1.. SPDX-License-Identifier: GPL-2.0 2 3========= 4IP Sysctl 5========= 6 7/proc/sys/net/ipv4/* Variables 8============================== 9 10ip_forward - BOOLEAN 11 Forward Packets between interfaces. 12 13 This variable is special, its change resets all configuration 14 parameters to their default state (RFC1122 for hosts, RFC1812 15 for routers) 16 17 Possible values: 18 19 - 0 (disabled) 20 - 1 (enabled) 21 22 Default: 0 (disabled) 23 24ip_default_ttl - INTEGER 25 Default value of TTL field (Time To Live) for outgoing (but not 26 forwarded) IP packets. Should be between 1 and 255 inclusive. 27 Default: 64 (as recommended by RFC1700) 28 29ip_no_pmtu_disc - INTEGER 30 Disable Path MTU Discovery. If enabled in mode 1 and a 31 fragmentation-required ICMP is received, the PMTU to this 32 destination will be set to the smallest of the old MTU to 33 this destination and min_pmtu (see below). You will need 34 to raise min_pmtu to the smallest interface MTU on your system 35 manually if you want to avoid locally generated fragments. 36 37 In mode 2 incoming Path MTU Discovery messages will be 38 discarded. Outgoing frames are handled the same as in mode 1, 39 implicitly setting IP_PMTUDISC_DONT on every created socket. 40 41 Mode 3 is a hardened pmtu discover mode. The kernel will only 42 accept fragmentation-needed errors if the underlying protocol 43 can verify them besides a plain socket lookup. Current 44 protocols for which pmtu events will be honored are TCP and 45 SCTP as they verify e.g. the sequence number or the 46 association. This mode should not be enabled globally but is 47 only intended to secure e.g. name servers in namespaces where 48 TCP path mtu must still work but path MTU information of other 49 protocols should be discarded. If enabled globally this mode 50 could break other protocols. 51 52 Possible values: 0-3 53 54 Default: FALSE 55 56min_pmtu - INTEGER 57 default 552 - minimum Path MTU. Unless this is changed manually, 58 each cached pmtu will never be lower than this setting. 59 60ip_forward_use_pmtu - BOOLEAN 61 By default we don't trust protocol path MTUs while forwarding 62 because they could be easily forged and can lead to unwanted 63 fragmentation by the router. 64 You only need to enable this if you have user-space software 65 which tries to discover path mtus by itself and depends on the 66 kernel honoring this information. This is normally not the 67 case. 68 69 Possible values: 70 71 - 0 (disabled) 72 - 1 (enabled) 73 74 Default: 0 (disabled) 75 76fwmark_reflect - BOOLEAN 77 Controls the fwmark of kernel-generated IPv4 reply packets that are not 78 associated with a socket for example, TCP RSTs or ICMP echo replies). 79 If disabled, these packets have a fwmark of zero. If enabled, they have the 80 fwmark of the packet they are replying to. 81 82 Possible values: 83 84 - 0 (disabled) 85 - 1 (enabled) 86 87 Default: 0 (disabled) 88 89fib_multipath_use_neigh - BOOLEAN 90 Use status of existing neighbor entry when determining nexthop for 91 multipath routes. If disabled, neighbor information is not used and 92 packets could be directed to a failed nexthop. Only valid for kernels 93 built with CONFIG_IP_ROUTE_MULTIPATH enabled. 94 95 Possible values: 96 97 - 0 (disabled) 98 - 1 (enabled) 99 100 Default: 0 (disabled) 101 102fib_multipath_hash_policy - INTEGER 103 Controls which hash policy to use for multipath routes. Only valid 104 for kernels built with CONFIG_IP_ROUTE_MULTIPATH enabled. 105 106 Default: 0 (Layer 3) 107 108 Possible values: 109 110 - 0 - Layer 3 111 - 1 - Layer 4 112 - 2 - Layer 3 or inner Layer 3 if present 113 - 3 - Custom multipath hash. Fields used for multipath hash calculation 114 are determined by fib_multipath_hash_fields sysctl 115 116fib_multipath_hash_fields - UNSIGNED INTEGER 117 When fib_multipath_hash_policy is set to 3 (custom multipath hash), the 118 fields used for multipath hash calculation are determined by this 119 sysctl. 120 121 This value is a bitmask which enables various fields for multipath hash 122 calculation. 123 124 Possible fields are: 125 126 ====== ============================ 127 0x0001 Source IP address 128 0x0002 Destination IP address 129 0x0004 IP protocol 130 0x0008 Unused (Flow Label) 131 0x0010 Source port 132 0x0020 Destination port 133 0x0040 Inner source IP address 134 0x0080 Inner destination IP address 135 0x0100 Inner IP protocol 136 0x0200 Inner Flow Label 137 0x0400 Inner source port 138 0x0800 Inner destination port 139 ====== ============================ 140 141 Default: 0x0007 (source IP, destination IP and IP protocol) 142 143fib_multipath_hash_seed - UNSIGNED INTEGER 144 The seed value used when calculating hash for multipath routes. Applies 145 to both IPv4 and IPv6 datapath. Only present for kernels built with 146 CONFIG_IP_ROUTE_MULTIPATH enabled. 147 148 When set to 0, the seed value used for multipath routing defaults to an 149 internal random-generated one. 150 151 The actual hashing algorithm is not specified -- there is no guarantee 152 that a next hop distribution effected by a given seed will keep stable 153 across kernel versions. 154 155 Default: 0 (random) 156 157fib_sync_mem - UNSIGNED INTEGER 158 Amount of dirty memory from fib entries that can be backlogged before 159 synchronize_rcu is forced. 160 161 Default: 512kB Minimum: 64kB Maximum: 64MB 162 163ip_forward_update_priority - INTEGER 164 Whether to update SKB priority from "TOS" field in IPv4 header after it 165 is forwarded. The new SKB priority is mapped from TOS field value 166 according to an rt_tos2priority table (see e.g. man tc-prio). 167 168 Default: 1 (Update priority.) 169 170 Possible values: 171 172 - 0 - Do not update priority. 173 - 1 - Update priority. 174 175route/max_size - INTEGER 176 Maximum number of routes allowed in the kernel. Increase 177 this when using large numbers of interfaces and/or routes. 178 179 From linux kernel 3.6 onwards, this is deprecated for ipv4 180 as route cache is no longer used. 181 182 From linux kernel 6.3 onwards, this is deprecated for ipv6 183 as garbage collection manages cached route entries. 184 185neigh/default/gc_thresh1 - INTEGER 186 Minimum number of entries to keep. Garbage collector will not 187 purge entries if there are fewer than this number. 188 189 Default: 128 190 191neigh/default/gc_thresh2 - INTEGER 192 Threshold when garbage collector becomes more aggressive about 193 purging entries. Entries older than 5 seconds will be cleared 194 when over this number. 195 196 Default: 512 197 198neigh/default/gc_thresh3 - INTEGER 199 Maximum number of non-PERMANENT neighbor entries allowed. Increase 200 this when using large numbers of interfaces and when communicating 201 with large numbers of directly-connected peers. 202 203 Default: 1024 204 205neigh/default/unres_qlen_bytes - INTEGER 206 The maximum number of bytes which may be used by packets 207 queued for each unresolved address by other network layers. 208 (added in linux 3.3) 209 210 Setting negative value is meaningless and will return error. 211 212 Default: SK_WMEM_DEFAULT, (same as net.core.wmem_default). 213 214 Exact value depends on architecture and kernel options, 215 but should be enough to allow queuing 256 packets 216 of medium size. 217 218neigh/default/unres_qlen - INTEGER 219 The maximum number of packets which may be queued for each 220 unresolved address by other network layers. 221 222 (deprecated in linux 3.3) : use unres_qlen_bytes instead. 223 224 Prior to linux 3.3, the default value is 3 which may cause 225 unexpected packet loss. The current default value is calculated 226 according to default value of unres_qlen_bytes and true size of 227 packet. 228 229 Default: 101 230 231neigh/default/interval_probe_time_ms - INTEGER 232 The probe interval for neighbor entries with NTF_MANAGED flag, 233 the min value is 1. 234 235 Default: 5000 236 237mtu_expires - INTEGER 238 Time, in seconds, that cached PMTU information is kept. 239 240min_adv_mss - INTEGER 241 The advertised MSS depends on the first hop route MTU, but will 242 never be lower than this setting. 243 244fib_notify_on_flag_change - INTEGER 245 Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/ 246 RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed. 247 248 After installing a route to the kernel, user space receives an 249 acknowledgment, which means the route was installed in the kernel, 250 but not necessarily in hardware. 