brintos / linux-shallow public Read only

1.. SPDX-License-Identifier: GPL-2.02 3=========4IP Sysctl5=========6 7/proc/sys/net/ipv4/* Variables8==============================9 10ip_forward - BOOLEAN11	- 0 - disabled (default)12	- not 0 - enabled13 14	Forward Packets between interfaces.15 16	This variable is special, its change resets all configuration17	parameters to their default state (RFC1122 for hosts, RFC181218	for routers)19 20ip_default_ttl - INTEGER21	Default value of TTL field (Time To Live) for outgoing (but not22	forwarded) IP packets. Should be between 1 and 255 inclusive.23	Default: 64 (as recommended by RFC1700)24 25ip_no_pmtu_disc - INTEGER26	Disable Path MTU Discovery. If enabled in mode 1 and a27	fragmentation-required ICMP is received, the PMTU to this28	destination will be set to the smallest of the old MTU to29	this destination and min_pmtu (see below). You will need30	to raise min_pmtu to the smallest interface MTU on your system31	manually if you want to avoid locally generated fragments.32 33	In mode 2 incoming Path MTU Discovery messages will be34	discarded. Outgoing frames are handled the same as in mode 1,35	implicitly setting IP_PMTUDISC_DONT on every created socket.36 37	Mode 3 is a hardened pmtu discover mode. The kernel will only38	accept fragmentation-needed errors if the underlying protocol39	can verify them besides a plain socket lookup. Current40	protocols for which pmtu events will be honored are TCP, SCTP41	and DCCP as they verify e.g. the sequence number or the42	association. This mode should not be enabled globally but is43	only intended to secure e.g. name servers in namespaces where44	TCP path mtu must still work but path MTU information of other45	protocols should be discarded. If enabled globally this mode46	could break other protocols.47 48	Possible values: 0-349 50	Default: FALSE51 52min_pmtu - INTEGER53	default 552 - minimum Path MTU. Unless this is changed manually,54	each cached pmtu will never be lower than this setting.55 56ip_forward_use_pmtu - BOOLEAN57	By default we don't trust protocol path MTUs while forwarding58	because they could be easily forged and can lead to unwanted59	fragmentation by the router.60	You only need to enable this if you have user-space software61	which tries to discover path mtus by itself and depends on the62	kernel honoring this information. This is normally not the63	case.64 65	Default: 0 (disabled)66 67	Possible values:68 69	- 0 - disabled70	- 1 - enabled71 72fwmark_reflect - BOOLEAN73	Controls the fwmark of kernel-generated IPv4 reply packets that are not74	associated with a socket for example, TCP RSTs or ICMP echo replies).75	If unset, these packets have a fwmark of zero. If set, they have the76	fwmark of the packet they are replying to.77 78	Default: 079 80fib_multipath_use_neigh - BOOLEAN81	Use status of existing neighbor entry when determining nexthop for82	multipath routes. If disabled, neighbor information is not used and83	packets could be directed to a failed nexthop. Only valid for kernels84	built with CONFIG_IP_ROUTE_MULTIPATH enabled.85 86	Default: 0 (disabled)87 88	Possible values:89 90	- 0 - disabled91	- 1 - enabled92 93fib_multipath_hash_policy - INTEGER94	Controls which hash policy to use for multipath routes. Only valid95	for kernels built with CONFIG_IP_ROUTE_MULTIPATH enabled.96 97	Default: 0 (Layer 3)98 99	Possible values:100 101	- 0 - Layer 3102	- 1 - Layer 4103	- 2 - Layer 3 or inner Layer 3 if present104	- 3 - Custom multipath hash. Fields used for multipath hash calculation105	  are determined by fib_multipath_hash_fields sysctl106 107fib_multipath_hash_fields - UNSIGNED INTEGER108	When fib_multipath_hash_policy is set to 3 (custom multipath hash), the109	fields used for multipath hash calculation are determined by this110	sysctl.111 112	This value is a bitmask which enables various fields for multipath hash113	calculation.114 115	Possible fields are:116 117	====== ============================118	0x0001 Source IP address119	0x0002 Destination IP address120	0x0004 IP protocol121	0x0008 Unused (Flow Label)122	0x0010 Source port123	0x0020 Destination port124	0x0040 Inner source IP address125	0x0080 Inner destination IP address126	0x0100 Inner IP protocol127	0x0200 Inner Flow Label128	0x0400 Inner source port129	0x0800 Inner destination port130	====== ============================131 132	Default: 0x0007 (source IP, destination IP and IP protocol)133 134fib_multipath_hash_seed - UNSIGNED INTEGER135	The seed value used when calculating hash for multipath routes. Applies136	to both IPv4 and IPv6 datapath. Only present for kernels built with137	CONFIG_IP_ROUTE_MULTIPATH enabled.138 139	When set to 0, the seed value used for multipath routing defaults to an140	internal random-generated one.141 142	The actual hashing algorithm is not specified -- there is no guarantee143	that a next hop distribution effected by a given seed will keep stable144	across kernel versions.145 146	Default: 0 (random)147 148fib_sync_mem - UNSIGNED INTEGER149	Amount of dirty memory from fib entries that can be backlogged before150	synchronize_rcu is forced.151 152	Default: 512kB   Minimum: 64kB   Maximum: 64MB153 154ip_forward_update_priority - INTEGER155	Whether to update SKB priority from "TOS" field in IPv4 header after it156	is forwarded. The new SKB priority is mapped from TOS field value157	according to an rt_tos2priority table (see e.g. man tc-prio).158 159	Default: 1 (Update priority.)160 161	Possible values:162 163	- 0 - Do not update priority.164	- 1 - Update priority.165 166route/max_size - INTEGER167	Maximum number of routes allowed in the kernel.  Increase168	this when using large numbers of interfaces and/or routes.169 170	From linux kernel 3.6 onwards, this is deprecated for ipv4171	as route cache is no longer used.172 173	From linux kernel 6.3 onwards, this is deprecated for ipv6174	as garbage collection manages cached route entries.175 176neigh/default/gc_thresh1 - INTEGER177	Minimum number of entries to keep.  Garbage collector will not178	purge entries if there are fewer than this number.179 180	Default: 128181 182neigh/default/gc_thresh2 - INTEGER183	Threshold when garbage collector becomes more aggressive about184	purging entries. Entries older than 5 seconds will be cleared185	when over this number.186 187	Default: 512188 189neigh/default/gc_thresh3 - INTEGER190	Maximum number of non-PERMANENT neighbor entries allowed.  Increase191	this when using large numbers of interfaces and when communicating192	with large numbers of directly-connected peers.193 194	Default: 1024195 196neigh/default/unres_qlen_bytes - INTEGER197	The maximum number of bytes which may be used by packets198	queued for each	unresolved address by other network layers.199	(added in linux 3.3)200 201	Setting negative value is meaningless and will return error.202 203	Default: SK_WMEM_MAX, (same as net.core.wmem_default).204 205		Exact value depends on architecture and kernel options,206		but should be enough to allow queuing 256 packets207		of medium size.208 209neigh/default/unres_qlen - INTEGER210	The maximum number of packets which may be queued for each211	unresolved address by other network layers.212 213	(deprecated in linux 3.3) : use unres_qlen_bytes instead.214 215	Prior to linux 3.3, the default value is 3 which may cause216	unexpected packet loss. The current default value is calculated217	according to default value of unres_qlen_bytes and true size of218	packet.219 220	Default: 101221 222neigh/default/interval_probe_time_ms - INTEGER223	The probe interval for neighbor entries with NTF_MANAGED flag,224	the min value is 1.225 226	Default: 5000227 228mtu_expires - INTEGER229	Time, in seconds, that cached PMTU information is kept.230 231min_adv_mss - INTEGER232	The advertised MSS depends on the first hop route MTU, but will233	never be lower than this setting.234 235fib_notify_on_flag_change - INTEGER236        Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/237        RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed.238 239        After installing a route to the kernel, user space receives an240        acknowledgment, which means the route was installed in the kernel,241        but not necessarily in hardware.242        It is also possible for a route already installed in hardware to change243        its action and therefore its flags. For example, a host route that is244        trapping packets can be "promoted" to perform decapsulation following245        the installation of an IPinIP/VXLAN tunnel.246        The notifications will indicate to user-space the state of the route.247 248        Default: 0 (Do not emit notifications.)249 250        Possible values:251 252        - 0 - Do not emit notifications.253        - 1 - Emit notifications.254        - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change.255 256IP Fragmentation:257 258ipfrag_high_thresh - LONG INTEGER259	Maximum memory used to reassemble IP fragments.260 261ipfrag_low_thresh - LONG INTEGER262	(Obsolete since linux-4.17)263	Maximum memory used to reassemble IP fragments before the kernel264	begins to remove incomplete fragment queues to free up resources.265	The kernel still accepts new fragments for defragmentation.266 267ipfrag_time - INTEGER268	Time in seconds to keep an IP fragment in memory.269 270ipfrag_max_dist - INTEGER271	ipfrag_max_dist is a non-negative integer value which defines the272	maximum "disorder" which is allowed among fragments which share a273	common IP source address. Note that reordering of packets is274	not unusual, but if a large number of fragments arrive from a source275	IP address while a particular fragment queue remains incomplete, it276	probably indicates that one or more fragments belonging to that queue277	have been lost. When ipfrag_max_dist is positive, an additional check278	is done on fragments before they are added to a reassembly queue - if279	ipfrag_max_dist (or more) fragments have arrived from a particular IP280	address between additions to any IP fragment queue using that source281	address, it's presumed that one or more fragments in the queue are282	lost. The existing fragment queue will be dropped, and a new one283	started. An ipfrag_max_dist value of zero disables this check.284 285	Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can286	result in unnecessarily dropping fragment queues when normal287	reordering of packets occurs, which could lead to poor application288	performance. Using a very large value, e.g. 50000, increases the289	likelihood of incorrectly reassembling IP fragments that originate290	from different IP datagrams, which could result in data corruption.291	Default: 64292 293bc_forwarding - INTEGER294	bc_forwarding enables the feature described in rfc1812#section-5.3.5.2295	and rfc2644. It allows the router to forward directed broadcast.296	To enable this feature, the 'all' entry and the input interface entry297	should be set to 1.298	Default: 0299 300INET peer storage301=================302 303inet_peer_threshold - INTEGER304	The approximate size of the storage.  Starting from this threshold305	entries will be thrown aggressively.  This threshold also determines306	entries' time-to-live and time intervals between garbage collection307	passes.  More entries, less time-to-live, less GC interval.308 309inet_peer_minttl - INTEGER310	Minimum time-to-live of entries.  Should be enough to cover fragment311	time-to-live on the reassembling side.  This minimum time-to-live  is312	guaranteed if the pool size is less than inet_peer_threshold.313	Measured in seconds.314 315inet_peer_maxttl - INTEGER316	Maximum time-to-live of entries.  Unused entries will expire after317	this period of time if there is no memory pressure on the pool (i.e.318	when the number of entries in the pool is very small).319	Measured in seconds.320 321TCP variables322=============323 324somaxconn - INTEGER325	Limit of socket listen() backlog, known in userspace as SOMAXCONN.326	Defaults to 4096. (Was 128 before linux-5.4)327	See also tcp_max_syn_backlog for additional tuning for TCP sockets.328 329tcp_abort_on_overflow - BOOLEAN330	If listening service is too slow to accept new connections,331	reset them. Default state is FALSE. It means that if overflow332	occurred due to a burst, connection will recover. Enable this333	option _only_ if you are really sure that listening daemon334	cannot be tuned to accept connections faster. Enabling this335	option can harm clients of your server.336 337tcp_adv_win_scale - INTEGER338	Obsolete since linux-6.6339	Count buffering overhead as bytes/2^tcp_adv_win_scale340	(if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),341	if it is <= 0.342 343	Possible values are [-31, 31], inclusive.344 345	Default: 1346 347tcp_allowed_congestion_control - STRING348	Show/set the congestion control choices available to non-privileged349	processes. The list is a subset of those listed in350	tcp_available_congestion_control.351 352	Default is "reno" and the default setting (tcp_congestion_control).353 354tcp_app_win - INTEGER355	Reserve max(window/2^tcp_app_win, mss) of window for application356	buffer. Value 0 is special, it means that nothing is reserved.357 358	Possible values are [0, 31], inclusive.359 360	Default: 31361 362tcp_autocorking - BOOLEAN363	Enable TCP auto corking :364	When applications do consecutive small write()/sendmsg() system calls,365	we try to coalesce these small writes as much as possible, to lower366	total amount of sent packets. This is done if at least one prior367	packet for the flow is waiting in Qdisc queues or device transmit368	queue. Applications can still use TCP_CORK for optimal behavior369	when they know how/when to uncork their sockets.370 371	Default : 1372 373tcp_available_congestion_control - STRING374	Shows the available congestion control choices that are registered.375	More congestion control algorithms may be available as modules,376	but not loaded.377 378tcp_base_mss - INTEGER379	The initial value of search_low to be used by the packetization layer380	Path MTU discovery (MTU probing).  If MTU probing is enabled,381	this is the initial MSS used by the connection.382 383tcp_mtu_probe_floor - INTEGER384	If MTU probing is enabled this caps the minimum MSS used for search_low385	for the connection.386 387	Default : 48388 389tcp_min_snd_mss - INTEGER390	TCP SYN and SYNACK messages usually advertise an ADVMSS option,391	as described in RFC 1122 and RFC 6691.392 393	If this ADVMSS option is smaller than tcp_min_snd_mss,394	it is silently capped to tcp_min_snd_mss.395 396	Default : 48 (at least 8 bytes of payload per segment)397 398tcp_congestion_control - STRING399	Set the congestion control algorithm to be used for new400	connections. The algorithm "reno" is always available, but401	additional choices may be available based on kernel configuration.402	Default is set as part of kernel configuration.403	For passive connections, the listener congestion control choice404	is inherited.405 406	[see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]407 408tcp_dsack - BOOLEAN409	Allows TCP to send "duplicate" SACKs.410 411tcp_early_retrans - INTEGER412	Tail loss probe (TLP) converts RTOs occurring due to tail413	losses into fast recovery (draft-ietf-tcpm-rack). Note that414	TLP requires RACK to function properly (see tcp_recovery below)415 416	Possible values:417 418		- 0 disables TLP419		- 3 or 4 enables TLP420 421	Default: 3422 423tcp_ecn - INTEGER424	Control use of Explicit Congestion Notification (ECN) by TCP.425	ECN is used only when both ends of the TCP connection indicate426	support for it.  