retries - INTEGER
1458 Controls the number of retries to generate a stable privacy
1459 address if a DAD conflict is detected.
1460 Default: 3 (as specified in RFC7217)
1463 Controls the MLD query robustness variable (see RFC3810 9.1).
1464 Default: 2 (as specified by RFC3810 9.1)
1465 Minimum: 1 (as specified by RFC6636 4.5)
1467 max_dst_opts_number - INTEGER
1468 Maximum number of non-padding TLVs allowed in a Destination
1469 options extension header. If this value is less than zero
1470 then unknown options are disallowed and the number of known
1471 TLVs allowed is the absolute value of this number.
1474 max_hbh_opts_number - INTEGER
1475 Maximum number of non-padding TLVs allowed in a Hop-by-Hop
1476 options extension header. If this value is less than zero
1477 then unknown options are disallowed and the number of known
1478 TLVs allowed is the absolute value of this number.
1481 max_dst_opts_length - INTEGER
1482 Maximum length allowed for a Destination options extension
1484 Default: INT_MAX (unlimited)
1486 max_hbh_length - INTEGER
1487 Maximum length allowed for a Hop-by-Hop options extension
1489 Default: INT_MAX (unlimited)
1491 skip_notify_on_dev_down - BOOLEAN
1492 Controls whether an RTM_DELROUTE message is generated for routes
1493 removed when a device is taken down or deleted. IPv4 does not
1494 generate this message; IPv6 does by default. Setting this sysctl
1495 to true skips the message, making IPv4 and IPv6 on par in relying
1496 on userspace caches to track link events and evict routes.
1497 Default: false (generate message)
1501 ip6frag_high_thresh - INTEGER
1502 Maximum memory used to reassemble IPv6 fragments. When
1503 ip6frag_high_thresh bytes of memory is allocated for this purpose,
1504 the fragment handler will toss packets until ip6frag_low_thresh
1507 ip6frag_low_thresh - INTEGER
1508 See ip6frag_high_thresh
1510 ip6frag_time - INTEGER
1511 Time in seconds to keep an IPv6 fragment in memory.
1513 IPv6 Segment Routing:
1515 seg6_flowlabel - INTEGER
1516 Controls the behaviour of computing the flowlabel of outer
1517 IPv6 header in case of SR T.encaps
1519 -1 set flowlabel to zero.
1520 0 copy flowlabel from Inner packet in case of Inner IPv6
1521 (Set flowlabel to 0 in case IPv4/L2)
1522 1 Compute the flowlabel using seg6_make_flowlabel()
1527 Change the interface-specific default settings.
1531 Change all the interface-specific settings.
1533 [XXX: Other special features than forwarding?]
1535 conf/all/forwarding - BOOLEAN
1536 Enable global IPv6 forwarding between all interfaces.
1538 IPv4 and IPv6 work differently here; e.g. netfilter must be used
1539 to control which interfaces may forward packets and which not.
1541 This also sets all interfaces' Host/Router setting
1542 'forwarding' to the specified value. See below for details.
1544 This referred to as global forwarding.
1549 fwmark_reflect - BOOLEAN
1550 Controls the fwmark of kernel-generated IPv6 reply packets that are not
1551 associated with a socket for example, TCP RSTs or ICMPv6 echo replies).
1552 If unset, these packets have a fwmark of zero. If set, they have the
1553 fwmark of the packet they are replying to.
1557 Change special settings per interface.
1559 The functional behaviour for certain settings is different
1560 depending on whether local forwarding is enabled or not.
1563 Accept Router Advertisements; autoconfigure using them.
1565 It also determines whether or not to transmit Router
1566 Solicitations. If and only if the functional setting is to
1567 accept Router Advertisements, Router Solicitations will be
1570 Possible values are:
1571 0 Do not accept Router Advertisements.
1572 1 Accept Router Advertisements if forwarding is disabled.
1573 2 Overrule forwarding behaviour. Accept Router Advertisements
1574 even if forwarding is enabled.
1576 Functional default: enabled if local forwarding is disabled.
1577 disabled if local forwarding is enabled.
1579 accept_ra_defrtr - BOOLEAN
1580 Learn default router in Router Advertisement.
1582 Functional default: enabled if accept_ra is enabled.
1583 disabled if accept_ra is disabled.
1585 accept_ra_from_local - BOOLEAN
1586 Accept RA with source-address that is found on local machine
1587 if the RA is otherwise proper and able to be accepted.