251 It is also possible for a route already installed in hardware to change 252 its action and therefore its flags. For example, a host route that is 253 trapping packets can be "promoted" to perform decapsulation following 254 the installation of an IPinIP/VXLAN tunnel. 255 The notifications will indicate to user-space the state of the route. 256 257 Default: 0 (Do not emit notifications.) 258 259 Possible values: 260 261 - 0 - Do not emit notifications. 262 - 1 - Emit notifications. 263 - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change. 264 265IP Fragmentation: 266 267ipfrag_high_thresh - LONG INTEGER 268 Maximum memory used to reassemble IP fragments. 269 270ipfrag_low_thresh - LONG INTEGER 271 (Obsolete since linux-4.17) 272 Maximum memory used to reassemble IP fragments before the kernel 273 begins to remove incomplete fragment queues to free up resources. 274 The kernel still accepts new fragments for defragmentation. 275 276ipfrag_time - INTEGER 277 Time in seconds to keep an IP fragment in memory. 278 279ipfrag_max_dist - INTEGER 280 ipfrag_max_dist is a non-negative integer value which defines the 281 maximum "disorder" which is allowed among fragments which share a 282 common IP source address. Note that reordering of packets is 283 not unusual, but if a large number of fragments arrive from a source 284 IP address while a particular fragment queue remains incomplete, it 285 probably indicates that one or more fragments belonging to that queue 286 have been lost. When ipfrag_max_dist is positive, an additional check 287 is done on fragments before they are added to a reassembly queue - if 288 ipfrag_max_dist (or more) fragments have arrived from a particular IP 289 address between additions to any IP fragment queue using that source 290 address, it's presumed that one or more fragments in the queue are 291 lost. The existing fragment queue will be dropped, and a new one 292 started. An ipfrag_max_dist value of zero disables this check. 293 294 Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can 295 result in unnecessarily dropping fragment queues when normal 296 reordering of packets occurs, which could lead to poor application 297 performance. Using a very large value, e.g. 50000, increases the 298 likelihood of incorrectly reassembling IP fragments that originate 299 from different IP datagrams, which could result in data corruption. 300 Default: 64 301 302bc_forwarding - INTEGER 303 bc_forwarding enables the feature described in rfc1812#section-5.3.5.2 304 and rfc2644. It allows the router to forward directed broadcast. 305 To enable this feature, the 'all' entry and the input interface entry 306 should be set to 1. 307 Default: 0 308 309INET peer storage 310================= 311 312inet_peer_threshold - INTEGER 313 The approximate size of the storage. Starting from this threshold 314 entries will be thrown aggressively. This threshold also determines 315 entries' time-to-live and time intervals between garbage collection 316 passes. More entries, less time-to-live, less GC interval. 317 318inet_peer_minttl - INTEGER 319 Minimum time-to-live of entries. Should be enough to cover fragment 320 time-to-live on the reassembling side. This minimum time-to-live is 321 guaranteed if the pool size is less than inet_peer_threshold. 322 Measured in seconds. 323 324inet_peer_maxttl - INTEGER 325 Maximum time-to-live of entries. Unused entries will expire after 326 this period of time if there is no memory pressure on the pool (i.e. 327 when the number of entries in the pool is very small). 328 Measured in seconds. 329 330TCP variables 331============= 332 333somaxconn - INTEGER 334 Limit of socket listen() backlog, known in userspace as SOMAXCONN. 335 Defaults to 4096. (Was 128 before linux-5.4) 336 See also tcp_max_syn_backlog for additional tuning for TCP sockets. 337 338tcp_abort_on_overflow - BOOLEAN 339 If listening service is too slow to accept new connections, 340 reset them. Default state is FALSE. It means that if overflow 341 occurred due to a burst, connection will recover. Enable this 342 option _only_ if you are really sure that listening daemon 343 cannot be tuned to accept connections faster. Enabling this 344 option can harm clients of your server. 345 346tcp_adv_win_scale - INTEGER 347 Obsolete since linux-6.6 348 Count buffering overhead as bytes/2^tcp_adv_win_scale 349 (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale), 350 if it is <= 0. 351 352 Possible values are [-31, 31], inclusive. 353 354 Default: 1 355 356tcp_allowed_congestion_control - STRING 357 Show/set the congestion control choices available to non-privileged 358 processes. The list is a subset of those listed in 359 tcp_available_congestion_control. 360 361 Default is "reno" and the default setting (tcp_congestion_control). 362 363tcp_app_win - INTEGER 364 Reserve max(window/2^tcp_app_win, mss) of window for application 365 buffer. Value 0 is special, it means that nothing is reserved. 366 367 Possible values are [0, 31], inclusive. 368 369 Default: 31 370 371tcp_autocorking - BOOLEAN 372 Enable TCP auto corking : 373 When applications do consecutive small write()/sendmsg() system calls, 374 we try to coalesce these small writes as much as possible, to lower 375 total amount of sent packets. This is done if at least one prior 376 packet for the flow is waiting in Qdisc queues or device transmit 377 queue. Applications can still use TCP_CORK for optimal behavior 378 when they know how/when to uncork their sockets. 379 380 Possible values: 381 382 - 0 (disabled) 383 - 1 (enabled) 384 385 Default: 1 (enabled) 386 387tcp_available_congestion_control - STRING 388 Shows the available congestion control choices that are registered. 389 More congestion control algorithms may be available as modules, 390 but not loaded. 391 392tcp_base_mss - INTEGER 393 The initial value of search_low to be used by the packetization layer 394 Path MTU discovery (MTU probing). If MTU probing is enabled, 395 this is the initial MSS used by the connection. 396 397tcp_mtu_probe_floor - INTEGER 398 If MTU probing is enabled this caps the minimum MSS used for search_low 399 for the connection. 400 401 Default : 48 402 403tcp_min_snd_mss - INTEGER 404 TCP SYN and SYNACK messages usually advertise an ADVMSS option, 405 as described in RFC 1122 and RFC 6691. 406 407 If this ADVMSS option is smaller than tcp_min_snd_mss, 408 it is silently capped to tcp_min_snd_mss. 409 410 Default : 48 (at least 8 bytes of payload per segment) 411 412tcp_congestion_control - STRING 413 Set the congestion control algorithm to be used for new 414 connections. The algorithm "reno" is always available, but 415 additional choices may be available based on kernel configuration. 416 Default is set as part of kernel configuration. 417 For passive connections, the listener congestion control choice 418 is inherited. 419 420 [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ] 421 422tcp_dsack - BOOLEAN 423 Allows TCP to send "duplicate" SACKs. 424 425 Possible values: 426 427 - 0 (disabled) 428 - 1 (enabled) 429 430 Default: 1 (enabled) 431 432tcp_early_retrans - INTEGER 433 Tail loss probe (TLP) converts RTOs occurring due to tail 434 losses into fast recovery (RFC8985). Note that 435 TLP requires RACK to function properly (see tcp_recovery below) 436 437 Possible values: 438 439 - 0 disables TLP 440 - 3 or 4 enables TLP 441 442 Default: 3 443 444tcp_ecn - INTEGER 445 Control use of Explicit Congestion Notification (ECN) by TCP. 446 ECN is used only when both ends of the TCP connection indicate support 447 for it. This feature is useful in avoiding losses due to congestion by 448 allowing supporting routers to signal congestion before having to drop 449 packets. A host that supports ECN both sends ECN at the IP layer and 450 feeds back ECN at the TCP layer. The highest variant of ECN feedback 451 that both peers support is chosen by the ECN negotiation (Accurate ECN, 452 ECN, or no ECN). 453 454 The highest negotiated variant for incoming connection requests 455 and the highest variant requested by outgoing connection 456 attempts: 457 458 ===== ==================== ==================== 459 Value Incoming connections Outgoing connections 460 ===== ==================== ==================== 461 0 No ECN No ECN 462 1 ECN ECN 463 2 ECN No ECN 464 3 AccECN AccECN 465 4 AccECN ECN 466 5 AccECN No ECN 467 ===== ==================== ==================== 468 469 Default: 2 470 471tcp_ecn_option - INTEGER 472 Control Accurate ECN (AccECN) option sending when AccECN has been 473 successfully negotiated during handshake. Send logic inhibits 474 sending AccECN options regarless of this setting when no AccECN 475 option has been seen for the reverse direction. 