This feature is useful in avoiding losses due427	to congestion by allowing supporting routers to signal428	congestion before having to drop packets.429 430	Possible values are:431 432		=  =====================================================433		0  Disable ECN.  Neither initiate nor accept ECN.434		1  Enable ECN when requested by incoming connections and435		   also request ECN on outgoing connection attempts.436		2  Enable ECN when requested by incoming connections437		   but do not request ECN on outgoing connections.438		=  =====================================================439 440	Default: 2441 442tcp_ecn_fallback - BOOLEAN443	If the kernel detects that ECN connection misbehaves, enable fall444	back to non-ECN. Currently, this knob implements the fallback445	from RFC3168, section 6.1.1.1., but we reserve that in future,446	additional detection mechanisms could be implemented under this447	knob. The value	is not used, if tcp_ecn or per route (or congestion448	control) ECN settings are disabled.449 450	Default: 1 (fallback enabled)451 452tcp_fack - BOOLEAN453	This is a legacy option, it has no effect anymore.454 455tcp_fin_timeout - INTEGER456	The length of time an orphaned (no longer referenced by any457	application) connection will remain in the FIN_WAIT_2 state458	before it is aborted at the local end.  While a perfectly459	valid "receive only" state for an un-orphaned connection, an460	orphaned connection in FIN_WAIT_2 state could otherwise wait461	forever for the remote to close its end of the connection.462 463	Cf. tcp_max_orphans464 465	Default: 60 seconds466 467tcp_frto - INTEGER468	Enables Forward RTO-Recovery (F-RTO) defined in RFC5682.469	F-RTO is an enhanced recovery algorithm for TCP retransmission470	timeouts.  It is particularly beneficial in networks where the471	RTT fluctuates (e.g., wireless). F-RTO is sender-side only472	modification. It does not require any support from the peer.473 474	By default it's enabled with a non-zero value. 0 disables F-RTO.475 476tcp_fwmark_accept - BOOLEAN477	If set, incoming connections to listening sockets that do not have a478	socket mark will set the mark of the accepting socket to the fwmark of479	the incoming SYN packet. This will cause all packets on that connection480	(starting from the first SYNACK) to be sent with that fwmark. The481	listening socket's mark is unchanged. Listening sockets that already482	have a fwmark set via setsockopt(SOL_SOCKET, SO_MARK, ...) are483	unaffected.484 485	Default: 0486 487tcp_invalid_ratelimit - INTEGER488	Limit the maximal rate for sending duplicate acknowledgments489	in response to incoming TCP packets that are for an existing490	connection but that are invalid due to any of these reasons:491 492	  (a) out-of-window sequence number,493	  (b) out-of-window acknowledgment number, or494	  (c) PAWS (Protection Against Wrapped Sequence numbers) check failure495 496	This can help mitigate simple "ack loop" DoS attacks, wherein497	a buggy or malicious middlebox or man-in-the-middle can498	rewrite TCP header fields in manner that causes each endpoint499	to think that the other is sending invalid TCP segments, thus500	causing each side to send an unterminating stream of duplicate501	acknowledgments for invalid segments.502 503	Using 0 disables rate-limiting of dupacks in response to504	invalid segments; otherwise this value specifies the minimal505	space between sending such dupacks, in milliseconds.506 507	Default: 500 (milliseconds).508 509tcp_keepalive_time - INTEGER510	How often TCP sends out keepalive messages when keepalive is enabled.511	Default: 2hours.512 513tcp_keepalive_probes - INTEGER514	How many keepalive probes TCP sends out, until it decides that the515	connection is broken. Default value: 9.516 517tcp_keepalive_intvl - INTEGER518	How frequently the probes are send out. Multiplied by519	tcp_keepalive_probes it is time to kill not responding connection,520	after probes started. Default value: 75sec i.e. connection521	will be aborted after ~11 minutes of retries.522 523tcp_l3mdev_accept - BOOLEAN524	Enables child sockets to inherit the L3 master device index.525	Enabling this option allows a "global" listen socket to work526	across L3 master domains (e.g., VRFs) with connected sockets527	derived from the listen socket to be bound to the L3 domain in528	which the packets originated. Only valid when the kernel was529	compiled with CONFIG_NET_L3_MASTER_DEV.530 531	Default: 0 (disabled)532 533tcp_low_latency - BOOLEAN534	This is a legacy option, it has no effect anymore.535 536tcp_max_orphans - INTEGER537	Maximal number of TCP sockets not attached to any user file handle,538	held by system.	If this number is exceeded orphaned connections are539	reset immediately and warning is printed. This limit exists540	only to prevent simple DoS attacks, you _must_ not rely on this541	or lower the limit artificially, but rather increase it542	(probably, after increasing installed memory),543	if network conditions require more than default value,544	and tune network services to linger and kill such states545	more aggressively. Let me to remind again: each orphan eats546	up to ~64K of unswappable memory.547 548tcp_max_syn_backlog - INTEGER549	Maximal number of remembered connection requests (SYN_RECV),550	which have not received an acknowledgment from connecting client.551 552	This is a per-listener limit.553 554	The minimal value is 128 for low memory machines, and it will555	increase in proportion to the memory of machine.556 557	If server suffers from overload, try increasing this number.558 559	Remember to also check /proc/sys/net/core/somaxconn560	A SYN_RECV request socket consumes about 304 bytes of memory.561 562tcp_max_tw_buckets - INTEGER563	Maximal number of timewait sockets held by system simultaneously.564	If this number is exceeded time-wait socket is immediately destroyed565	and warning is printed. This limit exists only to prevent566	simple DoS attacks, you _must_ not lower the limit artificially,567	but rather increase it (probably, after increasing installed memory),568	if network conditions require more than default value.569 570tcp_mem - vector of 3 INTEGERs: min, pressure, max571	min: below this number of pages TCP is not bothered about its572	memory appetite.573 574	pressure: when amount of memory allocated by TCP exceeds this number575	of pages, TCP moderates its memory consumption and enters memory576	pressure mode, which is exited when memory consumption falls577	under "min".578 579	max: number of pages allowed for queueing by all TCP sockets.580 581	Defaults are calculated at boot time from amount of available582	memory.583 584tcp_min_rtt_wlen - INTEGER585	The window length of the windowed min filter to track the minimum RTT.586	A shorter window lets a flow more quickly pick up new (higher)587	minimum RTT when it is moved to a longer path (e.g., due to traffic588	engineering). A longer window makes the filter more resistant to RTT589	inflations such as transient congestion. The unit is seconds.590 591	Possible values: 0 - 86400 (1 day)592 593	Default: 300594 595tcp_moderate_rcvbuf - BOOLEAN596	If set, TCP performs receive buffer auto-tuning, attempting to597	automatically size the buffer (no greater than tcp_rmem[2]) to598	match the size required by the path for full throughput.  Enabled by599	default.600 601tcp_mtu_probing - INTEGER602	Controls TCP Packetization-Layer Path MTU Discovery.  Takes three603	values:604 605	- 0 - Disabled606	- 1 - Disabled by default, enabled when an ICMP black hole detected607	- 2 - Always enabled, use initial MSS of tcp_base_mss.608 609tcp_probe_interval - UNSIGNED INTEGER610	Controls how often to start TCP Packetization-Layer Path MTU611	Discovery reprobe. The default is reprobing every 10 minutes as612	per RFC4821.613 614tcp_probe_threshold - INTEGER615	Controls when TCP Packetization-Layer Path MTU Discovery probing616	will stop in respect to the width of search range in bytes. Default617	is 8 bytes.618 619tcp_no_metrics_save - BOOLEAN620	By default, TCP saves various connection metrics in the route cache621	when the connection closes, so that connections established in the622	near future can use these to set initial conditions.  Usually, this623	increases overall performance, but may sometimes cause performance624	degradation.  If set, TCP will not cache metrics on closing625	connections.626 627tcp_no_ssthresh_metrics_save - BOOLEAN628	Controls whether TCP saves ssthresh metrics in the route cache.629 630	Default is 1, which disables ssthresh metrics.631 632tcp_orphan_retries - INTEGER633	This value influences the timeout of a locally closed TCP connection,634	when RTO retransmissions remain unacknowledged.635	See tcp_retries2 for more details.636 637	The default value is 8.638 639	If your machine is a loaded WEB server,640	you should think about lowering this value, such sockets641	may consume significant resources. Cf. tcp_max_orphans.642 643tcp_recovery - INTEGER644	This value is a bitmap to enable various experimental loss recovery645	features.646 647	=========   =============================================================648	RACK: 0x1   enables the RACK loss detection for fast detection of lost649		    retransmissions and tail drops. It also subsumes and disables650		    RFC6675 recovery for SACK connections.651 652	RACK: 0x2   makes RACK's reordering window static (min_rtt/4).653 654	RACK: 0x4   disables RACK's DUPACK threshold heuristic655	=========   =============================================================656 657	Default: 0x1658 659tcp_reflect_tos - BOOLEAN660	For listening sockets, reuse the DSCP value of the initial SYN message661	for outgoing packets. This allows to have both directions of a TCP662	stream to use the same DSCP value, assuming DSCP remains unchanged for663	the lifetime of the connection.664 665	This options affects both IPv4 and IPv6.666 667	Default: 0 (disabled)668 669tcp_reordering - INTEGER670	Initial reordering level of packets in a TCP stream.671	TCP stack can then dynamically adjust flow reordering level672	between this initial value and tcp_max_reordering673 674	Default: 3675 676tcp_max_reordering - INTEGER677	Maximal reordering level of packets in a TCP stream.678	300 is a fairly conservative value, but you might increase it679	if paths are using per packet load balancing (like bonding rr mode)680 681	Default: 300682 683tcp_retrans_collapse - BOOLEAN684	Bug-to-bug compatibility with some broken printers.685	On retransmit try to send bigger packets to work around bugs in686	certain TCP stacks.687 688tcp_retries1 - INTEGER689	This value influences the time, after which TCP decides, that690	something is wrong due to unacknowledged RTO retransmissions,691	and reports this suspicion to the network layer.692	See tcp_retries2 for more details.693 694	RFC 1122 recommends at least 3 retransmissions, which is the695	default.696 697tcp_retries2 - INTEGER698	This value influences the timeout of an alive TCP connection,699	when RTO retransmissions remain unacknowledged.700	Given a value of N, a hypothetical TCP connection following701	exponential backoff with an initial RTO of TCP_RTO_MIN would702	retransmit N times before killing the connection at the (N+1)th RTO.703 704	The default value of 15 yields a hypothetical timeout of 924.6705	seconds and is a lower bound for the effective timeout.706	TCP will effectively time out at the first RTO which exceeds the707	hypothetical timeout.708 709	RFC 1122 recommends at least 100 seconds for the timeout,710	which corresponds to a value of at least 8.711 712tcp_rfc1337 - BOOLEAN713	If set, the TCP stack behaves conforming to RFC1337. If unset,714	we are not conforming to RFC, but prevent TCP TIME_WAIT715	assassination.716 717	Default: 0718 719tcp_rmem - vector of 3 INTEGERs: min, default, max720	min: Minimal size of receive buffer used by TCP sockets.721	It is guaranteed to each TCP socket, even under moderate memory722	pressure.723 724	Default: 4K725 726	default: initial size of receive buffer used by TCP sockets.727	This value overrides net.core.rmem_default used by other protocols.728	Default: 131072 bytes.729	This value results in initial window of 65535.730 731	max: maximal size of receive buffer allowed for automatically732	selected receiver buffers for TCP socket. This value does not override733	net.core.rmem_max.  Calling setsockopt() with SO_RCVBUF disables734	automatic tuning of that socket's receive buffer size, in which735	case this value is ignored.736	Default: between 131072 and 6MB, depending on RAM size.737 738tcp_sack - BOOLEAN739	Enable select acknowledgments (SACKS).740 741tcp_comp_sack_delay_ns - LONG INTEGER742	TCP tries to reduce number of SACK sent, using a timer743	based on 5% of SRTT, capped by this sysctl, in nano seconds.744	The default is 1ms, based on TSO autosizing period.745 746	Default : 1,000,000 ns (1 ms)747 748tcp_comp_sack_slack_ns - LONG INTEGER749	This sysctl control the slack used when arming the750	timer used by SACK compression. This gives extra time751	for small RTT flows, and reduces system overhead by allowing752	opportunistic reduction of timer interrupts.753 754	Default : 100,000 ns (100 us)755 756tcp_comp_sack_nr - INTEGER757	Max number of SACK that can be compressed.758	Using 0 disables SACK compression.759 760	Default : 44761 762tcp_backlog_ack_defer - BOOLEAN763	If set, user thread processing socket backlog tries sending764	one ACK for the whole queue. This helps to avoid potential765	long latencies at end of a TCP socket syscall.766 767	Default : true768 769tcp_slow_start_after_idle - BOOLEAN770	If set, provide RFC2861 behavior and time out the congestion771	window after an idle period.  An idle period is defined at772	the current RTO.  