1588 Default is to NOT accept these as it may be an un-intended
1592 enabled if accept_ra_from_local is enabled
1593 on a specific interface.
1594 disabled if accept_ra_from_local is disabled
1595 on a specific interface.
1597 accept_ra_min_hop_limit - INTEGER
1598 Minimum hop limit Information in Router Advertisement.
1600 Hop limit Information in Router Advertisement less than this
1601 variable shall be ignored.
1605 accept_ra_pinfo - BOOLEAN
1606 Learn Prefix Information in Router Advertisement.
1608 Functional default: enabled if accept_ra is enabled.
1609 disabled if accept_ra is disabled.
1611 accept_ra_rt_info_min_plen - INTEGER
1612 Minimum prefix length of Route Information in RA.
1614 Route Information w/ prefix smaller than this variable shall
1617 Functional default: 0 if accept_ra_rtr_pref is enabled.
1618 -1 if accept_ra_rtr_pref is disabled.
1620 accept_ra_rt_info_max_plen - INTEGER
1621 Maximum prefix length of Route Information in RA.
1623 Route Information w/ prefix larger than this variable shall
1626 Functional default: 0 if accept_ra_rtr_pref is enabled.
1627 -1 if accept_ra_rtr_pref is disabled.
1629 accept_ra_rtr_pref - BOOLEAN
1630 Accept Router Preference in RA.
1632 Functional default: enabled if accept_ra is enabled.
1633 disabled if accept_ra is disabled.
1635 accept_ra_mtu - BOOLEAN
1636 Apply the MTU value specified in RA option 5 (RFC4861). If
1637 disabled, the MTU specified in the RA will be ignored.
1639 Functional default: enabled if accept_ra is enabled.
1640 disabled if accept_ra is disabled.
1642 accept_redirects - BOOLEAN
1645 Functional default: enabled if local forwarding is disabled.
1646 disabled if local forwarding is enabled.
1648 accept_source_route - INTEGER
1649 Accept source routing (routing extension header).
1651 >= 0: Accept only routing header type 2.
1652 < 0: Do not accept routing header.
1657 Autoconfigure addresses using Prefix Information in Router
1660 Functional default: enabled if accept_ra_pinfo is enabled.
1661 disabled if accept_ra_pinfo is disabled.
1663 dad_transmits - INTEGER
1664 The amount of Duplicate Address Detection probes to send.
1667 forwarding - INTEGER
1668 Configure interface-specific Host/Router behaviour.
1670 Note: It is recommended to have the same setting on all
1671 interfaces; mixed router/host scenarios are rather uncommon.
1673 Possible values are:
1674 0 Forwarding disabled
1675 1 Forwarding enabled
1679 By default, Host behaviour is assumed. This means:
1681 1. IsRouter flag is not set in Neighbour Advertisements.
1682 2. If accept_ra is TRUE (default), transmit Router
1684 3. If accept_ra is TRUE (default), accept Router
1685 Advertisements (and do autoconfiguration).
1686 4. If accept_redirects is TRUE (default), accept Redirects.
1690 If local forwarding is enabled, Router behaviour is assumed.
1691 This means exactly the reverse from the above:
1693 1. IsRouter flag is set in Neighbour Advertisements.
1694 2. Router Solicitations are not sent unless accept_ra is 2.
1695 3. Router Advertisements are ignored unless accept_ra is 2.
1696 4. Redirects are ignored.
1698 Default: 0 (disabled) if global forwarding is disabled (default),
1699 otherwise 1 (enabled).
1702 Default Hop Limit to set.
1706 Default Maximum Transfer Unit
1707 Default: 1280 (IPv6 required minimum)
1709 ip_nonlocal_bind - BOOLEAN
1710 If set, allows processes to bind() to non-local IPv6 addresses,
1711 which can be quite useful - but may break some applications.
1714 router_probe_interval - INTEGER
1715 Minimum interval (in seconds) between Router Probing described
1720 router_solicitation_delay - INTEGER
1721 Number of seconds to wait after interface is brought up
1722 before sending Router Solicitations.
1725 router_solicitation_interval - INTEGER
1726 Number of seconds to wait between Router Solicitations.
1729 router_solicitations - INTEGER
1730 Number of Router Solicitations to send until assuming no
1731 routers are present.
1734 use_oif_addrs_only - BOOLEAN
1735 When enabled, the candidate source addresses for destinations
1736 routed via this interface are restricted to the set of addresses
1737 configured on this interface (vis. RFC 6724, section 4).