476 477 Possible values are: 478 479 = ============================================================ 480 0 Never send AccECN option. This also disables sending AccECN 481 option in SYN/ACK during handshake. 482 1 Send AccECN option sparingly according to the minimum option 483 rules outlined in draft-ietf-tcpm-accurate-ecn. 484 2 Send AccECN option on every packet whenever it fits into TCP 485 option space. 486 = ============================================================ 487 488 Default: 2 489 490tcp_ecn_option_beacon - INTEGER 491 Control Accurate ECN (AccECN) option sending frequency per RTT and it 492 takes effect only when tcp_ecn_option is set to 2. 493 494 Default: 3 (AccECN will be send at least 3 times per RTT) 495 496tcp_ecn_fallback - BOOLEAN 497 If the kernel detects that ECN connection misbehaves, enable fall 498 back to non-ECN. Currently, this knob implements the fallback 499 from RFC3168, section 6.1.1.1., but we reserve that in future, 500 additional detection mechanisms could be implemented under this 501 knob. The value is not used, if tcp_ecn or per route (or congestion 502 control) ECN settings are disabled. 503 504 Possible values: 505 506 - 0 (disabled) 507 - 1 (enabled) 508 509 Default: 1 (enabled) 510 511tcp_fack - BOOLEAN 512 This is a legacy option, it has no effect anymore. 513 514tcp_fin_timeout - INTEGER 515 The length of time an orphaned (no longer referenced by any 516 application) connection will remain in the FIN_WAIT_2 state 517 before it is aborted at the local end. While a perfectly 518 valid "receive only" state for an un-orphaned connection, an 519 orphaned connection in FIN_WAIT_2 state could otherwise wait 520 forever for the remote to close its end of the connection. 521 522 Cf. tcp_max_orphans 523 524 Default: 60 seconds 525 526tcp_frto - INTEGER 527 Enables Forward RTO-Recovery (F-RTO) defined in RFC5682. 528 F-RTO is an enhanced recovery algorithm for TCP retransmission 529 timeouts. It is particularly beneficial in networks where the 530 RTT fluctuates (e.g., wireless). F-RTO is sender-side only 531 modification. It does not require any support from the peer. 532 533 By default it's enabled with a non-zero value. 0 disables F-RTO. 534 535tcp_fwmark_accept - BOOLEAN 536 If enabled, incoming connections to listening sockets that do not have a 537 socket mark will set the mark of the accepting socket to the fwmark of 538 the incoming SYN packet. This will cause all packets on that connection 539 (starting from the first SYNACK) to be sent with that fwmark. The 540 listening socket's mark is unchanged. Listening sockets that already 541 have a fwmark set via setsockopt(SOL_SOCKET, SO_MARK, ...) are 542 unaffected. 543 544 Possible values: 545 546 - 0 (disabled) 547 - 1 (enabled) 548 549 Default: 0 (disabled) 550 551tcp_invalid_ratelimit - INTEGER 552 Limit the maximal rate for sending duplicate acknowledgments 553 in response to incoming TCP packets that are for an existing 554 connection but that are invalid due to any of these reasons: 555 556 (a) out-of-window sequence number, 557 (b) out-of-window acknowledgment number, or 558 (c) PAWS (Protection Against Wrapped Sequence numbers) check failure 559 560 This can help mitigate simple "ack loop" DoS attacks, wherein 561 a buggy or malicious middlebox or man-in-the-middle can 562 rewrite TCP header fields in manner that causes each endpoint 563 to think that the other is sending invalid TCP segments, thus 564 causing each side to send an unterminating stream of duplicate 565 acknowledgments for invalid segments. 566 567 Using 0 disables rate-limiting of dupacks in response to 568 invalid segments; otherwise this value specifies the minimal 569 space between sending such dupacks, in milliseconds. 570 571 Default: 500 (milliseconds). 572 573tcp_keepalive_time - INTEGER 574 How often TCP sends out keepalive messages when keepalive is enabled. 575 Default: 2hours. 576 577tcp_keepalive_probes - INTEGER 578 How many keepalive probes TCP sends out, until it decides that the 579 connection is broken. Default value: 9. 580 581tcp_keepalive_intvl - INTEGER 582 How frequently the probes are send out. Multiplied by 583 tcp_keepalive_probes it is time to kill not responding connection, 584 after probes started. Default value: 75sec i.e. connection 585 will be aborted after ~11 minutes of retries. 586 587tcp_l3mdev_accept - BOOLEAN 588 Enables child sockets to inherit the L3 master device index. 589 Enabling this option allows a "global" listen socket to work 590 across L3 master domains (e.g., VRFs) with connected sockets 591 derived from the listen socket to be bound to the L3 domain in 592 which the packets originated. Only valid when the kernel was 593 compiled with CONFIG_NET_L3_MASTER_DEV. 594 595 Possible values: 596 597 - 0 (disabled) 598 - 1 (enabled) 599 600 Default: 0 (disabled) 601 602tcp_low_latency - BOOLEAN 603 This is a legacy option, it has no effect anymore. 604 605tcp_max_orphans - INTEGER 606 Maximal number of TCP sockets not attached to any user file handle, 607 held by system. If this number is exceeded orphaned connections are 608 reset immediately and warning is printed. This limit exists 609 only to prevent simple DoS attacks, you _must_ not rely on this 610 or lower the limit artificially, but rather increase it 611 (probably, after increasing installed memory), 612 if network conditions require more than default value, 613 and tune network services to linger and kill such states 614 more aggressively. Let me to remind again: each orphan eats 615 up to ~64K of unswappable memory. 616 617tcp_max_syn_backlog - INTEGER 618 Maximal number of remembered connection requests (SYN_RECV), 619 which have not received an acknowledgment from connecting client. 620 621 This is a per-listener limit. 622 623 The minimal value is 128 for low memory machines, and it will 624 increase in proportion to the memory of machine. 625 626 If server suffers from overload, try increasing this number. 627 628 Remember to also check /proc/sys/net/core/somaxconn 629 A SYN_RECV request socket consumes about 304 bytes of memory. 630 631tcp_max_tw_buckets - INTEGER 632 Maximal number of timewait sockets held by system simultaneously. 633 If this number is exceeded time-wait socket is immediately destroyed 634 and warning is printed. This limit exists only to prevent 635 simple DoS attacks, you _must_ not lower the limit artificially, 636 but rather increase it (probably, after increasing installed memory), 637 if network conditions require more than default value. 638 639tcp_mem - vector of 3 INTEGERs: min, pressure, max 640 min: below this number of pages TCP is not bothered about its 641 memory appetite. 642 643 pressure: when amount of memory allocated by TCP exceeds this number 644 of pages, TCP moderates its memory consumption and enters memory 645 pressure mode, which is exited when memory consumption falls 646 under "min". 647 648 max: number of pages allowed for queueing by all TCP sockets. 649 650 Defaults are calculated at boot time from amount of available 651 memory. 652 653tcp_min_rtt_wlen - INTEGER 654 The window length of the windowed min filter to track the minimum RTT. 655 A shorter window lets a flow more quickly pick up new (higher) 656 minimum RTT when it is moved to a longer path (e.g., due to traffic 657 engineering). A longer window makes the filter more resistant to RTT 658 inflations such as transient congestion. The unit is seconds. 659 660 Possible values: 0 - 86400 (1 day) 661 662 Default: 300 663 664tcp_moderate_rcvbuf - BOOLEAN 665 If enabled, TCP performs receive buffer auto-tuning, attempting to 666 automatically size the buffer (no greater than tcp_rmem[2]) to 667 match the size required by the path for full throughput. 668 669 Possible values: 670 671 - 0 (disabled) 672 - 1 (enabled) 673 674 Default: 1 (enabled) 675 676tcp_mtu_probing - INTEGER 677 Controls TCP Packetization-Layer Path MTU Discovery. Takes three 678 values: 679 680 - 0 - Disabled 681 - 1 - Disabled by default, enabled when an ICMP black hole detected 682 - 2 - Always enabled, use initial MSS of tcp_base_mss. 683 684tcp_probe_interval - UNSIGNED INTEGER 685 Controls how often to start TCP Packetization-Layer Path MTU 686 Discovery reprobe. The default is reprobing every 10 minutes as 687 per RFC4821. 