If unset, the congestion window will not773	be timed out after an idle period.774 775	Default: 1776 777tcp_stdurg - BOOLEAN778	Use the Host requirements interpretation of the TCP urgent pointer field.779	Most hosts use the older BSD interpretation, so if you turn this on780	Linux might not communicate correctly with them.781 782	Default: FALSE783 784tcp_synack_retries - INTEGER785	Number of times SYNACKs for a passive TCP connection attempt will786	be retransmitted. Should not be higher than 255. Default value787	is 5, which corresponds to 31seconds till the last retransmission788	with the current initial RTO of 1second. With this the final timeout789	for a passive TCP connection will happen after 63seconds.790 791tcp_syncookies - INTEGER792	Only valid when the kernel was compiled with CONFIG_SYN_COOKIES793	Send out syncookies when the syn backlog queue of a socket794	overflows. This is to prevent against the common 'SYN flood attack'795	Default: 1796 797	Note, that syncookies is fallback facility.798	It MUST NOT be used to help highly loaded servers to stand799	against legal connection rate. If you see SYN flood warnings800	in your logs, but investigation	shows that they occur801	because of overload with legal connections, you should tune802	another parameters until this warning disappear.803	See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.804 805	syncookies seriously violate TCP protocol, do not allow806	to use TCP extensions, can result in serious degradation807	of some services (f.e. SMTP relaying), visible not by you,808	but your clients and relays, contacting you. While you see809	SYN flood warnings in logs not being really flooded, your server810	is seriously misconfigured.811 812	If you want to test which effects syncookies have to your813	network connections you can set this knob to 2 to enable814	unconditionally generation of syncookies.815 816tcp_migrate_req - BOOLEAN817	The incoming connection is tied to a specific listening socket when818	the initial SYN packet is received during the three-way handshake.819	When a listener is closed, in-flight request sockets during the820	handshake and established sockets in the accept queue are aborted.821 822	If the listener has SO_REUSEPORT enabled, other listeners on the823	same port should have been able to accept such connections. This824	option makes it possible to migrate such child sockets to another825	listener after close() or shutdown().826 827	The BPF_SK_REUSEPORT_SELECT_OR_MIGRATE type of eBPF program should828	usually be used to define the policy to pick an alive listener.829	Otherwise, the kernel will randomly pick an alive listener only if830	this option is enabled.831 832	Note that migration between listeners with different settings may833	crash applications. Let's say migration happens from listener A to834	B, and only B has TCP_SAVE_SYN enabled. B cannot read SYN data from835	the requests migrated from A. To avoid such a situation, cancel836	migration by returning SK_DROP in the type of eBPF program, or837	disable this option.838 839	Default: 0840 841tcp_fastopen - INTEGER842	Enable TCP Fast Open (RFC7413) to send and accept data in the opening843	SYN packet.844 845	The client support is enabled by flag 0x1 (on by default). The client846	then must use sendmsg() or sendto() with the MSG_FASTOPEN flag,847	rather than connect() to send data in SYN.848 849	The server support is enabled by flag 0x2 (off by default). Then850	either enable for all listeners with another flag (0x400) or851	enable individual listeners via TCP_FASTOPEN socket option with852	the option value being the length of the syn-data backlog.853 854	The values (bitmap) are855 856	=====  ======== ======================================================857	  0x1  (client) enables sending data in the opening SYN on the client.858	  0x2  (server) enables the server support, i.e., allowing data in859			a SYN packet to be accepted and passed to the860			application before 3-way handshake finishes.861	  0x4  (client) send data in the opening SYN regardless of cookie862			availability and without a cookie option.863	0x200  (server) accept data-in-SYN w/o any cookie option present.864	0x400  (server) enable all listeners to support Fast Open by865			default without explicit TCP_FASTOPEN socket option.866	=====  ======== ======================================================867 868	Default: 0x1869 870	Note that additional client or server features are only871	effective if the basic support (0x1 and 0x2) are enabled respectively.872 873tcp_fastopen_blackhole_timeout_sec - INTEGER874	Initial time period in second to disable Fastopen on active TCP sockets875	when a TFO firewall blackhole issue happens.876	This time period will grow exponentially when more blackhole issues877	get detected right after Fastopen is re-enabled and will reset to878	initial value when the blackhole issue goes away.879	0 to disable the blackhole detection.880 881	By default, it is set to 0 (feature is disabled).882 883tcp_fastopen_key - list of comma separated 32-digit hexadecimal INTEGERs884	The list consists of a primary key and an optional backup key. The885	primary key is used for both creating and validating cookies, while the886	optional backup key is only used for validating cookies. The purpose of887	the backup key is to maximize TFO validation when keys are rotated.888 889	A randomly chosen primary key may be configured by the kernel if890	the tcp_fastopen sysctl is set to 0x400 (see above), or if the891	TCP_FASTOPEN setsockopt() optname is set and a key has not been892	previously configured via sysctl. If keys are configured via893	setsockopt() by using the TCP_FASTOPEN_KEY optname, then those894	per-socket keys will be used instead of any keys that are specified via895	sysctl.896 897	A key is specified as 4 8-digit hexadecimal integers which are separated898	by a '-' as: xxxxxxxx-xxxxxxxx-xxxxxxxx-xxxxxxxx. Leading zeros may be899	omitted. A primary and a backup key may be specified by separating them900	by a comma. If only one key is specified, it becomes the primary key and901	any previously configured backup keys are removed.902 903tcp_syn_retries - INTEGER904	Number of times initial SYNs for an active TCP connection attempt905	will be retransmitted. Should not be higher than 127. Default value906	is 6, which corresponds to 67seconds (with tcp_syn_linear_timeouts = 4)907	till the last retransmission with the current initial RTO of 1second.908	With this the final timeout for an active TCP connection attempt909	will happen after 131seconds.910 911tcp_timestamps - INTEGER912	Enable timestamps as defined in RFC1323.913 914	- 0: Disabled.915	- 1: Enable timestamps as defined in RFC1323 and use random offset for916	  each connection rather than only using the current time.917	- 2: Like 1, but without random offsets.918 919	Default: 1920 921tcp_min_tso_segs - INTEGER922	Minimal number of segments per TSO frame.923 924	Since linux-3.12, TCP does an automatic sizing of TSO frames,925	depending on flow rate, instead of filling 64Kbytes packets.926	For specific usages, it's possible to force TCP to build big927	TSO frames. Note that TCP stack might split too big TSO packets928	if available window is too small.929 930	Default: 2931 932tcp_tso_rtt_log - INTEGER933	Adjustment of TSO packet sizes based on min_rtt934 935	Starting from linux-5.18, TCP autosizing can be tweaked936	for flows having small RTT.937 938	Old autosizing was splitting the pacing budget to send 1024 TSO939	per second.940 941	tso_packet_size = sk->sk_pacing_rate / 1024;942 943	With the new mechanism, we increase this TSO sizing using:944 945	distance = min_rtt_usec / (2^tcp_tso_rtt_log)946	tso_packet_size += gso_max_size >> distance;947 948	This means that flows between very close hosts can use bigger949	TSO packets, reducing their cpu costs.950 951	If you want to use the old autosizing, set this sysctl to 0.952 953	Default: 9  (2^9 = 512 usec)954 955tcp_pacing_ss_ratio - INTEGER956	sk->sk_pacing_rate is set by TCP stack using a ratio applied957	to current rate. (current_rate = cwnd * mss / srtt)958	If TCP is in slow start, tcp_pacing_ss_ratio is applied959	to let TCP probe for bigger speeds, assuming cwnd can be960	doubled every other RTT.961 962	Default: 200963 964tcp_pacing_ca_ratio - INTEGER965	sk->sk_pacing_rate is set by TCP stack using a ratio applied966	to current rate. (current_rate = cwnd * mss / srtt)967	If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio968	is applied to conservatively probe for bigger throughput.969 970	Default: 120971 972tcp_syn_linear_timeouts - INTEGER973	The number of times for an active TCP connection to retransmit SYNs with974	a linear backoff timeout before defaulting to an exponential backoff975	timeout. This has no effect on SYNACK at the passive TCP side.976 977	With an initial RTO of 1 and tcp_syn_linear_timeouts = 4 we would978	expect SYN RTOs to be: 1, 1, 1, 1, 1, 2, 4, ... (4 linear timeouts,979	and the first exponential backoff using 2^0 * initial_RTO).980	Default: 4981 982tcp_tso_win_divisor - INTEGER983	This allows control over what percentage of the congestion window984	can be consumed by a single TSO frame.985	The setting of this parameter is a choice between burstiness and986	building larger TSO frames.987 988	Default: 3989 990tcp_tw_reuse - INTEGER991	Enable reuse of TIME-WAIT sockets for new connections when it is992	safe from protocol viewpoint.993 994	- 0 - disable995	- 1 - global enable996	- 2 - enable for loopback traffic only997 998	It should not be changed without advice/request of technical999	experts.1000 1001	Default: 21002 1003tcp_window_scaling - BOOLEAN1004	Enable window scaling as defined in RFC1323.1005 1006tcp_shrink_window - BOOLEAN1007	This changes how the TCP receive window is calculated.1008 1009	RFC 7323, section 2.4, says there are instances when a retracted1010	window can be offered, and that TCP implementations MUST ensure1011	that they handle a shrinking window, as specified in RFC 1122.1012 1013	- 0 - Disabled.	The window is never shrunk.1014	- 1 - Enabled.	The window is shrunk when necessary to remain within1015			the memory limit set by autotuning (sk_rcvbuf).1016			This only occurs if a non-zero receive window1017			scaling factor is also in effect.1018 1019	Default: 01020 1021tcp_wmem - vector of 3 INTEGERs: min, default, max1022	min: Amount of memory reserved for send buffers for TCP sockets.1023	Each TCP socket has rights to use it due to fact of its birth.1024 1025	Default: 4K1026 1027	default: initial size of send buffer used by TCP sockets.  This1028	value overrides net.core.wmem_default used by other protocols.1029 1030	It is usually lower than net.core.wmem_default.1031 1032	Default: 16K1033 1034	max: Maximal amount of memory allowed for automatically tuned1035	send buffers for TCP sockets. This value does not override1036	net.core.wmem_max.  Calling setsockopt() with SO_SNDBUF disables1037	automatic tuning of that socket's send buffer size, in which case1038	this value is ignored.1039 1040	Default: between 64K and 4MB, depending on RAM size.1041 1042tcp_notsent_lowat - UNSIGNED INTEGER1043	A TCP socket can control the amount of unsent bytes in its write queue,1044	thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll()1045	reports POLLOUT events if the amount of unsent bytes is below a per1046	socket value, and if the write queue is not full. sendmsg() will1047	also not add new buffers if the limit is hit.1048 1049	This global variable controls the amount of unsent data for1050	sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change1051	to the global variable has immediate effect.1052 1053	Default: UINT_MAX (0xFFFFFFFF)1054 1055tcp_workaround_signed_windows - BOOLEAN1056	If set, assume no receipt of a window scaling option means the1057	remote TCP is broken and treats the window as a signed quantity.1058	If unset, assume the remote TCP is not broken even if we do1059	not receive a window scaling option from them.1060 1061	Default: 01062 1063tcp_thin_linear_timeouts - BOOLEAN1064	Enable dynamic triggering of linear timeouts for thin streams.1065	If set, a check is performed upon retransmission by timeout to1066	determine if the stream is thin (less than 4 packets in flight).1067	As long as the stream is found to be thin, up to 6 linear1068	timeouts may be performed before exponential backoff mode is1069	initiated. This improves retransmission latency for1070	non-aggressive thin streams, often found to be time-dependent.1071	For more information on thin streams, see1072	Documentation/networking/tcp-thin.rst1073 1074	Default: 01075 1076tcp_limit_output_bytes - INTEGER1077	Controls TCP Small Queue limit per tcp socket.1078	TCP bulk sender tends to increase packets in flight until it1079	gets losses notifications. With SNDBUF autotuning, this can1080	result in a large amount of packets queued on the local machine1081	(e.g.: qdiscs, CPU backlog, or device) hurting latency of other1082	flows, for typical pfifo_fast qdiscs.  tcp_limit_output_bytes1083	limits the number of bytes on qdisc or device to reduce artificial1084	RTT/cwnd and reduce bufferbloat.1085 1086	Default: 1048576 (16 * 65536)1087 1088tcp_challenge_ack_limit - INTEGER1089	Limits number of Challenge ACK sent per second, as recommended1090	in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)1091	Note that this per netns rate limit can allow some side channel1092	attacks and probably should not be enabled.1093	TCP stack implements per TCP socket limits anyway.1094	Default: INT_MAX (unlimited)1095 1096tcp_ehash_entries - INTEGER1097	Show the number of hash buckets for TCP sockets in the current1098	networking namespace.1099 1100	A negative value means the networking namespace does not own its1101	hash buckets and shares the initial networking namespace's one.1102 1103tcp_child_ehash_entries - INTEGER1104	Control the number of hash buckets for TCP sockets in the child1105	networking namespace, which must be set before clone() or unshare().1106 1107	If the value is not 0, the kernel uses a value rounded up to 2^n1108	as the actual hash bucket size.  0 is a special value, meaning1109	the child networking namespace will share the initial networking1110	namespace's hash buckets.1111 1112	Note that the child will use the global one in case the kernel1113	fails to allocate enough memory.  In addition, the global hash1114	buckets are spread over available NUMA nodes, but the allocation1115	of the child hash table depends on the current process's NUMA1116	policy, which could result in performance differences.1117 1118	Note also that the default value of tcp_max_tw_buckets and1119	tcp_max_syn_backlog depend on the hash bucket size.1120 1121	Possible values: 0, 2^n (n: 0 - 24 (16Mi))1122 1123	Default: 01124 1125tcp_plb_enabled - BOOLEAN1126	If set and the underlying congestion control (e.g. DCTCP) supports1127	and enables PLB feature, TCP PLB (Protective Load Balancing) is1128	enabled. PLB is described in the following paper:1129	https://doi.org/10.1145/3544216.3544226. Based on PLB parameters,1130	upon sensing sustained congestion, TCP triggers a change in1131	flow label field for outgoing IPv6 packets. A change in flow label1132	field potentially changes the path of outgoing packets for switches1133	that use ECMP/WCMP for routing.1134 1135	PLB changes socket txhash which results in a change in IPv6 Flow Label1136	field, and currently no-op for IPv4 headers. It is possible1137	to apply PLB for IPv4 with other network header fields (e.g. TCP1138	or IPv4 options) or using encapsulation where outer header is used1139	by switches to determine next hop. In either case, further host1140	and switch side changes will be needed.1141 1142	When set, PLB assumes that congestion signal (e.g. ECN) is made1143	available and used by congestion control module to estimate a1144	congestion measure (e.g. ce_ratio). PLB needs a congestion measure to1145	make repathing decisions.1146 1147	Default: FALSE1148 1149tcp_plb_idle_rehash_rounds - INTEGER1150	Number of consecutive congested rounds (RTT) seen after which1151	a rehash can be performed, given there are no packets in flight.1152	This is referred to as M in PLB paper:1153	https://doi.org/10.1145/3544216.3544226.1154 1155	Possible Values: 0 - 311156 1157	Default: 31158 1159tcp_plb_rehash_rounds - INTEGER1160	Number of consecutive congested rounds (RTT) seen after which1161	a forced rehash can be performed. Be careful when setting this1162	parameter, as a small value increases the risk of retransmissions.1163	This is referred to as N in PLB paper:1164	https://doi.org/10.1145/3544216.3544226.1165 1166	Possible Values: 0 - 311167 1168	Default: 121169 1170tcp_plb_suspend_rto_sec - INTEGER1171	Time, in seconds, to suspend PLB in event of an RTO. In order to avoid1172	having PLB repath onto a connectivity "black hole", after an RTO a TCP1173	connection suspends PLB repathing for a random duration between 1x and1174	2x of this parameter. Randomness is added to avoid concurrent rehashing1175	of multiple TCP connections. This should be set corresponding to the1176	amount of time it takes to repair a failed link.1177 1178	Possible Values: 0 - 2551179 1180	Default: 601181 1182tcp_plb_cong_thresh - INTEGER1183	Fraction of packets marked with congestion over a round (RTT) to1184	