1741 use_tempaddr - INTEGER
1742 Preference for Privacy Extensions (RFC3041).
1743 <= 0 : disable Privacy Extensions
1744 == 1 : enable Privacy Extensions, but prefer public
1745 addresses over temporary addresses.
1746 > 1 : enable Privacy Extensions and prefer temporary
1747 addresses over public addresses.
1748 Default: 0 (for most devices)
1749 -1 (for point-to-point devices and loopback devices)
1751 temp_valid_lft - INTEGER
1752 valid lifetime (in seconds) for temporary addresses.
1753 Default: 604800 (7 days)
1755 temp_prefered_lft - INTEGER
1756 Preferred lifetime (in seconds) for temporary addresses.
1757 Default: 86400 (1 day)
1759 keep_addr_on_down - INTEGER
1760 Keep all IPv6 addresses on an interface down event. If set static
1761 global addresses with no expiration time are not flushed.
1766 Default: 0 (addresses are removed)
1768 max_desync_factor - INTEGER
1769 Maximum value for DESYNC_FACTOR, which is a random value
1770 that ensures that clients don't synchronize with each
1771 other and generate new addresses at exactly the same time.
1772 value is in seconds.
1775 regen_max_retry - INTEGER
1776 Number of attempts before give up attempting to generate
1777 valid temporary addresses.
1780 max_addresses - INTEGER
1781 Maximum number of autoconfigured addresses per interface. Setting
1782 to zero disables the limitation. It is not recommended to set this
1783 value too large (or to zero) because it would be an easy way to
1784 crash the kernel by allowing too many addresses to be created.
1787 disable_ipv6 - BOOLEAN
1788 Disable IPv6 operation. If accept_dad is set to 2, this value
1789 will be dynamically set to TRUE if DAD fails for the link-local
1791 Default: FALSE (enable IPv6 operation)
1793 When this value is changed from 1 to 0 (IPv6 is being enabled),
1794 it will dynamically create a link-local address on the given
1795 interface and start Duplicate Address Detection, if necessary.
1797 When this value is changed from 0 to 1 (IPv6 is being disabled),
1798 it will dynamically delete all addresses and routes on the given
1799 interface. From now on it will not possible to add addresses/routes
1800 to the selected interface.
1802 accept_dad - INTEGER
1803 Whether to accept DAD (Duplicate Address Detection).
1805 1: Enable DAD (default)
1806 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1807 link-local address has been found.
1809 DAD operation and mode on a given interface will be selected according
1810 to the maximum value of conf/{all,interface}/accept_dad.
1812 force_tllao - BOOLEAN
1813 Enable sending the target link-layer address option even when
1814 responding to a unicast neighbor solicitation.
1817 Quoting from RFC 2461, section 4.4, Target link-layer address:
1819 "The option MUST be included for multicast solicitations in order to
1820 avoid infinite Neighbor Solicitation "recursion" when the peer node
1821 does not have a cache entry to return a Neighbor Advertisements
1822 message. When responding to unicast solicitations, the option can be
1823 omitted since the sender of the solicitation has the correct link-
1824 layer address; otherwise it would not have be able to send the unicast
1825 solicitation in the first place. However, including the link-layer
1826 address in this case adds little overhead and eliminates a potential
1827 race condition where the sender deletes the cached link-layer address
1828 prior to receiving a response to a previous solicitation."
1830 ndisc_notify - BOOLEAN
1831 Define mode for notification of address and device changes.
1832 0 - (default): do nothing
1833 1 - Generate unsolicited neighbour advertisements when device is brought
1834 up or hardware address changes.