688 689tcp_probe_threshold - INTEGER 690 Controls when TCP Packetization-Layer Path MTU Discovery probing 691 will stop in respect to the width of search range in bytes. Default 692 is 8 bytes. 693 694tcp_no_metrics_save - BOOLEAN 695 By default, TCP saves various connection metrics in the route cache 696 when the connection closes, so that connections established in the 697 near future can use these to set initial conditions. Usually, this 698 increases overall performance, but may sometimes cause performance 699 degradation. If enabled, TCP will not cache metrics on closing 700 connections. 701 702 Possible values: 703 704 - 0 (disabled) 705 - 1 (enabled) 706 707 Default: 0 (disabled) 708 709tcp_no_ssthresh_metrics_save - BOOLEAN 710 Controls whether TCP saves ssthresh metrics in the route cache. 711 If enabled, ssthresh metrics are disabled. 712 713 Possible values: 714 715 - 0 (disabled) 716 - 1 (enabled) 717 718 Default: 1 (enabled) 719 720tcp_orphan_retries - INTEGER 721 This value influences the timeout of a locally closed TCP connection, 722 when RTO retransmissions remain unacknowledged. 723 See tcp_retries2 for more details. 724 725 The default value is 8. 726 727 If your machine is a loaded WEB server, 728 you should think about lowering this value, such sockets 729 may consume significant resources. Cf. tcp_max_orphans. 730 731tcp_recovery - INTEGER 732 This value is a bitmap to enable various experimental loss recovery 733 features. 734 735 ========= ============================================================= 736 RACK: 0x1 enables RACK loss detection, for fast detection of lost 737 retransmissions and tail drops, and resilience to 738 reordering. currently, setting this bit to 0 has no 739 effect, since RACK is the only supported loss detection 740 algorithm. 741 742 RACK: 0x2 makes RACK's reordering window static (min_rtt/4). 743 744 RACK: 0x4 disables RACK's DUPACK threshold heuristic 745 ========= ============================================================= 746 747 Default: 0x1 748 749tcp_reflect_tos - BOOLEAN 750 For listening sockets, reuse the DSCP value of the initial SYN message 751 for outgoing packets. This allows to have both directions of a TCP 752 stream to use the same DSCP value, assuming DSCP remains unchanged for 753 the lifetime of the connection. 754 755 This options affects both IPv4 and IPv6. 756 757 Possible values: 758 759 - 0 (disabled) 760 - 1 (enabled) 761 762 Default: 0 (disabled) 763 764tcp_reordering - INTEGER 765 Initial reordering level of packets in a TCP stream. 766 TCP stack can then dynamically adjust flow reordering level 767 between this initial value and tcp_max_reordering 768 769 Default: 3 770 771tcp_max_reordering - INTEGER 772 Maximal reordering level of packets in a TCP stream. 773 300 is a fairly conservative value, but you might increase it 774 if paths are using per packet load balancing (like bonding rr mode) 775 776 Default: 300 777 778tcp_retrans_collapse - BOOLEAN 779 Bug-to-bug compatibility with some broken printers. 780 On retransmit try to send bigger packets to work around bugs in 781 certain TCP stacks. 782 783 Possible values: 784 785 - 0 (disabled) 786 - 1 (enabled) 787 788 Default: 1 (enabled) 789 790tcp_retries1 - INTEGER 791 This value influences the time, after which TCP decides, that 792 something is wrong due to unacknowledged RTO retransmissions, 793 and reports this suspicion to the network layer. 794 See tcp_retries2 for more details. 795 796 RFC 1122 recommends at least 3 retransmissions, which is the 797 default. 798 799tcp_retries2 - INTEGER 800 This value influences the timeout of an alive TCP connection, 801 when RTO retransmissions remain unacknowledged. 802 Given a value of N, a hypothetical TCP connection following 803 exponential backoff with an initial RTO of TCP_RTO_MIN would 804 retransmit N times before killing the connection at the (N+1)th RTO. 805 806 The default value of 15 yields a hypothetical timeout of 924.6 807 seconds and is a lower bound for the effective timeout. 808 TCP will effectively time out at the first RTO which exceeds the 809 hypothetical timeout. 810 If tcp_rto_max_ms is decreased, it is recommended to also 811 change tcp_retries2. 812 813 RFC 1122 recommends at least 100 seconds for the timeout, 814 which corresponds to a value of at least 8. 815 816tcp_rfc1337 - BOOLEAN 817 If enabled, the TCP stack behaves conforming to RFC1337. If unset, 818 we are not conforming to RFC, but prevent TCP TIME_WAIT 819 assassination. 820 821 Possible values: 822 823 - 0 (disabled) 824 - 1 (enabled) 825 826 Default: 0 (disabled) 827 828tcp_rmem - vector of 3 INTEGERs: min, default, max 829 min: Minimal size of receive buffer used by TCP sockets. 830 It is guaranteed to each TCP socket, even under moderate memory 831 pressure. 832 833 Default: 4K 834 835 default: initial size of receive buffer used by TCP sockets. 836 This value overrides net.core.rmem_default used by other protocols. 837 Default: 131072 bytes. 838 This value results in initial window of 65535. 839 840 max: maximal size of receive buffer allowed for automatically 841 selected receiver buffers for TCP socket. 842 Calling setsockopt() with SO_RCVBUF disables 843 automatic tuning of that socket's receive buffer size, in which 844 case this value is ignored. 845 Default: between 131072 and 32MB, depending on RAM size. 846 847tcp_sack - BOOLEAN 848 Enable select acknowledgments (SACKS). 849 850 Possible values: 851 852 - 0 (disabled) 853 - 1 (enabled) 854 855 Default: 1 (enabled) 856 857tcp_comp_sack_delay_ns - LONG INTEGER 858 TCP tries to reduce number of SACK sent, using a timer 859 based on 5% of SRTT, capped by this sysctl, in nano seconds. 860 The default is 1ms, based on TSO autosizing period. 861 862 Default : 1,000,000 ns (1 ms) 863 864tcp_comp_sack_slack_ns - LONG INTEGER 865 This sysctl control the slack used when arming the 866 timer used by SACK compression. This gives extra time 867 for small RTT flows, and reduces system overhead by allowing 868 opportunistic reduction of timer interrupts. 869 870 Default : 100,000 ns (100 us) 871 872tcp_comp_sack_nr - INTEGER 873 Max number of SACK that can be compressed. 874 Using 0 disables SACK compression. 875 876 Default : 44 877 878tcp_backlog_ack_defer - BOOLEAN 879 If enabled, user thread processing socket backlog tries sending 880 one ACK for the whole queue. This helps to avoid potential 881 long latencies at end of a TCP socket syscall. 882 883 Possible values: 884 885 - 0 (disabled) 886 - 1 (enabled) 887 888 Default: 1 (enabled) 889 890tcp_slow_start_after_idle - BOOLEAN 891 If enabled, provide RFC2861 behavior and time out the congestion 892 window after an idle period. An idle period is defined at 893 the current RTO. If unset, the congestion window will not 894 be timed out after an idle period. 895 896 Possible values: 897 898 - 0 (disabled) 899 - 1 (enabled) 900 901 Default: 1 (enabled) 902 903tcp_stdurg - BOOLEAN 904 Use the Host requirements interpretation of the TCP urgent pointer field. 905 Most hosts use the older BSD interpretation, so if enabled, 906 Linux might not communicate correctly with them. 907 908 Possible values: 909 910 - 0 (disabled) 911 - 1 (enabled) 912 913 Default: 0 (disabled) 914 915tcp_synack_retries - INTEGER 916 Number of times SYNACKs for a passive TCP connection attempt will 917 be retransmitted. Should not be higher than 255. Default value 918 is 5, which corresponds to 31seconds till the last retransmission 919 with the current initial RTO of 1second. With this the final timeout 920 for a passive TCP connection will happen after 63seconds. 921 922tcp_syncookies - INTEGER 923 Only valid when the kernel was compiled with CONFIG_SYN_COOKIES 924 Send out syncookies when the syn backlog queue of a socket 925 overflows. This is to prevent against the common 'SYN flood attack' 926 Default: 1 927 928 Note, that syncookies is fallback facility. 929 It MUST NOT be used to help highly loaded servers to stand 930 against legal connection rate. If you see SYN flood warnings 931 in your logs, but investigation shows that they occur 932 because of overload with legal connections, you should tune 933 another parameters until this warning disappear. 