tag that round as congested. This is referred to as K in the PLB paper:1185	https://doi.org/10.1145/3544216.3544226.1186 1187	The 0-1 fraction range is mapped to 0-256 range to avoid floating1188	point operations. For example, 128 means that if at least 50% of1189	the packets in a round were marked as congested then the round1190	will be tagged as congested.1191 1192	Setting threshold to 0 means that PLB repaths every RTT regardless1193	of congestion. This is not intended behavior for PLB and should be1194	used only for experimentation purpose.1195 1196	Possible Values: 0 - 2561197 1198	Default: 1281199 1200tcp_pingpong_thresh - INTEGER1201	The number of estimated data replies sent for estimated incoming data1202	requests that must happen before TCP considers that a connection is a1203	"ping-pong" (request-response) connection for which delayed1204	acknowledgments can provide benefits.1205 1206	This threshold is 1 by default, but some applications may need a higher1207	threshold for optimal performance.1208 1209	Possible Values: 1 - 2551210 1211	Default: 11212 1213tcp_rto_min_us - INTEGER1214	Minimal TCP retransmission timeout (in microseconds). Note that the1215	rto_min route option has the highest precedence for configuring this1216	setting, followed by the TCP_BPF_RTO_MIN socket option, followed by1217	this tcp_rto_min_us sysctl.1218 1219	The recommended practice is to use a value less or equal to 2000001220	microseconds.1221 1222	Possible Values: 1 - INT_MAX1223 1224	Default: 2000001225 1226UDP variables1227=============1228 1229udp_l3mdev_accept - BOOLEAN1230	Enabling this option allows a "global" bound socket to work1231	across L3 master domains (e.g., VRFs) with packets capable of1232	being received regardless of the L3 domain in which they1233	originated. Only valid when the kernel was compiled with1234	CONFIG_NET_L3_MASTER_DEV.1235 1236	Default: 0 (disabled)1237 1238udp_mem - vector of 3 INTEGERs: min, pressure, max1239	Number of pages allowed for queueing by all UDP sockets.1240 1241	min: Number of pages allowed for queueing by all UDP sockets.1242 1243	pressure: This value was introduced to follow format of tcp_mem.1244 1245	max: This value was introduced to follow format of tcp_mem.1246 1247	Default is calculated at boot time from amount of available memory.1248 1249udp_rmem_min - INTEGER1250	Minimal size of receive buffer used by UDP sockets in moderation.1251	Each UDP socket is able to use the size for receiving data, even if1252	total pages of UDP sockets exceed udp_mem pressure. The unit is byte.1253 1254	Default: 4K1255 1256udp_wmem_min - INTEGER1257	UDP does not have tx memory accounting and this tunable has no effect.1258 1259udp_hash_entries - INTEGER1260	Show the number of hash buckets for UDP sockets in the current1261	networking namespace.1262 1263	A negative value means the networking namespace does not own its1264	hash buckets and shares the initial networking namespace's one.1265 1266udp_child_ehash_entries - INTEGER1267	Control the number of hash buckets for UDP sockets in the child1268	networking namespace, which must be set before clone() or unshare().1269 1270	If the value is not 0, the kernel uses a value rounded up to 2^n1271	as the actual hash bucket size.  0 is a special value, meaning1272	the child networking namespace will share the initial networking1273	namespace's hash buckets.1274 1275	Note that the child will use the global one in case the kernel1276	fails to allocate enough memory.  In addition, the global hash1277	buckets are spread over available NUMA nodes, but the allocation1278	of the child hash table depends on the current process's NUMA1279	policy, which could result in performance differences.1280 1281	Possible values: 0, 2^n (n: 7 (128) - 16 (64K))1282 1283	Default: 01284 1285 1286RAW variables1287=============1288 1289raw_l3mdev_accept - BOOLEAN1290	Enabling this option allows a "global" bound socket to work1291	across L3 master domains (e.g., VRFs) with packets capable of1292	being received regardless of the L3 domain in which they1293	originated. Only valid when the kernel was compiled with1294	CONFIG_NET_L3_MASTER_DEV.1295 1296	Default: 1 (enabled)1297 1298CIPSOv4 Variables1299=================1300 1301cipso_cache_enable - BOOLEAN1302	If set, enable additions to and lookups from the CIPSO label mapping1303	cache.  If unset, additions are ignored and lookups always result in a1304	miss.  However, regardless of the setting the cache is still1305	invalidated when required when means you can safely toggle this on and1306	off and the cache will always be "safe".1307 1308	Default: 11309 1310cipso_cache_bucket_size - INTEGER1311	The CIPSO label cache consists of a fixed size hash table with each1312	hash bucket containing a number of cache entries.  This variable limits1313	the number of entries in each hash bucket; the larger the value is, the1314	more CIPSO label mappings that can be cached.  When the number of1315	entries in a given hash bucket reaches this limit adding new entries1316	causes the oldest entry in the bucket to be removed to make room.1317 1318	Default: 101319 1320cipso_rbm_optfmt - BOOLEAN1321	Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of1322	the CIPSO draft specification (see Documentation/netlabel for details).1323	This means that when set the CIPSO tag will be padded with empty1324	categories in order to make the packet data 32-bit aligned.1325 1326	Default: 01327 1328cipso_rbm_structvalid - BOOLEAN1329	If set, do a very strict check of the CIPSO option when1330	ip_options_compile() is called.  If unset, relax the checks done during1331	ip_options_compile().  Either way is "safe" as errors are caught else1332	where in the CIPSO processing code but setting this to 0 (False) should1333	result in less work (i.e. it should be faster) but could cause problems1334	with other implementations that require strict checking.1335 1336	Default: 01337 1338IP Variables1339============1340 1341ip_local_port_range - 2 INTEGERS1342	Defines the local port range that is used by TCP and UDP to1343	choose the local port. The first number is the first, the1344	second the last local port number.1345	If possible, it is better these numbers have different parity1346	(one even and one odd value).1347	Must be greater than or equal to ip_unprivileged_port_start.1348	The default values are 32768 and 60999 respectively.1349 1350ip_local_reserved_ports - list of comma separated ranges1351	Specify the ports which are reserved for known third-party1352	applications. These ports will not be used by automatic port1353	assignments (e.g. when calling connect() or bind() with port1354	number 0). Explicit port allocation behavior is unchanged.1355 1356	The format used for both input and output is a comma separated1357	list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and1358	10). Writing to the file will clear all previously reserved1359	ports and update the current list with the one given in the1360	input.1361 1362	Note that ip_local_port_range and ip_local_reserved_ports1363	settings are independent and both are considered by the kernel1364	when determining which ports are available for automatic port1365	assignments.1366 1367	You can reserve ports which are not in the current1368	ip_local_port_range, e.g.::1369 1370	    $ cat /proc/sys/net/ipv4/ip_local_port_range1371	    32000	609991372	    $ cat /proc/sys/net/ipv4/ip_local_reserved_ports1373	    8080,91481374 1375	although this is redundant. However such a setting is useful1376	if later the port range is changed to a value that will1377	include the reserved ports. Also keep in mind, that overlapping1378	of these ranges may affect probability of selecting ephemeral1379	ports which are right after block of reserved ports.1380 1381	Default: Empty1382 1383ip_unprivileged_port_start - INTEGER1384	This is a per-namespace sysctl.  It defines the first1385	unprivileged port in the network namespace.  Privileged ports1386	require root or CAP_NET_BIND_SERVICE in order to bind to them.1387	To disable all privileged ports, set this to 0.  They must not1388	overlap with the ip_local_port_range.1389 1390	Default: 10241391 1392ip_nonlocal_bind - BOOLEAN1393	If set, allows processes to bind() to non-local IP addresses,1394	which can be quite useful - but may break some applications.1395 1396	Default: 01397 1398ip_autobind_reuse - BOOLEAN1399	By default, bind() does not select the ports automatically even if1400	the new socket and all sockets bound to the port have SO_REUSEADDR.1401	ip_autobind_reuse allows bind() to reuse the port and this is useful1402	when you use bind()+connect(), but may break some applications.1403	The preferred solution is to use IP_BIND_ADDRESS_NO_PORT and this1404	option should only be set by experts.1405	Default: 01406 1407ip_dynaddr - INTEGER1408	If set non-zero, enables support for dynamic addresses.1409	If set to a non-zero value larger than 1, a kernel log1410	message will be printed when dynamic address rewriting1411	occurs.1412 1413	Default: 01414 1415ip_early_demux - BOOLEAN1416	Optimize input packet processing down to one demux for1417	certain kinds of local sockets.  Currently we only do this1418	for established TCP and connected UDP sockets.1419 1420	It may add an additional cost for pure routing workloads that1421	reduces overall throughput, in such case you should disable it.1422 1423	Default: 11424 1425ping_group_range - 2 INTEGERS1426	Restrict ICMP_PROTO datagram sockets to users in the group range.1427	The default is "1 0", meaning, that nobody (not even root) may1428	create ping sockets.  Setting it to "100 100" would grant permissions1429	to the single group. "0 4294967294" would enable it for the world, "1001430	4294967294" would enable it for the users, but not daemons.1431 1432tcp_early_demux - BOOLEAN1433	Enable early demux for established TCP sockets.1434 1435	Default: 11436 1437udp_early_demux - BOOLEAN1438	Enable early demux for connected UDP sockets. Disable this if1439	your system could experience more unconnected load.1440 1441	Default: 11442 1443icmp_echo_ignore_all - BOOLEAN1444	If set non-zero, then the kernel will ignore all ICMP ECHO1445	requests sent to it.1446 1447	Default: 01448 1449icmp_echo_enable_probe - BOOLEAN1450        If set to one, then the kernel will respond to RFC 8335 PROBE1451        requests sent to it.1452 1453        Default: 01454 1455icmp_echo_ignore_broadcasts - BOOLEAN1456	If set non-zero, then the kernel will ignore all ICMP ECHO and1457	TIMESTAMP requests sent to it via broadcast/multicast.1458 1459	Default: 11460 1461icmp_ratelimit - INTEGER1462	Limit the maximal rates for sending ICMP packets whose type matches1463	icmp_ratemask (see below) to specific targets.1464	0 to disable any limiting,1465	otherwise the minimal space between responses in milliseconds.1466	Note that another sysctl, icmp_msgs_per_sec limits the number1467	of ICMP packets	sent on all targets.1468 1469	Default: 10001470 1471icmp_msgs_per_sec - INTEGER1472	Limit maximal number of ICMP packets sent per second from this host.1473	Only messages whose type matches icmp_ratemask (see below) are1474	controlled by this limit. For security reasons, the precise count1475	of messages per second is randomized.1476 1477	Default: 10001478 1479icmp_msgs_burst - INTEGER1480	icmp_msgs_per_sec controls number of ICMP packets sent per second,1481	while icmp_msgs_burst controls the burst size of these packets.1482	For security reasons, the precise burst size is randomized.1483 1484	Default: 501485 1486icmp_ratemask - INTEGER1487	Mask made of ICMP types for which rates are being limited.1488 1489	Significant bits: IHGFEDCBA98765432101490 1491	Default mask:     0000001100000011000 (6168)1492 1493	Bit definitions (see include/linux/icmp.h):1494 1495		= =========================1496		0 Echo Reply1497		3 Destination Unreachable [1]_1498		4 Source Quench [1]_1499		5 Redirect1500		8 Echo Request1501		B Time Exceeded [1]_1502		C Parameter Problem [1]_1503		D Timestamp Request1504		E Timestamp Reply1505		F Info Request1506		G Info Reply1507		H Address Mask Request1508		I Address Mask Reply1509		= =========================1510 1511	.. [1] These are rate limited by default (see default mask above)1512 1513icmp_ignore_bogus_error_responses - BOOLEAN1514	Some routers violate RFC1122 by sending bogus responses to broadcast1515	frames.  Such violations are normally logged via a kernel warning.1516	If this is set to TRUE, the kernel will not give such warnings, which1517	will avoid log file clutter.1518 1519	Default: 11520 1521icmp_errors_use_inbound_ifaddr - BOOLEAN1522 1523	If zero, icmp error messages are sent with the primary address of1524	the exiting interface.1525 1526	If non-zero, the message will be sent with the primary address of1527	the interface that received the packet that caused the icmp error.1528	This is the behaviour many network administrators will expect from1529	a router. And it can make debugging complicated network layouts1530	much easier.1531 1532	Note that if no primary address exists for the interface selected,1533	then the primary address of the first non-loopback interface that1534	has one will be used regardless of this setting.1535 1536	Default: 01537 1538igmp_max_memberships - INTEGER1539	Change the maximum number of multicast groups we can subscribe to.1540	Default: 201541 1542	Theoretical maximum value is bounded by having to send a membership1543	report in a single datagram (i.e. the report can't span multiple1544	datagrams, or risk confusing the switch and leaving groups you don't1545	intend to).1546 1547	The number of supported groups 'M' is bounded by the number of group1548	report entries you can fit into a single datagram of 65535 bytes.1549 1550	M = 65536-sizeof (ip header)/(sizeof(Group record))1551 1552	Group records are variable length, with a minimum of 12 bytes.1553	So net.ipv4.igmp_max_memberships should not be set higher than:1554 1555	(65536-24) / 12 = 54591556 1557	The value 5459 assumes no IP header options, so in practice1558	this number may be lower.1559 1560igmp_max_msf - INTEGER1561	Maximum number of addresses allowed in the source filter list for a1562	multicast group.1563 1564	Default: 101565 1566igmp_qrv - INTEGER1567	Controls the IGMP query robustness variable (see RFC2236 8.1).1568 1569	Default: 2 (as specified by RFC2236 8.1)1570 1571	Minimum: 1 (as specified by RFC6636 4.5)1572 1573force_igmp_version - INTEGER1574	- 0 - (default) No enforcement of a IGMP version, IGMPv1/v2 fallback1575	  allowed. Will back to IGMPv3 mode again if all IGMPv1/v2 Querier1576	  Present timer expires.1577	- 1 - Enforce to use IGMP version 1. Will also reply IGMPv1 report if1578	  receive IGMPv2/v3 query.1579	- 2 - Enforce to use IGMP version 2. Will fallback to IGMPv1 if receive1580	  IGMPv1 query message. Will reply report if receive IGMPv3 query.1581	- 3 - Enforce to use IGMP version 3. The same react with default 0.1582 1583	.. note::1584 1585	   this is not the same with force_mld_version because IGMPv3 RFC33761586	   Security Considerations does not have clear description that we could1587	   ignore other version messages completely as MLDv2 RFC3810. So make1588	   this value as default 0 is recommended.1589 1590``conf/interface/*``1591	changes special settings per interface (where1592	interface" is the name of your network interface)1593 1594``conf/all/*``1595	  is special, changes the settings for all interfaces1596 1597log_martians - BOOLEAN1598	Log packets with impossible addresses to kernel log.1599	log_martians for the interface will be enabled if at least one of1600	conf/{all,interface}/log_martians is set to TRUE,1601	it will be disabled otherwise1602 1603accept_redirects - BOOLEAN1604	Accept ICMP redirect messages.1605	accept_redirects for the interface will be enabled if:1606 1607	- both conf/{all,interface}/accept_redirects are TRUE in the case1608	  forwarding for the interface is enabled1609 1610	or1611 1612	- at least one of conf/{all,interface}/accept_redirects is TRUE in the1613	  case forwarding for the interface is disabled1614 1615	accept_redirects for the interface will be disabled otherwise1616 1617	default:1618 1619		- TRUE (host)1620		- FALSE (router)1621 1622forwarding - BOOLEAN1623	Enable IP forwarding on this interface.  