1836 ndisc_tclass - INTEGER
1837 The IPv6 Traffic Class to use by default when sending IPv6 Neighbor
1838 Discovery (Router Solicitation, Router Advertisement, Neighbor
1839 Solicitation, Neighbor Advertisement, Redirect) messages.
1840 These 8 bits can be interpreted as 6 high order bits holding the DSCP
1841 value and 2 low order bits representing ECN (which you probably want
1845 mldv1_unsolicited_report_interval - INTEGER
1846 The interval in milliseconds in which the next unsolicited
1847 MLDv1 report retransmit will take place.
1848 Default: 10000 (10 seconds)
1850 mldv2_unsolicited_report_interval - INTEGER
1851 The interval in milliseconds in which the next unsolicited
1852 MLDv2 report retransmit will take place.
1853 Default: 1000 (1 second)
1855 force_mld_version - INTEGER
1856 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed
1857 1 - Enforce to use MLD version 1
1858 2 - Enforce to use MLD version 2
1860 suppress_frag_ndisc - INTEGER
1861 Control RFC 6980 (Security Implications of IPv6 Fragmentation
1862 with IPv6 Neighbor Discovery) behavior:
1863 1 - (default) discard fragmented neighbor discovery packets
1864 0 - allow fragmented neighbor discovery packets
1866 optimistic_dad - BOOLEAN
1867 Whether to perform Optimistic Duplicate Address Detection (RFC 4429).
1868 0: disabled (default)
1871 Optimistic Duplicate Address Detection for the interface will be enabled
1872 if at least one of conf/{all,interface}/optimistic_dad is set to 1,
1873 it will be disabled otherwise.
1875 use_optimistic - BOOLEAN
1876 If enabled, do not classify optimistic addresses as deprecated during
1877 source address selection. Preferred addresses will still be chosen
1878 before optimistic addresses, subject to other ranking in the source
1879 address selection algorithm.
1880 0: disabled (default)
1883 This will be enabled if at least one of
1884 conf/{all,interface}/use_optimistic is set to 1, disabled otherwise.
1886 stable_secret - IPv6 address
1887 This IPv6 address will be used as a secret to generate IPv6
1888 addresses for link-local addresses and autoconfigured
1889 ones. All addresses generated after setting this secret will
1890 be stable privacy ones by default. This can be changed via the
1891 addrgenmode ip-link. conf/default/stable_secret is used as the
1892 secret for the namespace, the interface specific ones can
1893 overwrite that. Writes to conf/all/stable_secret are refused.
1895 It is recommended to generate this secret during installation
1896 of a system and keep it stable after that.
1898 By default the stable secret is unset.
1900 addr_gen_mode - INTEGER
1901 Defines how link-local and autoconf addresses are generated.
1903 0: generate address based on EUI64 (default)
1904 1: do no generate a link-local address, use EUI64 for addresses generated
1906 2: generate stable privacy addresses, using the secret from
1907 stable_secret (RFC7217)
1908 3: generate stable privacy addresses, using a random secret if unset
1910 drop_unicast_in_l2_multicast - BOOLEAN
1911 Drop any unicast IPv6 packets that are received in link-layer
1912 multicast (or broadcast) frames.
1914 By default this is turned off.
1916 drop_unsolicited_na - BOOLEAN
1917 Drop all unsolicited neighbor advertisements, for example if there's
1918 a known good NA proxy on the network and such frames need not be used
1919 (or in the case of 802.11, must not be used to prevent attacks.)
1921 By default this is turned off.
1923 enhanced_dad - BOOLEAN
1924 Include a nonce option in the IPv6 neighbor solicitation messages used for
1925 duplicate address detection per RFC7527. A received DAD NS will only signal
1926 a duplicate address if the nonce is different. This avoids any false
1927 detection of duplicates due to loopback of the NS messages that we send.
1928 The nonce option will be sent on an interface unless both of
1929 conf/{all,interface}/enhanced_dad are set to FALSE.
1934 Limit the maximal rates for sending ICMPv6 messages.
1935 0 to disable any limiting,
1936 otherwise the minimal space between responses in milliseconds.
1939 ratemask - list of comma separated ranges
1940 For ICMPv6 message types matching the ranges in the ratemask, limit
1941 the sending of the message according to ratelimit parameter.
1943 The format used for both input and output is a comma separated
1944 list of ranges (e.g. "0-127,129" for ICMPv6 message type 0 to 127 and
1945 129). Writing to the file will clear all previous ranges of ICMPv6
1946 message types and update the current list with the input.
1948 Refer to: https://www.iana.org/assignments/icmpv6-parameters/icmpv6-parameters.xhtml
1949 for numerical values of ICMPv6 message types, e.g. echo request is 128
1950 and echo reply is 129.
1952 Default: 0-1,3-127 (rate limit ICMPv6 errors except Packet Too Big)
1954 echo_ignore_all - BOOLEAN
1955 If set non-zero, then the kernel will ignore all ICMP ECHO
1956 requests sent to it over the IPv6 protocol.
1959 echo_ignore_multicast - BOOLEAN
1960 If set non-zero, then the kernel will ignore all ICMP ECHO
1961 requests sent to it over the IPv6 protocol via multicast.