934 See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow. 935 936 syncookies seriously violate TCP protocol, do not allow 937 to use TCP extensions, can result in serious degradation 938 of some services (f.e. SMTP relaying), visible not by you, 939 but your clients and relays, contacting you. While you see 940 SYN flood warnings in logs not being really flooded, your server 941 is seriously misconfigured. 942 943 If you want to test which effects syncookies have to your 944 network connections you can set this knob to 2 to enable 945 unconditionally generation of syncookies. 946 947tcp_migrate_req - BOOLEAN 948 The incoming connection is tied to a specific listening socket when 949 the initial SYN packet is received during the three-way handshake. 950 When a listener is closed, in-flight request sockets during the 951 handshake and established sockets in the accept queue are aborted. 952 953 If the listener has SO_REUSEPORT enabled, other listeners on the 954 same port should have been able to accept such connections. This 955 option makes it possible to migrate such child sockets to another 956 listener after close() or shutdown(). 957 958 The BPF_SK_REUSEPORT_SELECT_OR_MIGRATE type of eBPF program should 959 usually be used to define the policy to pick an alive listener. 960 Otherwise, the kernel will randomly pick an alive listener only if 961 this option is enabled. 962 963 Note that migration between listeners with different settings may 964 crash applications. Let's say migration happens from listener A to 965 B, and only B has TCP_SAVE_SYN enabled. B cannot read SYN data from 966 the requests migrated from A. To avoid such a situation, cancel 967 migration by returning SK_DROP in the type of eBPF program, or 968 disable this option. 969 970 Possible values: 971 972 - 0 (disabled) 973 - 1 (enabled) 974 975 Default: 0 (disabled) 976 977tcp_fastopen - INTEGER 978 Enable TCP Fast Open (RFC7413) to send and accept data in the opening 979 SYN packet. 980 981 The client support is enabled by flag 0x1 (on by default). The client 982 then must use sendmsg() or sendto() with the MSG_FASTOPEN flag, 983 rather than connect() to send data in SYN. 984 985 The server support is enabled by flag 0x2 (off by default). Then 986 either enable for all listeners with another flag (0x400) or 987 enable individual listeners via TCP_FASTOPEN socket option with 988 the option value being the length of the syn-data backlog. 989 990 The values (bitmap) are 991 992 ===== ======== ====================================================== 993 0x1 (client) enables sending data in the opening SYN on the client. 994 0x2 (server) enables the server support, i.e., allowing data in 995 a SYN packet to be accepted and passed to the 996 application before 3-way handshake finishes. 997 0x4 (client) send data in the opening SYN regardless of cookie 998 availability and without a cookie option. 999 0x200 (server) accept data-in-SYN w/o any cookie option present. 1000 0x400 (server) enable all listeners to support Fast Open by 1001 default without explicit TCP_FASTOPEN socket option. 1002 ===== ======== ====================================================== 1003 1004 Default: 0x1 1005 1006 Note that additional client or server features are only 1007 effective if the basic support (0x1 and 0x2) are enabled respectively. 1008 1009tcp_fastopen_blackhole_timeout_sec - INTEGER 1010 Initial time period in second to disable Fastopen on active TCP sockets 1011 when a TFO firewall blackhole issue happens. 1012 This time period will grow exponentially when more blackhole issues 1013 get detected right after Fastopen is re-enabled and will reset to 1014 initial value when the blackhole issue goes away. 1015 0 to disable the blackhole detection. 1016 1017 By default, it is set to 0 (feature is disabled). 1018 1019tcp_fastopen_key - list of comma separated 32-digit hexadecimal INTEGERs 1020 The list consists of a primary key and an optional backup key. The 1021 primary key is used for both creating and validating cookies, while the 1022 optional backup key is only used for validating cookies. The purpose of 1023 the backup key is to maximize TFO validation when keys are rotated. 1024 1025 A randomly chosen primary key may be configured by the kernel if 1026 the tcp_fastopen sysctl is set to 0x400 (see above), or if the 1027 TCP_FASTOPEN setsockopt() optname is set and a key has not been 1028 previously configured via sysctl. If keys are configured via 1029 setsockopt() by using the TCP_FASTOPEN_KEY optname, then those 1030 per-socket keys will be used instead of any keys that are specified via 1031 sysctl. 1032 1033 A key is specified as 4 8-digit hexadecimal integers which are separated 1034 by a '-' as: xxxxxxxx-xxxxxxxx-xxxxxxxx-xxxxxxxx. Leading zeros may be 1035 omitted. A primary and a backup key may be specified by separating them 1036 by a comma. If only one key is specified, it becomes the primary key and 1037 any previously configured backup keys are removed. 1038 1039tcp_syn_retries - INTEGER 1040 Number of times initial SYNs for an active TCP connection attempt 1041 will be retransmitted. Should not be higher than 127. Default value 1042 is 6, which corresponds to 67seconds (with tcp_syn_linear_timeouts = 4) 1043 till the last retransmission with the current initial RTO of 1second. 1044 With this the final timeout for an active TCP connection attempt 1045 will happen after 131seconds. 1046 1047tcp_timestamps - INTEGER 1048 Enable timestamps as defined in RFC1323. 1049 1050 - 0: Disabled. 1051 - 1: Enable timestamps as defined in RFC1323 and use random offset for 1052 each connection rather than only using the current time. 1053 - 2: Like 1, but without random offsets. 1054 1055 Default: 1 1056 1057tcp_min_tso_segs - INTEGER 1058 Minimal number of segments per TSO frame. 1059 1060 Since linux-3.12, TCP does an automatic sizing of TSO frames, 1061 depending on flow rate, instead of filling 64Kbytes packets. 1062 For specific usages, it's possible to force TCP to build big 1063 TSO frames. Note that TCP stack might split too big TSO packets 1064 if available window is too small. 1065 1066 Default: 2 1067 1068tcp_tso_rtt_log - INTEGER 1069 Adjustment of TSO packet sizes based on min_rtt 1070 1071 Starting from linux-5.18, TCP autosizing can be tweaked 1072 for flows having small RTT. 1073 1074 Old autosizing was splitting the pacing budget to send 1024 TSO 1075 per second. 1076 1077 tso_packet_size = sk->sk_pacing_rate / 1024; 1078 1079 With the new mechanism, we increase this TSO sizing using: 1080 1081 distance = min_rtt_usec / (2^tcp_tso_rtt_log) 1082 tso_packet_size += gso_max_size >> distance; 1083 1084 This means that flows between very close hosts can use bigger 1085 TSO packets, reducing their cpu costs. 1086 1087 If you want to use the old autosizing, set this sysctl to 0. 1088 1089 Default: 9 (2^9 = 512 usec) 1090 1091tcp_pacing_ss_ratio - INTEGER 1092 sk->sk_pacing_rate is set by TCP stack using a ratio applied 1093 to current rate. (current_rate = cwnd * mss / srtt) 1094 If TCP is in slow start, tcp_pacing_ss_ratio is applied 1095 to let TCP probe for bigger speeds, assuming cwnd can be 1096 doubled every other RTT. 1097 1098 Default: 200 1099 1100tcp_pacing_ca_ratio - INTEGER 1101 sk->sk_pacing_rate is set by TCP stack using a ratio applied 1102 to current rate. (current_rate = cwnd * mss / srtt) 1103 If TCP is in congestion avoidance phase, tcp_pacing_ca_ratio 1104 is applied to conservatively probe for bigger throughput. 1105 1106 Default: 120 1107 1108tcp_syn_linear_timeouts - INTEGER 1109 The number of times for an active TCP connection to retransmit SYNs with 1110 a linear backoff timeout before defaulting to an exponential backoff 1111 timeout. This has no effect on SYNACK at the passive TCP side. 1112 1113 With an initial RTO of 1 and tcp_syn_linear_timeouts = 4 we would 1114 expect SYN RTOs to be: 1, 1, 1, 1, 1, 2, 4, ... (4 linear timeouts, 1115 and the first exponential backoff using 2^0 * initial_RTO). 1116 Default: 4 1117 1118tcp_tso_win_divisor - INTEGER 1119 This allows control over what percentage of the congestion window 1120 can be consumed by a single TSO frame. 1121 The setting of this parameter is a choice between burstiness and 1122 building larger TSO frames. 1123 1124 Default: 3 1125 1126tcp_tw_reuse - INTEGER 1127 Enable reuse of TIME-WAIT sockets for new connections when it is 1128 safe from protocol viewpoint. 