This controls whether packets1624	received _on_ this interface can be forwarded.1625 1626mc_forwarding - BOOLEAN1627	Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE1628	and a multicast routing daemon is required.1629	conf/all/mc_forwarding must also be set to TRUE to enable multicast1630	routing	for the interface1631 1632medium_id - INTEGER1633	Integer value used to differentiate the devices by the medium they1634	are attached to. Two devices can have different id values when1635	the broadcast packets are received only on one of them.1636	The default value 0 means that the device is the only interface1637	to its medium, value of -1 means that medium is not known.1638 1639	Currently, it is used to change the proxy_arp behavior:1640	the proxy_arp feature is enabled for packets forwarded between1641	two devices attached to different media.1642 1643proxy_arp - BOOLEAN1644	Do proxy arp.1645 1646	proxy_arp for the interface will be enabled if at least one of1647	conf/{all,interface}/proxy_arp is set to TRUE,1648	it will be disabled otherwise1649 1650proxy_arp_pvlan - BOOLEAN1651	Private VLAN proxy arp.1652 1653	Basically allow proxy arp replies back to the same interface1654	(from which the ARP request/solicitation was received).1655 1656	This is done to support (ethernet) switch features, like RFC1657	3069, where the individual ports are NOT allowed to1658	communicate with each other, but they are allowed to talk to1659	the upstream router.  As described in RFC 3069, it is possible1660	to allow these hosts to communicate through the upstream1661	router by proxy_arp'ing. Don't need to be used together with1662	proxy_arp.1663 1664	This technology is known by different names:1665 1666	  In RFC 3069 it is called VLAN Aggregation.1667	  Cisco and Allied Telesyn call it Private VLAN.1668	  Hewlett-Packard call it Source-Port filtering or port-isolation.1669	  Ericsson call it MAC-Forced Forwarding (RFC Draft).1670 1671proxy_delay - INTEGER1672	Delay proxy response.1673 1674	Delay response to a neighbor solicitation when proxy_arp1675	or proxy_ndp is enabled. A random value between [0, proxy_delay)1676	will be chosen, setting to zero means reply with no delay.1677	Value in jiffies. Defaults to 80.1678 1679shared_media - BOOLEAN1680	Send(router) or accept(host) RFC1620 shared media redirects.1681	Overrides secure_redirects.1682 1683	shared_media for the interface will be enabled if at least one of1684	conf/{all,interface}/shared_media is set to TRUE,1685	it will be disabled otherwise1686 1687	default TRUE1688 1689secure_redirects - BOOLEAN1690	Accept ICMP redirect messages only to gateways listed in the1691	interface's current gateway list. Even if disabled, RFC1122 redirect1692	rules still apply.1693 1694	Overridden by shared_media.1695 1696	secure_redirects for the interface will be enabled if at least one of1697	conf/{all,interface}/secure_redirects is set to TRUE,1698	it will be disabled otherwise1699 1700	default TRUE1701 1702send_redirects - BOOLEAN1703	Send redirects, if router.1704 1705	send_redirects for the interface will be enabled if at least one of1706	conf/{all,interface}/send_redirects is set to TRUE,1707	it will be disabled otherwise1708 1709	Default: TRUE1710 1711bootp_relay - BOOLEAN1712	Accept packets with source address 0.b.c.d destined1713	not to this host as local ones. It is supposed, that1714	BOOTP relay daemon will catch and forward such packets.1715	conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay1716	for the interface1717 1718	default FALSE1719 1720	Not Implemented Yet.1721 1722accept_source_route - BOOLEAN1723	Accept packets with SRR option.1724	conf/all/accept_source_route must also be set to TRUE to accept packets1725	with SRR option on the interface1726 1727	default1728 1729		- TRUE (router)1730		- FALSE (host)1731 1732accept_local - BOOLEAN1733	Accept packets with local source addresses. In combination with1734	suitable routing, this can be used to direct packets between two1735	local interfaces over the wire and have them accepted properly.1736	default FALSE1737 1738route_localnet - BOOLEAN1739	Do not consider loopback addresses as martian source or destination1740	while routing. This enables the use of 127/8 for local routing purposes.1741 1742	default FALSE1743 1744rp_filter - INTEGER1745	- 0 - No source validation.1746	- 1 - Strict mode as defined in RFC3704 Strict Reverse Path1747	  Each incoming packet is tested against the FIB and if the interface1748	  is not the best reverse path the packet check will fail.1749	  By default failed packets are discarded.1750	- 2 - Loose mode as defined in RFC3704 Loose Reverse Path1751	  Each incoming packet's source address is also tested against the FIB1752	  and if the source address is not reachable via any interface1753	  the packet check will fail.1754 1755	Current recommended practice in RFC3704 is to enable strict mode1756	to prevent IP spoofing from DDos attacks. If using asymmetric routing1757	or other complicated routing, then loose mode is recommended.1758 1759	The max value from conf/{all,interface}/rp_filter is used1760	when doing source validation on the {interface}.1761 1762	Default value is 0. Note that some distributions enable it1763	in startup scripts.1764 1765src_valid_mark - BOOLEAN1766	- 0 - The fwmark of the packet is not included in reverse path1767	  route lookup.  This allows for asymmetric routing configurations1768	  utilizing the fwmark in only one direction, e.g., transparent1769	  proxying.1770 1771	- 1 - The fwmark of the packet is included in reverse path route1772	  lookup.  This permits rp_filter to function when the fwmark is1773	  used for routing traffic in both directions.1774 1775	This setting also affects the utilization of fmwark when1776	performing source address selection for ICMP replies, or1777	determining addresses stored for the IPOPT_TS_TSANDADDR and1778	IPOPT_RR IP options.1779 1780	The max value from conf/{all,interface}/src_valid_mark is used.1781 1782	Default value is 0.1783 1784arp_filter - BOOLEAN1785	- 1 - Allows you to have multiple network interfaces on the same1786	  subnet, and have the ARPs for each interface be answered1787	  based on whether or not the kernel would route a packet from1788	  the ARP'd IP out that interface (therefore you must use source1789	  based routing for this to work). In other words it allows control1790	  of which cards (usually 1) will respond to an arp request.1791 1792	- 0 - (default) The kernel can respond to arp requests with addresses1793	  from other interfaces. This may seem wrong but it usually makes1794	  sense, because it increases the chance of successful communication.1795	  IP addresses are owned by the complete host on Linux, not by1796	  particular interfaces. Only for more complex setups like load-1797	  balancing, does this behaviour cause problems.1798 1799	arp_filter for the interface will be enabled if at least one of1800	conf/{all,interface}/arp_filter is set to TRUE,1801	it will be disabled otherwise1802 1803arp_announce - INTEGER1804	Define different restriction levels for announcing the local1805	source IP address from IP packets in ARP requests sent on1806	interface:1807 1808	- 0 - (default) Use any local address, configured on any interface1809	- 1 - Try to avoid local addresses that are not in the target's1810	  subnet for this interface. This mode is useful when target1811	  hosts reachable via this interface require the source IP1812	  address in ARP requests to be part of their logical network1813	  configured on the receiving interface. When we generate the1814	  request we will check all our subnets that include the1815	  target IP and will preserve the source address if it is from1816	  such subnet. If there is no such subnet we select source1817	  address according to the rules for level 2.1818	- 2 - Always use the best local address for this target.1819	  In this mode we ignore the source address in the IP packet1820	  and try to select local address that we prefer for talks with1821	  the target host. Such local address is selected by looking1822	  for primary IP addresses on all our subnets on the outgoing1823	  interface that include the target IP address. If no suitable1824	  local address is found we select the first local address1825	  we have on the outgoing interface or on all other interfaces,1826	  with the hope we will receive reply for our request and1827	  even sometimes no matter the source IP address we announce.1828 1829	The max value from conf/{all,interface}/arp_announce is used.1830 1831	Increasing the restriction level gives more chance for1832	receiving answer from the resolved target while decreasing1833	the level announces more valid sender's information.1834 1835arp_ignore - INTEGER1836	Define different modes for sending replies in response to1837	received ARP requests that resolve local target IP addresses:1838 1839	- 0 - (default): reply for any local target IP address, configured1840	  on any interface1841	- 1 - reply only if the target IP address is local address1842	  configured on the incoming interface1843	- 2 - reply only if the target IP address is local address1844	  configured on the incoming interface and both with the1845	  sender's IP address are part from same subnet on this interface1846	- 3 - do not reply for local addresses configured with scope host,1847	  only resolutions for global and link addresses are replied1848	- 4-7 - reserved1849	- 8 - do not reply for all local addresses1850 1851	The max value from conf/{all,interface}/arp_ignore is used1852	when ARP request is received on the {interface}1853 1854arp_notify - BOOLEAN1855	Define mode for notification of address and device changes.1856 1857	 ==  ==========================================================1858	  0  (default): do nothing1859	  1  Generate gratuitous arp requests when device is brought up1860	     or hardware address changes.1861	 ==  ==========================================================1862 1863arp_accept - INTEGER1864	Define behavior for accepting gratuitous ARP (garp) frames from devices1865	that are not already present in the ARP table:1866 1867	- 0 - don't create new entries in the ARP table1868	- 1 - create new entries in the ARP table1869	- 2 - create new entries only if the source IP address is in the same1870	  subnet as an address configured on the interface that received the1871	  garp message.1872 1873	Both replies and requests type gratuitous arp will trigger the1874	ARP table to be updated, if this setting is on.1875 1876	If the ARP table already contains the IP address of the1877	gratuitous arp frame, the arp table will be updated regardless1878	if this setting is on or off.1879 1880arp_evict_nocarrier - BOOLEAN1881	Clears the ARP cache on NOCARRIER events. This option is important for1882	wireless devices where the ARP cache should not be cleared when roaming1883	between access points on the same network. In most cases this should1884	remain as the default (1).1885 1886	- 1 - (default): Clear the ARP cache on NOCARRIER events1887	- 0 - Do not clear ARP cache on NOCARRIER events1888 1889mcast_solicit - INTEGER1890	The maximum number of multicast probes in INCOMPLETE state,1891	when the associated hardware address is unknown.  Defaults1892	to 3.1893 1894ucast_solicit - INTEGER1895	The maximum number of unicast probes in PROBE state, when1896	the hardware address is being reconfirmed.  Defaults to 3.1897 1898app_solicit - INTEGER1899	The maximum number of probes to send to the user space ARP daemon1900	via netlink before dropping back to multicast probes (see1901	mcast_resolicit).  Defaults to 0.1902 1903mcast_resolicit - INTEGER1904	The maximum number of multicast probes after unicast and1905	app probes in PROBE state.  Defaults to 0.1906 1907disable_policy - BOOLEAN1908	Disable IPSEC policy (SPD) for this interface1909 1910disable_xfrm - BOOLEAN1911	Disable IPSEC encryption on this interface, whatever the policy1912 1913igmpv2_unsolicited_report_interval - INTEGER1914	The interval in milliseconds in which the next unsolicited1915	IGMPv1 or IGMPv2 report retransmit will take place.1916 1917	Default: 10000 (10 seconds)1918 1919igmpv3_unsolicited_report_interval - INTEGER1920	The interval in milliseconds in which the next unsolicited1921	IGMPv3 report retransmit will take place.1922 1923	Default: 1000 (1 seconds)1924 1925ignore_routes_with_linkdown - BOOLEAN1926        Ignore routes whose link is down when performing a FIB lookup.1927 1928promote_secondaries - BOOLEAN1929	When a primary IP address is removed from this interface1930	promote a corresponding secondary IP address instead of1931	removing all the corresponding secondary IP addresses.1932 1933drop_unicast_in_l2_multicast - BOOLEAN1934	Drop any unicast IP packets that are received in link-layer1935	multicast (or broadcast) frames.1936 1937	This behavior (for multicast) is actually a SHOULD in RFC1938	1122, but is disabled by default for compatibility reasons.1939 1940	Default: off (0)1941 1942drop_gratuitous_arp - BOOLEAN1943	Drop all gratuitous ARP frames, for example if there's a known1944	good ARP proxy on the network and such frames need not be used1945	(or in the case of 802.11, must not be used to prevent attacks.)1946 1947	Default: off (0)1948 1949 1950tag - INTEGER1951	Allows you to write a number, which can be used as required.1952 1953	Default value is 0.1954 1955xfrm4_gc_thresh - INTEGER1956	(Obsolete since linux-4.14)1957	The threshold at which we will start garbage collecting for IPv41958	destination cache entries.  At twice this value the system will1959	refuse new allocations.1960 1961igmp_link_local_mcast_reports - BOOLEAN1962	Enable IGMP reports for link local multicast groups in the1963	224.0.0.X range.1964 1965	Default TRUE1966 1967Alexey Kuznetsov.1968kuznet@ms2.inr.ac.ru1969 1970Updated by:1971 1972- Andi Kleen1973  ak@muc.de1974- Nicolas Delon1975  delon.nicolas@wanadoo.fr1976 1977 1978 1979 1980/proc/sys/net/ipv6/* Variables1981==============================1982 1983IPv6 has no global variables such as tcp_*.  