1964 echo_ignore_anycast - BOOLEAN
1965 If set non-zero, then the kernel will ignore all ICMP ECHO
1966 requests sent to it over the IPv6 protocol destined to anycast address.
1969 xfrm6_gc_thresh - INTEGER
1970 (Obsolete since linux-4.14)
1971 The threshold at which we will start garbage collecting for IPv6
1972 destination cache entries. At twice this value the system will
1973 refuse new allocations.
1977 Pekka Savola <pekkas@netcore.fi>
1978 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1981 /proc/sys/net/bridge/* Variables:
1983 bridge-nf-call-arptables - BOOLEAN
1984 1 : pass bridged ARP traffic to arptables' FORWARD chain.
1988 bridge-nf-call-iptables - BOOLEAN
1989 1 : pass bridged IPv4 traffic to iptables' chains.
1993 bridge-nf-call-ip6tables - BOOLEAN
1994 1 : pass bridged IPv6 traffic to ip6tables' chains.
1998 bridge-nf-filter-vlan-tagged - BOOLEAN
1999 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
2003 bridge-nf-filter-pppoe-tagged - BOOLEAN
2004 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
2008 bridge-nf-pass-vlan-input-dev - BOOLEAN
2009 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
2010 interface on the bridge and set the netfilter input device to the vlan.
2011 This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
2012 target work with vlan-on-top-of-bridge interfaces. When no matching
2013 vlan interface is found, or this switch is off, the input device is
2014 set to the bridge interface.
2015 0: disable bridge netfilter vlan interface lookup.
2018 proc/sys/net/sctp/* Variables:
2020 addip_enable - BOOLEAN
2021 Enable or disable extension of Dynamic Address Reconfiguration
2022 (ADD-IP) functionality specified in RFC5061. This extension provides
2023 the ability to dynamically add and remove new addresses for the SCTP
2026 1: Enable extension.
2028 0: Disable extension.
2033 Enable or disable pf (pf is short for potentially failed) state. A value
2034 of pf_retrans > path_max_retrans also disables pf state. That is, one of
2035 both pf_enable and pf_retrans > path_max_retrans can disable pf state.
2036 Since pf_retrans and path_max_retrans can be changed by userspace
2037 application, sometimes user expects to disable pf state by the value of
2038 pf_retrans > path_max_retrans, but occasionally the value of pf_retrans
2039 or path_max_retrans is changed by the user application, this pf state is
2040 enabled. As such, it is necessary to add this to dynamically enable
2041 and disable pf state. See:
2042 https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for
2051 addip_noauth_enable - BOOLEAN
2052 Dynamic Address Reconfiguration (ADD-IP) requires the use of
2053 authentication to protect the operations of adding or removing new
2054 addresses. This requirement is mandated so that unauthorized hosts
2055 would not be able to hijack associations. However, older
2056 implementations may not have implemented this requirement while
2057 allowing the ADD-IP extension. For reasons of interoperability,
2058 we provide this variable to control the enforcement of the
2059 authentication requirement.
2061 1: Allow ADD-IP extension to be used without authentication. This
2062 should only be set in a closed environment for interoperability
2063 with older implementations.
2065 0: Enforce the authentication requirement
2069 auth_enable - BOOLEAN
2070 Enable or disable Authenticated Chunks extension. This extension
2071 provides the ability to send and receive authenticated chunks and is
2072 required for secure operation of Dynamic Address Reconfiguration
2075 1: Enable this extension.
2076 0: Disable this extension.
2080 prsctp_enable - BOOLEAN
2081 Enable or disable the Partial Reliability extension (RFC3758) which
2082 is used to notify peers that a given DATA should no longer be expected.
2090 The limit of the number of new packets that can be initially sent. It
2091 controls how bursty the generated traffic can be.
2095 association_max_retrans - INTEGER
2096 Set the maximum number for retransmissions that an association can
2097 attempt deciding that the remote end is unreachable. If this value
2098 is exceeded, the association is terminated.
2102 max_init_retransmits - INTEGER
2103 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
2104 that an association will attempt before declaring the destination
2105 unreachable and terminating.
2109 path_max_retrans - INTEGER
2110 The maximum number of retransmissions that will be attempted on a given
2111 path. Once this threshold is exceeded, the path is considered
2112 unreachable, and new traffic will use a different path when the
2113 association is multihomed.