1129 1130 - 0 - disable 1131 - 1 - global enable 1132 - 2 - enable for loopback traffic only 1133 1134 It should not be changed without advice/request of technical 1135 experts. 1136 1137 Default: 2 1138 1139tcp_tw_reuse_delay - UNSIGNED INTEGER 1140 The delay in milliseconds before a TIME-WAIT socket can be reused by a 1141 new connection, if TIME-WAIT socket reuse is enabled. The actual reuse 1142 threshold is within [N, N+1] range, where N is the requested delay in 1143 milliseconds, to ensure the delay interval is never shorter than the 1144 configured value. 1145 1146 This setting contains an assumption about the other TCP timestamp clock 1147 tick interval. It should not be set to a value lower than the peer's 1148 clock tick for PAWS (Protection Against Wrapped Sequence numbers) 1149 mechanism work correctly for the reused connection. 1150 1151 Default: 1000 (milliseconds) 1152 1153tcp_window_scaling - BOOLEAN 1154 Enable window scaling as defined in RFC1323. 1155 1156 Possible values: 1157 1158 - 0 (disabled) 1159 - 1 (enabled) 1160 1161 Default: 1 (enabled) 1162 1163tcp_shrink_window - BOOLEAN 1164 This changes how the TCP receive window is calculated. 1165 1166 RFC 7323, section 2.4, says there are instances when a retracted 1167 window can be offered, and that TCP implementations MUST ensure 1168 that they handle a shrinking window, as specified in RFC 1122. 1169 1170 Possible values: 1171 1172 - 0 (disabled) - The window is never shrunk. 1173 - 1 (enabled) - The window is shrunk when necessary to remain within 1174 the memory limit set by autotuning (sk_rcvbuf). 1175 This only occurs if a non-zero receive window 1176 scaling factor is also in effect. 1177 1178 Default: 0 (disabled) 1179 1180tcp_wmem - vector of 3 INTEGERs: min, default, max 1181 min: Amount of memory reserved for send buffers for TCP sockets. 1182 Each TCP socket has rights to use it due to fact of its birth. 1183 1184 Default: 4K 1185 1186 default: initial size of send buffer used by TCP sockets. This 1187 value overrides net.core.wmem_default used by other protocols. 1188 1189 It is usually lower than net.core.wmem_default. 1190 1191 Default: 16K 1192 1193 max: Maximal amount of memory allowed for automatically tuned 1194 send buffers for TCP sockets. This value does not override 1195 net.core.wmem_max. Calling setsockopt() with SO_SNDBUF disables 1196 automatic tuning of that socket's send buffer size, in which case 1197 this value is ignored. 1198 1199 Default: between 64K and 4MB, depending on RAM size. 1200 1201tcp_notsent_lowat - UNSIGNED INTEGER 1202 A TCP socket can control the amount of unsent bytes in its write queue, 1203 thanks to TCP_NOTSENT_LOWAT socket option. poll()/select()/epoll() 1204 reports POLLOUT events if the amount of unsent bytes is below a per 1205 socket value, and if the write queue is not full. sendmsg() will 1206 also not add new buffers if the limit is hit. 1207 1208 This global variable controls the amount of unsent data for 1209 sockets not using TCP_NOTSENT_LOWAT. For these sockets, a change 1210 to the global variable has immediate effect. 1211 1212 Default: UINT_MAX (0xFFFFFFFF) 1213 1214tcp_workaround_signed_windows - BOOLEAN 1215 If enabled, assume no receipt of a window scaling option means the 1216 remote TCP is broken and treats the window as a signed quantity. 1217 If disabled, assume the remote TCP is not broken even if we do 1218 not receive a window scaling option from them. 1219 1220 Possible values: 1221 1222 - 0 (disabled) 1223 - 1 (enabled) 1224 1225 Default: 0 (disabled) 1226 1227tcp_thin_linear_timeouts - BOOLEAN 1228 Enable dynamic triggering of linear timeouts for thin streams. 1229 If enabled, a check is performed upon retransmission by timeout to 1230 determine if the stream is thin (less than 4 packets in flight). 1231 As long as the stream is found to be thin, up to 6 linear 1232 timeouts may be performed before exponential backoff mode is 1233 initiated. This improves retransmission latency for 1234 non-aggressive thin streams, often found to be time-dependent. 1235 For more information on thin streams, see 1236 Documentation/networking/tcp-thin.rst 1237 1238 Possible values: 1239 1240 - 0 (disabled) 1241 - 1 (enabled) 1242 1243 Default: 0 (disabled) 1244 1245tcp_limit_output_bytes - INTEGER 1246 Controls TCP Small Queue limit per tcp socket. 1247 TCP bulk sender tends to increase packets in flight until it 1248 gets losses notifications. With SNDBUF autotuning, this can 1249 result in a large amount of packets queued on the local machine 1250 (e.g.: qdiscs, CPU backlog, or device) hurting latency of other 1251 flows, for typical pfifo_fast qdiscs. tcp_limit_output_bytes 1252 limits the number of bytes on qdisc or device to reduce artificial 1253 RTT/cwnd and reduce bufferbloat. 1254 1255 Default: 4194304 (4 MB) 1256 1257tcp_challenge_ack_limit - INTEGER 1258 Limits number of Challenge ACK sent per second, as recommended 1259 in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks) 1260 Note that this per netns rate limit can allow some side channel 1261 attacks and probably should not be enabled. 1262 TCP stack implements per TCP socket limits anyway. 1263 Default: INT_MAX (unlimited) 1264 1265tcp_ehash_entries - INTEGER 1266 Show the number of hash buckets for TCP sockets in the current 1267 networking namespace. 1268 1269 A negative value means the networking namespace does not own its 1270 hash buckets and shares the initial networking namespace's one. 1271 1272tcp_child_ehash_entries - INTEGER 1273 Control the number of hash buckets for TCP sockets in the child 1274 networking namespace, which must be set before clone() or unshare(). 1275 1276 If the value is not 0, the kernel uses a value rounded up to 2^n 1277 as the actual hash bucket size. 0 is a special value, meaning 1278 the child networking namespace will share the initial networking 1279 namespace's hash buckets. 1280 1281 Note that the child will use the global one in case the kernel 1282 fails to allocate enough memory. In addition, the global hash 1283 buckets are spread over available NUMA nodes, but the allocation 1284 of the child hash table depends on the current process's NUMA 1285 policy, which could result in performance differences. 1286 1287 Note also that the default value of tcp_max_tw_buckets and 1288 tcp_max_syn_backlog depend on the hash bucket size. 1289 1290 Possible values: 0, 2^n (n: 0 - 24 (16Mi)) 1291 1292 Default: 0 1293 1294tcp_plb_enabled - BOOLEAN 1295 If enabled and the underlying congestion control (e.g. DCTCP) supports 1296 and enables PLB feature, TCP PLB (Protective Load Balancing) is 1297 enabled. PLB is described in the following paper: 1298 https://doi.org/10.1145/3544216.3544226. Based on PLB parameters, 1299 upon sensing sustained congestion, TCP triggers a change in 1300 flow label field for outgoing IPv6 packets. A change in flow label 1301 field potentially changes the path of outgoing packets for switches 1302 that use ECMP/WCMP for routing. 1303 1304 PLB changes socket txhash which results in a change in IPv6 Flow Label 1305 field, and currently no-op for IPv4 headers. It is possible 1306 to apply PLB for IPv4 with other network header fields (e.g. TCP 1307 or IPv4 options) or using encapsulation where outer header is used 1308 by switches to determine next hop. In either case, further host 1309 and switch side changes will be needed. 1310 1311 If enabled, PLB assumes that congestion signal (e.g. ECN) is made 1312 available and used by congestion control module to estimate a 1313 congestion measure (e.g. ce_ratio). PLB needs a congestion measure to 1314 make repathing decisions. 1315 1316 Possible values: 1317 1318 - 0 (disabled) 1319 - 1 (enabled) 1320 1321 Default: 0 (disabled) 1322 1323tcp_plb_idle_rehash_rounds - INTEGER 1324 Number of consecutive congested rounds (RTT) seen after which 1325 a rehash can be performed, given there are no packets in flight. 1326 This is referred to as M in PLB paper: 1327 https://doi.org/10.1145/3544216.3544226. 1328 1329 Possible Values: 0 - 31 1330 1331 Default: 3 1332 1333tcp_plb_rehash_rounds - INTEGER 1334 Number of consecutive congested rounds (RTT) seen after which 1335 a forced rehash can be performed. Be careful when setting this 1336 parameter, as a small value increases the risk of retransmissions. 1337 This is referred to as N in PLB paper: 1338 https://doi.org/10.1145/3544216.3544226. 