tcp_* settings under ipv4/ also1984apply to IPv6 [XXX?].1985 1986bindv6only - BOOLEAN1987	Default value for IPV6_V6ONLY socket option,1988	which restricts use of the IPv6 socket to IPv6 communication1989	only.1990 1991		- TRUE: disable IPv4-mapped address feature1992		- FALSE: enable IPv4-mapped address feature1993 1994	Default: FALSE (as specified in RFC3493)1995 1996flowlabel_consistency - BOOLEAN1997	Protect the consistency (and unicity) of flow label.1998	You have to disable it to use IPV6_FL_F_REFLECT flag on the1999	flow label manager.2000 2001	- TRUE: enabled2002	- FALSE: disabled2003 2004	Default: TRUE2005 2006auto_flowlabels - INTEGER2007	Automatically generate flow labels based on a flow hash of the2008	packet. This allows intermediate devices, such as routers, to2009	identify packet flows for mechanisms like Equal Cost Multipath2010	Routing (see RFC 6438).2011 2012	=  ===========================================================2013	0  automatic flow labels are completely disabled2014	1  automatic flow labels are enabled by default, they can be2015	   disabled on a per socket basis using the IPV6_AUTOFLOWLABEL2016	   socket option2017	2  automatic flow labels are allowed, they may be enabled on a2018	   per socket basis using the IPV6_AUTOFLOWLABEL socket option2019	3  automatic flow labels are enabled and enforced, they cannot2020	   be disabled by the socket option2021	=  ===========================================================2022 2023	Default: 12024 2025flowlabel_state_ranges - BOOLEAN2026	Split the flow label number space into two ranges. 0-0x7FFFF is2027	reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF2028	is reserved for stateless flow labels as described in RFC6437.2029 2030	- TRUE: enabled2031	- FALSE: disabled2032 2033	Default: true2034 2035flowlabel_reflect - INTEGER2036	Control flow label reflection. Needed for Path MTU2037	Discovery to work with Equal Cost Multipath Routing in anycast2038	environments. See RFC 7690 and:2039	https://tools.ietf.org/html/draft-wang-6man-flow-label-reflection-012040 2041	This is a bitmask.2042 2043	- 1: enabled for established flows2044 2045	  Note that this prevents automatic flowlabel changes, as done2046	  in "tcp: change IPv6 flow-label upon receiving spurious retransmission"2047	  and "tcp: Change txhash on every SYN and RTO retransmit"2048 2049	- 2: enabled for TCP RESET packets (no active listener)2050	  If set, a RST packet sent in response to a SYN packet on a closed2051	  port will reflect the incoming flow label.2052 2053	- 4: enabled for ICMPv6 echo reply messages.2054 2055	Default: 02056 2057fib_multipath_hash_policy - INTEGER2058	Controls which hash policy to use for multipath routes.2059 2060	Default: 0 (Layer 3)2061 2062	Possible values:2063 2064	- 0 - Layer 3 (source and destination addresses plus flow label)2065	- 1 - Layer 4 (standard 5-tuple)2066	- 2 - Layer 3 or inner Layer 3 if present2067	- 3 - Custom multipath hash. Fields used for multipath hash calculation2068	  are determined by fib_multipath_hash_fields sysctl2069 2070fib_multipath_hash_fields - UNSIGNED INTEGER2071	When fib_multipath_hash_policy is set to 3 (custom multipath hash), the2072	fields used for multipath hash calculation are determined by this2073	sysctl.2074 2075	This value is a bitmask which enables various fields for multipath hash2076	calculation.2077 2078	Possible fields are:2079 2080	====== ============================2081	0x0001 Source IP address2082	0x0002 Destination IP address2083	0x0004 IP protocol2084	0x0008 Flow Label2085	0x0010 Source port2086	0x0020 Destination port2087	0x0040 Inner source IP address2088	0x0080 Inner destination IP address2089	0x0100 Inner IP protocol2090	0x0200 Inner Flow Label2091	0x0400 Inner source port2092	0x0800 Inner destination port2093	====== ============================2094 2095	Default: 0x0007 (source IP, destination IP and IP protocol)2096 2097anycast_src_echo_reply - BOOLEAN2098	Controls the use of anycast addresses as source addresses for ICMPv62099	echo reply2100 2101	- TRUE:  enabled2102	- FALSE: disabled2103 2104	Default: FALSE2105 2106idgen_delay - INTEGER2107	Controls the delay in seconds after which time to retry2108	privacy stable address generation if a DAD conflict is2109	detected.2110 2111	Default: 1 (as specified in RFC7217)2112 2113idgen_retries - INTEGER2114	Controls the number of retries to generate a stable privacy2115	address if a DAD conflict is detected.2116 2117	Default: 3 (as specified in RFC7217)2118 2119mld_qrv - INTEGER2120	Controls the MLD query robustness variable (see RFC3810 9.1).2121 2122	Default: 2 (as specified by RFC3810 9.1)2123 2124	Minimum: 1 (as specified by RFC6636 4.5)2125 2126max_dst_opts_number - INTEGER2127	Maximum number of non-padding TLVs allowed in a Destination2128	options extension header. If this value is less than zero2129	then unknown options are disallowed and the number of known2130	TLVs allowed is the absolute value of this number.2131 2132	Default: 82133 2134max_hbh_opts_number - INTEGER2135	Maximum number of non-padding TLVs allowed in a Hop-by-Hop2136	options extension header. If this value is less than zero2137	then unknown options are disallowed and the number of known2138	TLVs allowed is the absolute value of this number.2139 2140	Default: 82141 2142max_dst_opts_length - INTEGER2143	Maximum length allowed for a Destination options extension2144	header.2145 2146	Default: INT_MAX (unlimited)2147 2148max_hbh_length - INTEGER2149	Maximum length allowed for a Hop-by-Hop options extension2150	header.2151 2152	Default: INT_MAX (unlimited)2153 2154skip_notify_on_dev_down - BOOLEAN2155	Controls whether an RTM_DELROUTE message is generated for routes2156	removed when a device is taken down or deleted. IPv4 does not2157	generate this message; IPv6 does by default. Setting this sysctl2158	to true skips the message, making IPv4 and IPv6 on par in relying2159	on userspace caches to track link events and evict routes.2160 2161	Default: false (generate message)2162 2163nexthop_compat_mode - BOOLEAN2164	New nexthop API provides a means for managing nexthops independent of2165	prefixes. Backwards compatibility with old route format is enabled by2166	default which means route dumps and notifications contain the new2167	nexthop attribute but also the full, expanded nexthop definition.2168	Further, updates or deletes of a nexthop configuration generate route2169	notifications for each fib entry using the nexthop. Once a system2170	understands the new API, this sysctl can be disabled to achieve full2171	performance benefits of the new API by disabling the nexthop expansion2172	and extraneous notifications.2173	Default: true (backward compat mode)2174 2175fib_notify_on_flag_change - INTEGER2176        Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/2177        RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed.2178 2179        After installing a route to the kernel, user space receives an2180        acknowledgment, which means the route was installed in the kernel,2181        but not necessarily in hardware.2182        It is also possible for a route already installed in hardware to change2183        its action and therefore its flags. For example, a host route that is2184        trapping packets can be "promoted" to perform decapsulation following2185        the installation of an IPinIP/VXLAN tunnel.2186        The notifications will indicate to user-space the state of the route.2187 2188        Default: 0 (Do not emit notifications.)2189 2190        Possible values:2191 2192        - 0 - Do not emit notifications.2193        - 1 - Emit notifications.2194        - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change.2195 2196ioam6_id - INTEGER2197        Define the IOAM id of this node. Uses only 24 bits out of 32 in total.2198 2199        Min: 02200        Max: 0xFFFFFF2201 2202        Default: 0xFFFFFF2203 2204ioam6_id_wide - LONG INTEGER2205        Define the wide IOAM id of this node. Uses only 56 bits out of 64 in2206        total. Can be different from ioam6_id.2207 2208        Min: 02209        Max: 0xFFFFFFFFFFFFFF2210 2211        Default: 0xFFFFFFFFFFFFFF2212 2213IPv6 Fragmentation:2214 2215ip6frag_high_thresh - INTEGER2216	Maximum memory used to reassemble IPv6 fragments. When2217	ip6frag_high_thresh bytes of memory is allocated for this purpose,2218	the fragment handler will toss packets until ip6frag_low_thresh2219	is reached.2220 2221ip6frag_low_thresh - INTEGER2222	See ip6frag_high_thresh2223 2224ip6frag_time - INTEGER2225	Time in seconds to keep an IPv6 fragment in memory.2226 2227``conf/default/*``:2228	Change the interface-specific default settings.2229 2230	These settings would be used during creating new interfaces.2231 2232 2233``conf/all/*``:2234	Change all the interface-specific settings.2235 2236	[XXX:  Other special features than forwarding?]2237 2238conf/all/disable_ipv6 - BOOLEAN2239	Changing this value is same as changing ``conf/default/disable_ipv6``2240	setting and also all per-interface ``disable_ipv6`` settings to the same2241	value.2242 2243	Reading this value does not have any particular meaning. It does not say2244	whether IPv6 support is enabled or disabled. Returned value can be 12245	also in the case when some interface has ``disable_ipv6`` set to 0 and2246	has configured IPv6 addresses.2247 2248conf/all/forwarding - BOOLEAN2249	Enable global IPv6 forwarding between all interfaces.2250 2251	IPv4 and IPv6 work differently here; e.g. netfilter must be used2252	to control which interfaces may forward packets and which not.2253 2254	This also sets all interfaces' Host/Router setting2255	'forwarding' to the specified value.  See below for details.2256 2257	This referred to as global forwarding.2258 2259proxy_ndp - BOOLEAN2260	Do proxy ndp.2261 2262fwmark_reflect - BOOLEAN2263	Controls the fwmark of kernel-generated IPv6 reply packets that are not2264	associated with a socket for example, TCP RSTs or ICMPv6 echo replies).2265	If unset, these packets have a fwmark of zero. If set, they have the2266	fwmark of the packet they are replying to.2267 2268	Default: 02269 2270``conf/interface/*``:2271	Change special settings per interface.2272 2273	The functional behaviour for certain settings is different2274	depending on whether local forwarding is enabled or not.2275 2276accept_ra - INTEGER2277	Accept Router Advertisements; autoconfigure using them.2278 2279	It also determines whether or not to transmit Router2280	Solicitations. If and only if the functional setting is to2281	accept Router Advertisements, Router Solicitations will be2282	transmitted.2283 2284	Possible values are:2285 2286		==  ===========================================================2287		 0  Do not accept Router Advertisements.2288		 1  Accept Router Advertisements if forwarding is disabled.2289		 2  Overrule forwarding behaviour. Accept Router Advertisements2290		    even if forwarding is enabled.2291		==  ===========================================================2292 2293	Functional default:2294 2295		- enabled if local forwarding is disabled.2296		- disabled if local forwarding is enabled.2297 2298accept_ra_defrtr - BOOLEAN2299	Learn default router in Router Advertisement.2300 2301	Functional default:2302 2303		- enabled if accept_ra is enabled.2304		- disabled if accept_ra is disabled.2305 2306ra_defrtr_metric - UNSIGNED INTEGER2307	Route metric for default route learned in Router Advertisement. This value2308	will be assigned as metric for the default route learned via IPv6 Router2309	Advertisement. Takes affect only if accept_ra_defrtr is enabled.2310 2311	Possible values:2312		1 to 0xFFFFFFFF2313 2314		Default: IP6_RT_PRIO_USER i.e. 1024.2315 2316accept_ra_from_local - BOOLEAN2317	Accept RA with source-address that is found on local machine2318	if the RA is otherwise proper and able to be accepted.2319 2320	Default is to NOT accept these as it may be an un-intended2321	network loop.2322 2323	Functional default:2324 2325	   - enabled if accept_ra_from_local is enabled2326	     on a specific interface.2327	   - disabled if accept_ra_from_local is disabled2328	     on a specific interface.2329 2330accept_ra_min_hop_limit - INTEGER2331	Minimum hop limit Information in Router Advertisement.2332 2333	Hop limit Information in Router Advertisement less than this2334	variable shall be ignored.2335 2336	Default: 12337 2338accept_ra_min_lft - INTEGER2339	Minimum acceptable lifetime value in Router Advertisement.2340 2341	RA sections with a lifetime less than this value shall be2342	ignored. Zero lifetimes stay unaffected.2343 2344	Default: 02345 2346accept_ra_pinfo - BOOLEAN2347	Learn Prefix Information in Router Advertisement.2348 2349	Functional default:2350 2351		- enabled if accept_ra is enabled.2352		- disabled if accept_ra is disabled.2353 2354ra_honor_pio_life - BOOLEAN2355	Whether to use RFC4862 Section 5.5.3e to determine the valid2356	lifetime of an address matching a prefix sent in a Router2357	Advertisement Prefix Information Option.2358 2359	- If enabled, the PIO valid lifetime will always be honored.2360	- If disabled, RFC4862 section 5.5.3e is used to determine2361	  the valid lifetime of the address.2362 2363	Default: 0 (disabled)2364 2365ra_honor_pio_pflag - BOOLEAN2366	The Prefix Information Option P-flag indicates the network can2367	allocate a unique IPv6 prefix per client using DHCPv6-PD.2368	This sysctl can be enabled when a userspace DHCPv6-PD client2369	is running to cause the P-flag to take effect: i.e. the2370	P-flag suppresses any effects of the A-flag within the same2371	PIO. For a given PIO, P=1 and A=1 is treated as A=0.2372 2373	- If disabled, the P-flag is ignored.2374	- If enabled, the P-flag will disable SLAAC autoconfiguration2375	  for the given Prefix Information Option.2376 2377	Default: 0 (disabled)2378 2379accept_ra_rt_info_min_plen - INTEGER2380	Minimum prefix length of Route Information in RA.2381 2382	Route Information w/ prefix smaller than this variable shall2383	be ignored.2384 2385	Functional default:2386 2387		* 0 if accept_ra_rtr_pref is enabled.2388		* -1 if accept_ra_rtr_pref is disabled.2389 2390accept_ra_rt_info_max_plen - INTEGER2391	Maximum prefix length of Route Information in RA.2392 2393	Route Information w/ prefix larger than this variable shall2394	be ignored.2395 2396	Functional default:2397 2398		* 0 if accept_ra_rtr_pref is enabled.2399		* -1 if accept_ra_rtr_pref is disabled.2400 2401accept_ra_rtr_pref - BOOLEAN2402	Accept Router Preference in RA.2403 2404	Functional default:2405 2406		- enabled if accept_ra is enabled.2407		- disabled if accept_ra is disabled.2408 2409accept_ra_mtu - BOOLEAN2410	Apply the MTU value specified in RA option 5 (RFC4861). If2411	disabled, the MTU specified in the RA will be ignored.2412 2413	Functional default:2414 2415		- enabled if accept_ra is enabled.2416		- disabled if accept_ra is disabled.2417 2418accept_redirects - BOOLEAN2419	Accept Redirects.2420 2421	Functional default:2422 2423		- enabled if local forwarding is disabled.2424		- disabled if local forwarding is enabled.2425 2426accept_source_route - INTEGER2427	Accept source routing (routing extension header).2428 2429	- >= 0: Accept only routing header type 2.2430	- < 0: Do not accept routing header.2431 2432	Default: 02433 2434autoconf - BOOLEAN2435	Autoconfigure addresses using Prefix Information in Router2436	Advertisements.2437 2438	Functional default:2439 2440		- enabled if accept_ra_pinfo is enabled.2441		- disabled if accept_ra_pinfo is disabled.2442 2443dad_transmits - INTEGER2444	The amount of Duplicate Address Detection probes to send.2445 2446	Default: 12447 2448forwarding - INTEGER2449	Configure interface-specific Host/Router behaviour.2450 2451	.. note::2452 2453	   It is recommended to have the same setting on all2454	   interfaces; mixed router/host scenarios are rather uncommon.2455 2456	Possible values are:2457 2458		- 0 Forwarding disabled2459		- 1 Forwarding enabled2460 2461	**FALSE (0)**:2462 2463	By default, Host behaviour is assumed.  