2117 pf_retrans - INTEGER
2118 The number of retransmissions that will be attempted on a given path
2119 before traffic is redirected to an alternate transport (should one
2120 exist). Note this is distinct from path_max_retrans, as a path that
2121 passes the pf_retrans threshold can still be used. Its only
2122 deprioritized when a transmission path is selected by the stack. This
2123 setting is primarily used to enable fast failover mechanisms without
2124 having to reduce path_max_retrans to a very low value. See:
2125 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
2126 for details. Note also that a value of pf_retrans > path_max_retrans
2127 disables this feature. Since both pf_retrans and path_max_retrans can
2128 be changed by userspace application, a variable pf_enable is used to
2133 rto_initial - INTEGER
2134 The initial round trip timeout value in milliseconds that will be used
2135 in calculating round trip times. This is the initial time interval
2136 for retransmissions.
2141 The maximum value (in milliseconds) of the round trip timeout. This
2142 is the largest time interval that can elapse between retransmissions.
2147 The minimum value (in milliseconds) of the round trip timeout. This
2148 is the smallest time interval the can elapse between retransmissions.
2152 hb_interval - INTEGER
2153 The interval (in milliseconds) between HEARTBEAT chunks. These chunks
2154 are sent at the specified interval on idle paths to probe the state of
2155 a given path between 2 associations.
2159 sack_timeout - INTEGER
2160 The amount of time (in milliseconds) that the implementation will wait
2165 valid_cookie_life - INTEGER
2166 The default lifetime of the SCTP cookie (in milliseconds). The cookie
2167 is used during association establishment.
2171 cookie_preserve_enable - BOOLEAN
2172 Enable or disable the ability to extend the lifetime of the SCTP cookie
2173 that is used during the establishment phase of SCTP association
2175 1: Enable cookie lifetime extension.
2180 cookie_hmac_alg - STRING
2181 Select the hmac algorithm used when generating the cookie value sent by
2182 a listening sctp socket to a connecting client in the INIT-ACK chunk.
2187 Ability to assign md5 or sha1 as the selected alg is predicated on the
2188 configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
2189 CONFIG_CRYPTO_SHA1).
2191 Default: Dependent on configuration. MD5 if available, else SHA1 if
2192 available, else none.
2194 rcvbuf_policy - INTEGER
2195 Determines if the receive buffer is attributed to the socket or to
2196 association. SCTP supports the capability to create multiple
2197 associations on a single socket. When using this capability, it is
2198 possible that a single stalled association that's buffering a lot
2199 of data may block other associations from delivering their data by
2200 consuming all of the receive buffer space. To work around this,
2201 the rcvbuf_policy could be set to attribute the receiver buffer space
2202 to each association instead of the socket. This prevents the described
2205 1: rcvbuf space is per association
2206 0: rcvbuf space is per socket
2210 sndbuf_policy - INTEGER
2211 Similar to rcvbuf_policy above, this applies to send buffer space.
2213 1: Send buffer is tracked per association
2214 0: Send buffer is tracked per socket.
2218 sctp_mem - vector of 3 INTEGERs: min, pressure, max
2219 Number of pages allowed for queueing by all SCTP sockets.
2221 min: Below this number of pages SCTP is not bothered about its
2222 memory appetite. When amount of memory allocated by SCTP exceeds
2223 this number, SCTP starts to moderate memory usage.
2225 pressure: This value was introduced to follow format of tcp_mem.
2227 max: Number of pages allowed for queueing by all SCTP sockets.
2229 Default is calculated at boot time from amount of available memory.
2231 sctp_rmem - vector of 3 INTEGERs: min, default, max
2232 Only the first value ("min") is used, "default" and "max" are
2235 min: Minimal size of receive buffer used by SCTP socket.
2236 It is guaranteed to each SCTP socket (but not association) even
2237 under moderate memory pressure.
2241 sctp_wmem - vector of 3 INTEGERs: min, default, max
2242 Currently this tunable has no effect.
2244 addr_scope_policy - INTEGER
2245 Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
2247 0 - Disable IPv4 address scoping
2248 1 - Enable IPv4 address scoping
2249 2 - Follow draft but allow IPv4 private addresses
2250 3 - Follow draft but allow IPv4 link local addresses
2255 /proc/sys/net/core/*
2256 Please see: Documentation/sysctl/net.txt for descriptions of these entries.
2259 /proc/sys/net/unix/*
2260 max_dgram_qlen - INTEGER
2261 The maximum length of dgram socket receive queue