1339 1340 Possible Values: 0 - 31 1341 1342 Default: 12 1343 1344tcp_plb_suspend_rto_sec - INTEGER 1345 Time, in seconds, to suspend PLB in event of an RTO. In order to avoid 1346 having PLB repath onto a connectivity "black hole", after an RTO a TCP 1347 connection suspends PLB repathing for a random duration between 1x and 1348 2x of this parameter. Randomness is added to avoid concurrent rehashing 1349 of multiple TCP connections. This should be set corresponding to the 1350 amount of time it takes to repair a failed link. 1351 1352 Possible Values: 0 - 255 1353 1354 Default: 60 1355 1356tcp_plb_cong_thresh - INTEGER 1357 Fraction of packets marked with congestion over a round (RTT) to 1358 tag that round as congested. This is referred to as K in the PLB paper: 1359 https://doi.org/10.1145/3544216.3544226. 1360 1361 The 0-1 fraction range is mapped to 0-256 range to avoid floating 1362 point operations. For example, 128 means that if at least 50% of 1363 the packets in a round were marked as congested then the round 1364 will be tagged as congested. 1365 1366 Setting threshold to 0 means that PLB repaths every RTT regardless 1367 of congestion. This is not intended behavior for PLB and should be 1368 used only for experimentation purpose. 1369 1370 Possible Values: 0 - 256 1371 1372 Default: 128 1373 1374tcp_pingpong_thresh - INTEGER 1375 The number of estimated data replies sent for estimated incoming data 1376 requests that must happen before TCP considers that a connection is a 1377 "ping-pong" (request-response) connection for which delayed 1378 acknowledgments can provide benefits. 1379 1380 This threshold is 1 by default, but some applications may need a higher 1381 threshold for optimal performance. 1382 1383 Possible Values: 1 - 255 1384 1385 Default: 1 1386 1387tcp_rto_min_us - INTEGER 1388 Minimal TCP retransmission timeout (in microseconds). Note that the 1389 rto_min route option has the highest precedence for configuring this 1390 setting, followed by the TCP_BPF_RTO_MIN and TCP_RTO_MIN_US socket 1391 options, followed by this tcp_rto_min_us sysctl. 1392 1393 The recommended practice is to use a value less or equal to 200000 1394 microseconds. 1395 1396 Possible Values: 1 - INT_MAX 1397 1398 Default: 200000 1399 1400tcp_rto_max_ms - INTEGER 1401 Maximal TCP retransmission timeout (in ms). 1402 Note that TCP_RTO_MAX_MS socket option has higher precedence. 1403 1404 When changing tcp_rto_max_ms, it is important to understand 1405 that tcp_retries2 might need a change. 1406 1407 Possible Values: 1000 - 120,000 1408 1409 Default: 120,000 1410 1411UDP variables 1412============= 1413 1414udp_l3mdev_accept - BOOLEAN 1415 Enabling this option allows a "global" bound socket to work 1416 across L3 master domains (e.g., VRFs) with packets capable of 1417 being received regardless of the L3 domain in which they 1418 originated. Only valid when the kernel was compiled with 1419 CONFIG_NET_L3_MASTER_DEV. 1420 1421 Possible values: 1422 1423 - 0 (disabled) 1424 - 1 (enabled) 1425 1426 Default: 0 (disabled) 1427 1428udp_mem - vector of 3 INTEGERs: min, pressure, max 1429 Number of pages allowed for queueing by all UDP sockets. 1430 1431 min: Number of pages allowed for queueing by all UDP sockets. 1432 1433 pressure: This value was introduced to follow format of tcp_mem. 1434 1435 max: This value was introduced to follow format of tcp_mem. 1436 1437 Default is calculated at boot time from amount of available memory. 1438 1439udp_rmem_min - INTEGER 1440 Minimal size of receive buffer used by UDP sockets in moderation. 1441 Each UDP socket is able to use the size for receiving data, even if 1442 total pages of UDP sockets exceed udp_mem pressure. The unit is byte. 1443 1444 Default: 4K 1445 1446udp_wmem_min - INTEGER 1447 UDP does not have tx memory accounting and this tunable has no effect. 1448 1449udp_hash_entries - INTEGER 1450 Show the number of hash buckets for UDP sockets in the current 1451 networking namespace. 1452 1453 A negative value means the networking namespace does not own its 1454 hash buckets and shares the initial networking namespace's one. 1455 1456udp_child_hash_entries - INTEGER 1457 Control the number of hash buckets for UDP sockets in the child 1458 networking namespace, which must be set before clone() or unshare(). 1459 1460 If the value is not 0, the kernel uses a value rounded up to 2^n 1461 as the actual hash bucket size. 0 is a special value, meaning 1462 the child networking namespace will share the initial networking 1463 namespace's hash buckets. 1464 1465 Note that the child will use the global one in case the kernel 1466 fails to allocate enough memory. In addition, the global hash 1467 buckets are spread over available NUMA nodes, but the allocation 1468 of the child hash table depends on the current process's NUMA 1469 policy, which could result in performance differences. 1470 1471 Possible values: 0, 2^n (n: 7 (128) - 16 (64K)) 1472 1473 Default: 0 1474 1475 1476RAW variables 1477============= 1478 1479raw_l3mdev_accept - BOOLEAN 1480 Enabling this option allows a "global" bound socket to work 1481 across L3 master domains (e.g., VRFs) with packets capable of 1482 being received regardless of the L3 domain in which they 1483 originated. Only valid when the kernel was compiled with 1484 CONFIG_NET_L3_MASTER_DEV. 1485 1486 Possible values: 1487 1488 - 0 (disabled) 1489 - 1 (enabled) 1490 1491 Default: 1 (enabled) 1492 1493CIPSOv4 Variables 1494================= 1495 1496cipso_cache_enable - BOOLEAN 1497 If enabled, enable additions to and lookups from the CIPSO label mapping 1498 cache. If disabled, additions are ignored and lookups always result in a 1499 miss. However, regardless of the setting the cache is still 1500 invalidated when required when means you can safely toggle this on and 1501 off and the cache will always be "safe". 1502 1503 Possible values: 1504 1505 - 0 (disabled) 1506 - 1 (enabled) 1507 1508 Default: 1 (enabled) 1509 1510cipso_cache_bucket_size - INTEGER 1511 The CIPSO label cache consists of a fixed size hash table with each 1512 hash bucket containing a number of cache entries. This variable limits 1513 the number of entries in each hash bucket; the larger the value is, the 1514 more CIPSO label mappings that can be cached. When the number of 1515 entries in a given hash bucket reaches this limit adding new entries 1516 causes the oldest entry in the bucket to be removed to make room. 1517 1518 Default: 10 1519 1520cipso_rbm_optfmt - BOOLEAN 1521 Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of 1522 the CIPSO draft specification (see Documentation/netlabel for details). 1523 This means that when set the CIPSO tag will be padded with empty 1524 categories in order to make the packet data 32-bit aligned. 1525 1526 Possible values: 1527 1528 - 0 (disabled) 1529 - 1 (enabled) 1530 1531 Default: 0 (disabled) 1532 1533cipso_rbm_strictvalid - BOOLEAN 1534 If enabled, do a very strict check of the CIPSO option when 1535 ip_options_compile() is called. If disabled, relax the checks done during 1536 ip_options_compile(). Either way is "safe" as errors are caught else 1537 where in the CIPSO processing code but setting this to 0 (False) should 1538 result in less work (i.e. it should be faster) but could cause problems 1539 with other implementations that require strict checking. 1540 1541 Possible values: 1542 1543 - 0 (disabled) 1544 - 1 (enabled) 1545 1546 Default: 0 (disabled) 1547 1548IP Variables 1549============ 1550 1551ip_local_port_range - 2 INTEGERS 1552 Defines the local port range that is used by TCP and UDP to 1553 choose the local port. The first number is the first, the 1554 second the last local port number. 1555 If possible, it is better these numbers have different parity 1556 (one even and one odd value). 1557 Must be greater than or equal to ip_unprivileged_port_start. 1558 The default values are 32768 and 60999 respectively. 1559 1560ip_local_reserved_ports - list of comma separated ranges 1561 Specify the ports which are reserved for known third-party 1562 applications. These ports will not be used by automatic port 1563 assignments (e.g. when calling connect() or bind() with port 1564 number 0). Explicit port allocation behavior is unchanged. 1565 1566 The format used for both input and output is a comma separated 1567 list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and 1568 10). Writing to the file will clear all previously reserved 1569 ports and update the current list with the one given in the 1570 input. 