This means:2464 2465	1. IsRouter flag is not set in Neighbour Advertisements.2466	2. If accept_ra is TRUE (default), transmit Router2467	   Solicitations.2468	3. If accept_ra is TRUE (default), accept Router2469	   Advertisements (and do autoconfiguration).2470	4. If accept_redirects is TRUE (default), accept Redirects.2471 2472	**TRUE (1)**:2473 2474	If local forwarding is enabled, Router behaviour is assumed.2475	This means exactly the reverse from the above:2476 2477	1. IsRouter flag is set in Neighbour Advertisements.2478	2. Router Solicitations are not sent unless accept_ra is 2.2479	3. Router Advertisements are ignored unless accept_ra is 2.2480	4. Redirects are ignored.2481 2482	Default: 0 (disabled) if global forwarding is disabled (default),2483	otherwise 1 (enabled).2484 2485hop_limit - INTEGER2486	Default Hop Limit to set.2487 2488	Default: 642489 2490mtu - INTEGER2491	Default Maximum Transfer Unit2492 2493	Default: 1280 (IPv6 required minimum)2494 2495ip_nonlocal_bind - BOOLEAN2496	If set, allows processes to bind() to non-local IPv6 addresses,2497	which can be quite useful - but may break some applications.2498 2499	Default: 02500 2501router_probe_interval - INTEGER2502	Minimum interval (in seconds) between Router Probing described2503	in RFC4191.2504 2505	Default: 602506 2507router_solicitation_delay - INTEGER2508	Number of seconds to wait after interface is brought up2509	before sending Router Solicitations.2510 2511	Default: 12512 2513router_solicitation_interval - INTEGER2514	Number of seconds to wait between Router Solicitations.2515 2516	Default: 42517 2518router_solicitations - INTEGER2519	Number of Router Solicitations to send until assuming no2520	routers are present.2521 2522	Default: 32523 2524use_oif_addrs_only - BOOLEAN2525	When enabled, the candidate source addresses for destinations2526	routed via this interface are restricted to the set of addresses2527	configured on this interface (vis. RFC 6724, section 4).2528 2529	Default: false2530 2531use_tempaddr - INTEGER2532	Preference for Privacy Extensions (RFC3041).2533 2534	  * <= 0 : disable Privacy Extensions2535	  * == 1 : enable Privacy Extensions, but prefer public2536	    addresses over temporary addresses.2537	  * >  1 : enable Privacy Extensions and prefer temporary2538	    addresses over public addresses.2539 2540	Default:2541 2542		* 0 (for most devices)2543		* -1 (for point-to-point devices and loopback devices)2544 2545temp_valid_lft - INTEGER2546	valid lifetime (in seconds) for temporary addresses. If less than the2547	minimum required lifetime (typically 5-7 seconds), temporary addresses2548	will not be created.2549 2550	Default: 172800 (2 days)2551 2552temp_prefered_lft - INTEGER2553	Preferred lifetime (in seconds) for temporary addresses. If2554	temp_prefered_lft is less than the minimum required lifetime (typically2555	5-7 seconds), the preferred lifetime is the minimum required. If2556	temp_prefered_lft is greater than temp_valid_lft, the preferred lifetime2557	is temp_valid_lft.2558 2559	Default: 86400 (1 day)2560 2561keep_addr_on_down - INTEGER2562	Keep all IPv6 addresses on an interface down event. If set static2563	global addresses with no expiration time are not flushed.2564 2565	*   >0 : enabled2566	*    0 : system default2567	*   <0 : disabled2568 2569	Default: 0 (addresses are removed)2570 2571max_desync_factor - INTEGER2572	Maximum value for DESYNC_FACTOR, which is a random value2573	that ensures that clients don't synchronize with each2574	other and generate new addresses at exactly the same time.2575	value is in seconds.2576 2577	Default: 6002578 2579regen_min_advance - INTEGER2580	How far in advance (in seconds), at minimum, to create a new temporary2581	address before the current one is deprecated. This value is added to2582	the amount of time that may be required for duplicate address detection2583	to determine when to create a new address. Linux permits setting this2584	value to less than the default of 2 seconds, but a value less than 22585	does not conform to RFC 8981.2586 2587	Default: 22588 2589regen_max_retry - INTEGER2590	Number of attempts before give up attempting to generate2591	valid temporary addresses.2592 2593	Default: 52594 2595max_addresses - INTEGER2596	Maximum number of autoconfigured addresses per interface.  Setting2597	to zero disables the limitation.  It is not recommended to set this2598	value too large (or to zero) because it would be an easy way to2599	crash the kernel by allowing too many addresses to be created.2600 2601	Default: 162602 2603disable_ipv6 - BOOLEAN2604	Disable IPv6 operation.  If accept_dad is set to 2, this value2605	will be dynamically set to TRUE if DAD fails for the link-local2606	address.2607 2608	Default: FALSE (enable IPv6 operation)2609 2610	When this value is changed from 1 to 0 (IPv6 is being enabled),2611	it will dynamically create a link-local address on the given2612	interface and start Duplicate Address Detection, if necessary.2613 2614	When this value is changed from 0 to 1 (IPv6 is being disabled),2615	it will dynamically delete all addresses and routes on the given2616	interface. From now on it will not possible to add addresses/routes2617	to the selected interface.2618 2619accept_dad - INTEGER2620	Whether to accept DAD (Duplicate Address Detection).2621 2622	 == ==============================================================2623	  0  Disable DAD2624	  1  Enable DAD (default)2625	  2  Enable DAD, and disable IPv6 operation if MAC-based duplicate2626	     link-local address has been found.2627	 == ==============================================================2628 2629	DAD operation and mode on a given interface will be selected according2630	to the maximum value of conf/{all,interface}/accept_dad.2631 2632force_tllao - BOOLEAN2633	Enable sending the target link-layer address option even when2634	responding to a unicast neighbor solicitation.2635 2636	Default: FALSE2637 2638	Quoting from RFC 2461, section 4.4, Target link-layer address:2639 2640	"The option MUST be included for multicast solicitations in order to2641	avoid infinite Neighbor Solicitation "recursion" when the peer node2642	does not have a cache entry to return a Neighbor Advertisements2643	message.  When responding to unicast solicitations, the option can be2644	omitted since the sender of the solicitation has the correct link-2645	layer address; otherwise it would not have be able to send the unicast2646	solicitation in the first place. However, including the link-layer2647	address in this case adds little overhead and eliminates a potential2648	race condition where the sender deletes the cached link-layer address2649	prior to receiving a response to a previous solicitation."2650 2651ndisc_notify - BOOLEAN2652	Define mode for notification of address and device changes.2653 2654	* 0 - (default): do nothing2655	* 1 - Generate unsolicited neighbour advertisements when device is brought2656	  up or hardware address changes.2657 2658ndisc_tclass - INTEGER2659	The IPv6 Traffic Class to use by default when sending IPv6 Neighbor2660	Discovery (Router Solicitation, Router Advertisement, Neighbor2661	Solicitation, Neighbor Advertisement, Redirect) messages.2662	These 8 bits can be interpreted as 6 high order bits holding the DSCP2663	value and 2 low order bits representing ECN (which you probably want2664	to leave cleared).2665 2666	* 0 - (default)2667 2668ndisc_evict_nocarrier - BOOLEAN2669	Clears the neighbor discovery table on NOCARRIER events. This option is2670	important for wireless devices where the neighbor discovery cache should2671	not be cleared when roaming between access points on the same network.2672	In most cases this should remain as the default (1).2673 2674	- 1 - (default): Clear neighbor discover cache on NOCARRIER events.2675	- 0 - Do not clear neighbor discovery cache on NOCARRIER events.2676 2677mldv1_unsolicited_report_interval - INTEGER2678	The interval in milliseconds in which the next unsolicited2679	MLDv1 report retransmit will take place.2680 2681	Default: 10000 (10 seconds)2682 2683mldv2_unsolicited_report_interval - INTEGER2684	The interval in milliseconds in which the next unsolicited2685	MLDv2 report retransmit will take place.2686 2687	Default: 1000 (1 second)2688 2689force_mld_version - INTEGER2690	* 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed2691	* 1 - Enforce to use MLD version 12692	* 2 - Enforce to use MLD version 22693 2694suppress_frag_ndisc - INTEGER2695	Control RFC 6980 (Security Implications of IPv6 Fragmentation2696	with IPv6 Neighbor Discovery) behavior:2697 2698	* 1 - (default) discard fragmented neighbor discovery packets2699	* 0 - allow fragmented neighbor discovery packets2700 2701optimistic_dad - BOOLEAN2702	Whether to perform Optimistic Duplicate Address Detection (RFC 4429).2703 2704	* 0: disabled (default)2705	* 1: enabled2706 2707	Optimistic Duplicate Address Detection for the interface will be enabled2708	if at least one of conf/{all,interface}/optimistic_dad is set to 1,2709	it will be disabled otherwise.2710 2711use_optimistic - BOOLEAN2712	If enabled, do not classify optimistic addresses as deprecated during2713	source address selection.  Preferred addresses will still be chosen2714	before optimistic addresses, subject to other ranking in the source2715	address selection algorithm.2716 2717	* 0: disabled (default)2718	* 1: enabled2719 2720	This will be enabled if at least one of2721	conf/{all,interface}/use_optimistic is set to 1, disabled otherwise.2722 2723stable_secret - IPv6 address2724	This IPv6 address will be used as a secret to generate IPv62725	addresses for link-local addresses and autoconfigured2726	ones. All addresses generated after setting this secret will2727	be stable privacy ones by default. This can be changed via the2728	addrgenmode ip-link. conf/default/stable_secret is used as the2729	secret for the namespace, the interface specific ones can2730	overwrite that. Writes to conf/all/stable_secret are refused.2731 2732	It is recommended to generate this secret during installation2733	of a system and keep it stable after that.2734 2735	By default the stable secret is unset.2736 2737addr_gen_mode - INTEGER2738	Defines how link-local and autoconf addresses are generated.2739 2740	=  =================================================================2741	0  generate address based on EUI64 (default)2742	1  do no generate a link-local address, use EUI64 for addresses2743	   generated from autoconf2744	2  generate stable privacy addresses, using the secret from2745	   stable_secret (RFC7217)2746	3  generate stable privacy addresses, using a random secret if unset2747	=  =================================================================2748 2749drop_unicast_in_l2_multicast - BOOLEAN2750	Drop any unicast IPv6 packets that are received in link-layer2751	multicast (or broadcast) frames.2752 2753	By default this is turned off.2754 2755drop_unsolicited_na - BOOLEAN2756	Drop all unsolicited neighbor advertisements, for example if there's2757	a known good NA proxy on the network and such frames need not be used2758	(or in the case of 802.11, must not be used to prevent attacks.)2759 2760	By default this is turned off.2761 2762accept_untracked_na - INTEGER2763	Define behavior for accepting neighbor advertisements from devices that2764	are absent in the neighbor cache:2765 2766	- 0 - (default) Do not accept unsolicited and untracked neighbor2767	  advertisements.2768 2769	- 1 - Add a new neighbor cache entry in STALE state for routers on2770	  receiving a neighbor advertisement (either solicited or unsolicited)2771	  with target link-layer address option specified if no neighbor entry2772	  is already present for the advertised IPv6 address. Without this knob,2773	  NAs received for untracked addresses (absent in neighbor cache) are2774	  silently ignored.2775 2776	  This is as per router-side behavior documented in RFC9131.2777 2778	  This has lower precedence than drop_unsolicited_na.2779 2780	  This will optimize the return path for the initial off-link2781	  communication that is initiated by a directly connected host, by2782	  ensuring that the first-hop router which turns on this setting doesn't2783	  have to buffer the initial return packets to do neighbor-solicitation.2784	  The prerequisite is that the host is configured to send unsolicited2785	  neighbor advertisements on interface bringup. This setting should be2786	  used in conjunction with the ndisc_notify setting on the host to2787	  satisfy this prerequisite.2788 2789	- 2 - Extend option (1) to add a new neighbor cache entry only if the2790	  source IP address is in the same subnet as an address configured on2791	  the interface that received the neighbor advertisement.2792 2793enhanced_dad - BOOLEAN2794	Include a nonce option in the IPv6 neighbor solicitation messages used for2795	duplicate address detection per RFC7527. A received DAD NS will only signal2796	a duplicate address if the nonce is different. This avoids any false2797	detection of duplicates due to loopback of the NS messages that we send.2798	The nonce option will be sent on an interface unless both of2799	conf/{all,interface}/enhanced_dad are set to FALSE.2800 2801	Default: TRUE2802 2803``icmp/*``:2804===========2805 2806ratelimit - INTEGER2807	Limit the maximal rates for sending ICMPv6 messages.2808 2809	0 to disable any limiting,2810	otherwise the minimal space between responses in milliseconds.2811 2812	Default: 10002813 2814ratemask - list of comma separated ranges2815	For ICMPv6 message types matching the ranges in the ratemask, limit2816	the sending of the message according to ratelimit parameter.2817 2818	The format used for both input and output is a comma separated2819	list of ranges (e.g. "0-127,129" for ICMPv6 message type 0 to 127 and2820	129). Writing to the file will clear all previous ranges of ICMPv62821	message types and update the current list with the input.2822 2823	Refer to: https://www.iana.org/assignments/icmpv6-parameters/icmpv6-parameters.xhtml2824	for numerical values of ICMPv6 message types, e.g. echo request is 1282825	and echo reply is 129.2826 2827	Default: 0-1,3-127 (rate limit ICMPv6 errors except Packet Too Big)2828 2829echo_ignore_all - BOOLEAN2830	If set non-zero, then the kernel will ignore all ICMP ECHO2831	requests sent to it over the IPv6 protocol.2832 2833	Default: 02834 2835echo_ignore_multicast - BOOLEAN2836	If set non-zero, then the kernel will ignore all ICMP ECHO2837	requests sent to it over the IPv6 protocol via multicast.2838 2839	Default: 02840 2841echo_ignore_anycast - BOOLEAN2842	If set non-zero, then the kernel will ignore all ICMP ECHO2843	requests sent to it over the IPv6 protocol destined to anycast address.2844 2845	Default: 02846 2847error_anycast_as_unicast - BOOLEAN2848	If set to 1, then the kernel will respond with ICMP Errors2849	resulting from requests sent to it over the IPv6 protocol destined2850	to anycast address essentially treating anycast as unicast.2851 2852	Default: 02853 2854xfrm6_gc_thresh - INTEGER2855	(Obsolete since linux-4.14)2856	The threshold at which we will start garbage collecting for IPv62857	destination cache entries.  At twice this value the system will2858	refuse new allocations.2859 2860 2861IPv6 Update by:2862Pekka Savola <pekkas@netcore.fi>2863YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>2864 2865 2866/proc/sys/net/bridge/* Variables:2867=================================2868 2869bridge-nf-call-arptables - BOOLEAN2870	- 1 : pass bridged ARP traffic to arptables' FORWARD chain.2871	- 0 : disable this.2872 2873	Default: 12874 2875bridge-nf-call-iptables - BOOLEAN2876	- 1 : pass bridged IPv4 traffic to iptables' chains.2877	- 0 : disable this.2878 2879	Default: 12880 2881bridge-nf-call-ip6tables - BOOLEAN2882	- 1 : pass bridged IPv6 traffic to ip6tables' chains.2883	- 0 : disable this.2884 2885	Default: 12886 2887bridge-nf-filter-vlan-tagged - BOOLEAN2888	- 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.2889	- 0 : disable this.2890 2891	Default: 02892 2893bridge-nf-filter-pppoe-tagged - BOOLEAN2894	- 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.2895	- 0 : disable this.2896 2897	Default: 02898 2899bridge-nf-pass-vlan-input-dev - BOOLEAN2900	- 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan2901	  interface on the bridge and set the netfilter input device to the2902	  vlan. This allows use of e.g. "iptables -i br0.1" and makes the2903	  REDIRECT target work with vlan-on-top-of-bridge interfaces.  When no2904	  matching vlan interface is found, or this switch is off, the input2905	  device is set to the bridge interface.2906 2907	- 0: disable bridge netfilter vlan interface lookup.2908 2909	Default: 02910 2911``proc/sys/net/sctp/*`` Variables:2912==================================2913 2914addip_enable - BOOLEAN2915	Enable or disable extension of  Dynamic Address Reconfiguration2916	(ADD-IP) functionality specified in RFC5061.  This extension provides2917	the ability to dynamically add and remove new addresses for the SCTP2918	associations.2919 2920	1: Enable extension.2921 2922	0: Disable extension.2923 2924	Default: 02925 2926pf_enable - INTEGER2927	Enable or disable pf (pf is short for potentially failed) state. A value2928	of pf_retrans > path_max_retrans also disables pf state. That is, one of2929	both pf_enable and pf_retrans > path_max_retrans can disable pf state.2930	Since pf_retrans and path_max_retrans can be changed by userspace2931	application, sometimes user expects to disable pf state by the value of2932	pf_retrans > path_max_retrans, but occasionally the value of pf_retrans2933	or path_max_retrans is changed by the user application, this pf state is2934	enabled. As such, it is necessary to add this to dynamically enable2935	and disable pf state. See:2936	https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for2937	details.2938 2939	1: Enable pf.2940 2941	0: Disable pf.2942 2943	Default: 12944 2945pf_expose - INTEGER2946	Unset or enable/disable pf (pf is short for potentially failed) state2947	exposure.  Applications can control the exposure of the PF path state2948	in the SCTP_PEER_ADDR_CHANGE event and the SCTP_GET_PEER_ADDR_INFO2949	sockopt.   When it's unset, no SCTP_PEER_ADDR_CHANGE event with2950	SCTP_ADDR_PF state will be sent and a SCTP_PF-state transport info2951	can be got via SCTP_GET_PEER_ADDR_INFO sockopt;  When it's enabled,2952	a SCTP_PEER_ADDR_CHANGE event will be sent for a transport becoming2953	SCTP_PF state and a SCTP_PF-state transport info can be got via2954	SCTP_GET_PEER_ADDR_INFO sockopt;  When it's disabled, no2955	SCTP_PEER_ADDR_CHANGE event will be sent and it returns -EACCES when2956	trying to get a SCTP_PF-state transport info via SCTP_GET_PEER_ADDR_INFO2957	sockopt.2958 2959	0: Unset pf state exposure, Compatible with old applications.2960 2961	1: Disable pf state exposure.2962 2963	2: Enable pf state exposure.2964 2965	Default: 02966 2967addip_noauth_enable - BOOLEAN2968	Dynamic Address Reconfiguration (ADD-IP) requires the use of2969	authentication to protect the operations of adding or removing new2970	addresses.  This requirement is mandated so that unauthorized hosts2971	would not be able to hijack associations.  However, older2972	implementations may not have implemented this requirement while2973	allowing the ADD-IP extension.  For reasons of interoperability,2974	we provide this variable to control the enforcement of the2975	authentication requirement.2976 2977	== ===============================================================2978	1  Allow ADD-IP extension to be used without authentication.  This2979	   should only be set in a closed environment for interoperability2980	   with older implementations.2981 2982	0  Enforce the authentication requirement2983	== ===============================================================2984 2985	Default: 02986 2987auth_enable - BOOLEAN2988	Enable or disable Authenticated Chunks extension.  This extension2989	provides the ability to send and receive authenticated chunks and is2990	required for secure operation of Dynamic Address Reconfiguration2991	(ADD-IP) extension.2992 2993	- 1: Enable this extension.2994	- 0: Disable this extension.2995 2996	Default: 02997 2998prsctp_enable - BOOLEAN2999	Enable or disable the Partial Reliability extension (RFC3758) which3000	is used to notify peers that a given DATA should no longer be expected.3001 3002	- 1: Enable extension3003	- 0: Disable3004 3005	Default: 13006 3007max_burst - INTEGER3008	The limit of the number of new packets that can be initially sent.  It3009	controls how bursty the generated traffic can be.3010 3011	Default: 43012 3013association_max_retrans - INTEGER3014	Set the maximum number for retransmissions that an association can3015	attempt deciding that the remote end is unreachable.  If this value3016	is exceeded, the association is terminated.3017 3018	Default: 103019 3020max_init_retransmits - INTEGER3021	The maximum number of retransmissions of INIT and COOKIE-ECHO chunks3022	that an association will attempt before declaring the destination3023	unreachable and terminating.3024 3025	Default: 83026 3027path_max_retrans - INTEGER3028	The maximum number of retransmissions that will be attempted on a given3029	path.  Once this threshold is exceeded, the path is considered3030	unreachable, and new traffic will use a different path when the3031	association is multihomed.3032 3033	Default: 53034 3035pf_retrans - INTEGER3036	The number of retransmissions that will be attempted on a given path3037	before traffic is redirected to an alternate transport (should one3038	exist).  Note this is distinct from path_max_retrans, as a path that3039	passes the pf_retrans threshold can still be used.  Its only3040	deprioritized when a transmission path is selected by the stack.  This3041	setting is primarily used to enable fast failover mechanisms without3042	having to reduce path_max_retrans to a very low value.  See:3043	http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt3044	for details.  Note also that a value of pf_retrans > path_max_retrans3045	disables this feature. Since both pf_retrans and path_max_retrans can3046	be changed by userspace application, a variable pf_enable is used to3047	disable pf state.3048 3049	Default: 03050 3051ps_retrans - INTEGER3052	Primary.Switchover.Max.Retrans (PSMR), it's a tunable parameter coming3053	from section-5 "Primary Path Switchover" in rfc7829.  The primary path3054	will be changed to another active path when the path error counter on3055	the old primary path exceeds PSMR, so that "the SCTP sender is allowed3056	to continue data transmission on a new working path even when the old3057	primary destination address becomes active again".   Note this feature3058	is disabled by initializing 'ps_retrans' per netns as 0xffff by default,3059	and its value can't be less than 'pf_retrans' when changing by sysctl.3060 3061	Default: 0xffff3062 3063rto_initial - INTEGER3064	The initial round trip timeout value in milliseconds that will be used3065	in calculating round trip times.  This is the initial time interval3066	for retransmissions.3067 3068	Default: 30003069 3070rto_max - INTEGER3071	The maximum value (in milliseconds) of the round trip timeout.  This3072	is the largest time interval that can elapse between retransmissions.3073 3074	Default: 600003075 3076rto_min - INTEGER3077	The minimum value (in milliseconds) of the round trip timeout.  This3078	is the smallest time interval the can elapse between retransmissions.3079 3080	Default: 10003081 3082hb_interval - INTEGER3083	The interval (in milliseconds) between HEARTBEAT chunks.  These chunks3084	are sent at the specified interval on idle paths to probe the state of3085	a given path between 2 associations.3086 3087	Default: 300003088 3089sack_timeout - INTEGER3090	The amount of time (in milliseconds) that the implementation will wait3091	to send a SACK.3092 3093	Default: 2003094 3095valid_cookie_life - INTEGER3096	The default lifetime of the SCTP cookie (in milliseconds).  The cookie3097	is used during association establishment.3098 3099	Default: 600003100 3101cookie_preserve_enable - BOOLEAN3102	Enable or disable the ability to extend the lifetime of the SCTP cookie3103	that is used during the establishment phase of SCTP association3104 3105	- 1: Enable cookie lifetime extension.3106	- 0: Disable3107 3108	Default: 13109 3110cookie_hmac_alg - STRING3111	Select the hmac algorithm used when generating the cookie value sent by3112	a listening sctp socket to a connecting client in the INIT-ACK chunk.3113	Valid values are:3114 3115	* md53116	* sha13117	* none3118 3119	Ability to assign md5 or sha1 as the selected alg is predicated on the3120	configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and3121	CONFIG_CRYPTO_SHA1).3122 3123	Default: Dependent on configuration.  MD5 if available, else SHA1 if3124	available, else none.3125 3126rcvbuf_policy - INTEGER3127	Determines if the receive buffer is attributed to the socket or to3128	association.   SCTP supports the capability to create multiple3129	associations on a single socket.  When using this capability, it is3130	possible that a single stalled association that's buffering a lot3131	of data may block other associations from delivering their data by3132	consuming all of the receive buffer space.  To work around this,3133	the rcvbuf_policy could be set to attribute the receiver buffer space3134	to each association instead of the socket.  This prevents the described3135	blocking.3136 3137	- 1: rcvbuf space is per association3138	- 0: rcvbuf space is per socket3139 3140	Default: 03141 3142sndbuf_policy - INTEGER3143	Similar to rcvbuf_policy above, this applies to send buffer space.3144 3145	- 1: Send buffer is tracked per association3146	- 0: Send buffer is tracked per socket.3147 3148	Default: 03149 3150sctp_mem - vector of 3 INTEGERs: min, pressure, max3151	Number of pages allowed for queueing by all SCTP sockets.3152 3153	min: Below this number of pages SCTP is not bothered about its3154	memory appetite. When amount of memory allocated by SCTP exceeds3155	this number, SCTP starts to moderate memory usage.3156 3157	pressure: This value was introduced to follow format of tcp_mem.3158 3159	max: Number of pages allowed for queueing by all SCTP sockets.3160 3161	Default is calculated at boot time from amount of available memory.3162 3163sctp_rmem - vector of 3 INTEGERs: min, default, max3164	Only the first value ("min") is used, "default" and "max" are3165	ignored.3166 3167	min: Minimal size of receive buffer used by SCTP socket.3168	It is guaranteed to each SCTP socket (but not association) even3169	under moderate memory pressure.3170 3171	Default: 4K3172 3173sctp_wmem  - vector of 3 INTEGERs: min, default, max3174	Only the first value ("min") is used, "default" and "max" are3175	ignored.3176 3177	min: Minimum size of send buffer that can be used by SCTP sockets.3178	It is guaranteed to each SCTP socket (but not association) even3179	under moderate memory pressure.3180 3181	Default: 4K3182 3183addr_scope_policy - INTEGER3184	Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-003185 3186	- 0   - Disable IPv4 address scoping3187	- 1   - Enable IPv4 address scoping3188	- 2   - Follow draft but allow IPv4 private addresses3189	- 3   - Follow draft but allow IPv4 link local addresses3190 3191	Default: 13192 3193udp_port - INTEGER3194	The listening port for the local UDP tunneling sock. Normally it's3195	using the IANA-assigned UDP port number 9899 (sctp-tunneling).3196 3197	This UDP sock is used for processing the incoming UDP-encapsulated3198	SCTP packets (from RFC6951), and shared by all applications in the3199	same net namespace. This UDP sock will be closed when the value is3200	set to 0.3201 3202	The value will also be used to set the src port of the UDP header3203	for the outgoing UDP-encapsulated SCTP packets. For the dest port,3204	please refer to 'encap_port' below.3205 3206	Default: 03207 3208encap_port - INTEGER3209	The default remote UDP encapsulation port.3210 3211	This value is used to set the dest port of the UDP header for the3212	outgoing UDP-encapsulated SCTP packets by default. Users can also3213	change the value for each sock/asoc/transport by using setsockopt.3214	For further information, please refer to RFC6951.3215 3216	Note that when connecting to a remote server, the client should set3217	this to the port that the UDP tunneling sock on the peer server is3218	listening to and the local UDP tunneling sock on the client also3219	must be started. On the server, it would get the encap_port from3220	the incoming packet's source port.3221 3222	Default: 03223 3224plpmtud_probe_interval - INTEGER3225        The time interval (in milliseconds) for the PLPMTUD probe timer,3226        which is configured to expire after this period to receive an3227        acknowledgment to a probe packet. This is also the time interval3228        between the probes for the current pmtu when the probe search3229        is done.3230 3231        PLPMTUD will be disabled when 0 is set, and other values for it3232        must be >= 5000.3233 3234	Default: 03235 3236reconf_enable - BOOLEAN3237        Enable or disable extension of Stream Reconfiguration functionality3238        specified in RFC6525. This extension provides the ability to "reset"3239        a stream, and it includes the Parameters of "Outgoing/Incoming SSN3240        Reset", "SSN/TSN Reset" and "Add Outgoing/Incoming Streams".3241 3242	- 1: Enable extension.3243	- 0: Disable extension.3244 3245	Default: 03246 3247intl_enable - BOOLEAN3248        Enable or disable extension of User Message Interleaving functionality3249        specified in RFC8260. This extension allows the interleaving of user3250        messages sent on different streams. With this feature enabled, I-DATA3251        chunk will replace DATA chunk to carry user messages if also supported3252        by the peer. Note that to use this feature, one needs to set this option3253        to 1 and also needs to set socket options SCTP_FRAGMENT_INTERLEAVE to 23254        and SCTP_INTERLEAVING_SUPPORTED to 1.3255 3256	- 1: Enable extension.3257	- 0: Disable extension.3258 3259	Default: 03260 3261ecn_enable - BOOLEAN3262        Control use of Explicit Congestion Notification (ECN) by SCTP.3263        Like in TCP, ECN is used only when both ends of the SCTP connection3264        indicate support for it. This feature is useful in avoiding losses3265        due to congestion by allowing supporting routers to signal congestion3266        before having to drop packets.3267 3268        1: Enable ecn.3269        0: Disable ecn.3270 3271        Default: 13272 3273l3mdev_accept - BOOLEAN3274	Enabling this option allows a "global" bound socket to work3275	across L3 master domains (e.g., VRFs) with packets capable of3276	being received regardless of the L3 domain in which they3277	originated. Only valid when the kernel was compiled with3278	CONFIG_NET_L3_MASTER_DEV.3279 3280	Default: 1 (enabled)3281 3282 3283``/proc/sys/net/core/*``3284========================3285 3286	Please see: Documentation/admin-guide/sysctl/net.rst for descriptions of these entries.3287 3288 3289``/proc/sys/net/unix/*``3290========================3291 3292max_dgram_qlen - INTEGER3293	The maximum length of dgram socket receive queue3294 3295	Default: 103296 3297