1571 1572 Note that ip_local_port_range and ip_local_reserved_ports 1573 settings are independent and both are considered by the kernel 1574 when determining which ports are available for automatic port 1575 assignments. 1576 1577 You can reserve ports which are not in the current 1578 ip_local_port_range, e.g.:: 1579 1580 $ cat /proc/sys/net/ipv4/ip_local_port_range 1581 32000 60999 1582 $ cat /proc/sys/net/ipv4/ip_local_reserved_ports 1583 8080,9148 1584 1585 although this is redundant. However such a setting is useful 1586 if later the port range is changed to a value that will 1587 include the reserved ports. Also keep in mind, that overlapping 1588 of these ranges may affect probability of selecting ephemeral 1589 ports which are right after block of reserved ports. 1590 1591 Default: Empty 1592 1593ip_unprivileged_port_start - INTEGER 1594 This is a per-namespace sysctl. It defines the first 1595 unprivileged port in the network namespace. Privileged ports 1596 require root or CAP_NET_BIND_SERVICE in order to bind to them. 1597 To disable all privileged ports, set this to 0. They must not 1598 overlap with the ip_local_port_range. 1599 1600 Default: 1024 1601 1602ip_nonlocal_bind - BOOLEAN 1603 If enabled, allows processes to bind() to non-local IP addresses, 1604 which can be quite useful - but may break some applications. 1605 1606 Possible values: 1607 1608 - 0 (disabled) 1609 - 1 (enabled) 1610 1611 Default: 0 (disabled) 1612 1613ip_autobind_reuse - BOOLEAN 1614 By default, bind() does not select the ports automatically even if 1615 the new socket and all sockets bound to the port have SO_REUSEADDR. 1616 ip_autobind_reuse allows bind() to reuse the port and this is useful 1617 when you use bind()+connect(), but may break some applications. 1618 The preferred solution is to use IP_BIND_ADDRESS_NO_PORT and this 1619 option should only be set by experts. 1620 1621 Possible values: 1622 1623 - 0 (disabled) 1624 - 1 (enabled) 1625 1626 Default: 0 (disabled) 1627 1628ip_dynaddr - INTEGER 1629 If set non-zero, enables support for dynamic addresses. 1630 If set to a non-zero value larger than 1, a kernel log 1631 message will be printed when dynamic address rewriting 1632 occurs. 1633 1634 Default: 0 1635 1636ip_early_demux - BOOLEAN 1637 Optimize input packet processing down to one demux for 1638 certain kinds of local sockets. Currently we only do this 1639 for established TCP and connected UDP sockets. 1640 1641 It may add an additional cost for pure routing workloads that 1642 reduces overall throughput, in such case you should disable it. 1643 1644 Possible values: 1645 1646 - 0 (disabled) 1647 - 1 (enabled) 1648 1649 Default: 1 (enabled) 1650 1651ping_group_range - 2 INTEGERS 1652 Restrict ICMP_PROTO datagram sockets to users in the group range. 1653 The default is "1 0", meaning, that nobody (not even root) may 1654 create ping sockets. Setting it to "100 100" would grant permissions 1655 to the single group. "0 4294967294" would enable it for the world, "100 1656 4294967294" would enable it for the users, but not daemons. 1657 1658tcp_early_demux - BOOLEAN 1659 Enable early demux for established TCP sockets. 1660 1661 Possible values: 1662 1663 - 0 (disabled) 1664 - 1 (enabled) 1665 1666 Default: 1 (enabled) 1667 1668udp_early_demux - BOOLEAN 1669 Enable early demux for connected UDP sockets. Disable this if 1670 your system could experience more unconnected load. 1671 1672 Possible values: 1673 1674 - 0 (disabled) 1675 - 1 (enabled) 1676 1677 Default: 1 (enabled) 1678 1679icmp_echo_ignore_all - BOOLEAN 1680 If enabled, then the kernel will ignore all ICMP ECHO 1681 requests sent to it. 1682 1683 Possible values: 1684 1685 - 0 (disabled) 1686 - 1 (enabled) 1687 1688 Default: 0 (disabled) 1689 1690icmp_echo_enable_probe - BOOLEAN 1691 If enabled, then the kernel will respond to RFC 8335 PROBE 1692 requests sent to it. 1693 1694 Possible values: 1695 1696 - 0 (disabled) 1697 - 1 (enabled) 1698 1699 Default: 0 (disabled) 1700 1701icmp_echo_ignore_broadcasts - BOOLEAN 1702 If enabled, then the kernel will ignore all ICMP ECHO and 1703 TIMESTAMP requests sent to it via broadcast/multicast. 1704 1705 Possible values: 1706 1707 - 0 (disabled) 1708 - 1 (enabled) 1709 1710 Default: 1 (enabled) 1711 1712icmp_ratelimit - INTEGER 1713 Limit the maximal rates for sending ICMP packets whose type matches 1714 icmp_ratemask (see below) to specific targets. 1715 0 to disable any limiting, 1716 otherwise the minimal space between responses in milliseconds. 1717 Note that another sysctl, icmp_msgs_per_sec limits the number 1718 of ICMP packets sent on all targets. 1719 1720 Default: 1000 1721 1722icmp_msgs_per_sec - INTEGER 1723 Limit maximal number of ICMP packets sent per second from this host. 1724 Only messages whose type matches icmp_ratemask (see below) are 1725 controlled by this limit. For security reasons, the precise count 1726 of messages per second is randomized. 1727 1728 Default: 1000 1729 1730icmp_msgs_burst - INTEGER 1731 icmp_msgs_per_sec controls number of ICMP packets sent per second, 1732 while icmp_msgs_burst controls the burst size of these packets. 1733 For security reasons, the precise burst size is randomized. 1734 1735 Default: 50 1736 1737icmp_ratemask - INTEGER 1738 Mask made of ICMP types for which rates are being limited. 1739 1740 Significant bits: IHGFEDCBA9876543210 1741 1742 Default mask: 0000001100000011000 (6168) 1743 1744 Bit definitions (see include/linux/icmp.h): 1745 1746 = ========================= 1747 0 Echo Reply 1748 3 Destination Unreachable [1]_ 1749 4 Source Quench [1]_ 1750 5 Redirect 1751 8 Echo Request 1752 B Time Exceeded [1]_ 1753 C Parameter Problem [1]_ 1754 D Timestamp Request 1755 E Timestamp Reply 1756 F Info Request 1757 G Info Reply 1758 H Address Mask Request 1759 I Address Mask Reply 1760 = ========================= 1761 1762 .. [1] These are rate limited by default (see default mask above) 1763 1764icmp_ignore_bogus_error_responses - BOOLEAN 1765 Some routers violate RFC1122 by sending bogus responses to broadcast 1766 frames. Such violations are normally logged via a kernel warning. 1767 If enabled, the kernel will not give such warnings, which 1768 will avoid log file clutter. 1769 1770 Possible values: 1771 1772 - 0 (disabled) 1773 - 1 (enabled) 1774 1775 Default: 1 (enabled) 1776 1777icmp_errors_use_inbound_ifaddr - BOOLEAN 1778 1779 If disabled, icmp error messages are sent with the primary address of 1780 the exiting interface. 1781 1782 If enabled, the message will be sent with the primary address of 1783 the interface that received the packet that caused the icmp error. 1784 This is the behaviour many network administrators will expect from 1785 a router. And it can make debugging complicated network layouts 1786 much easier. 1787 1788 Note that if no primary address exists for the interface selected, 1789 then the primary address of the first non-loopback interface that 1790 has one will be used regardless of this setting. 1791 1792 Possible values: 1793 1794 - 0 (disabled) 1795 - 1 (enabled) 1796 1797 Default: 0 (disabled) 1798 1799igmp_max_memberships - INTEGER 1800 Change the maximum number of multicast groups we can subscribe to. 1801 Default: 20 1802 1803 Theoretical maximum value is bounded by having to send a membership 1804 report in a single datagram (i.e. the report can't span multiple 1805 datagrams, or risk confusing the switch and leaving groups you don't 1806 intend to). 1807 1808 The number of supported groups 'M' is bounded by the number of group 1809 report entries you can fit into a single datagram of 65535 bytes. 1810 1811 M = 65536-sizeof (ip header)/(sizeof(Group record)) 1812 1813 Group records are variable length, with a minimum of 12 bytes. 1814 So net.ipv4.igmp_max_memberships should not be set higher than: 1815 1816 (65536-24) / 12 = 5459 1817 1818 The value 5459 assumes no IP header options, so in practice 1819 this number may be lower. 1820 1821igmp_max_msf - INTEGER 1822 Maximum number of addresses allowed in the source filter list for a 1823 multicast group. 1824 1825 Default: 10 1826 1827igmp_qrv - INTEGER 1828 Controls the IGMP query robustness variable (see RFC2236 8.1). 1829 1830 Default: 2 (as specified by RFC2236 8.1) 1831 1832 Minimum: 1 (as specified by RFC6636 4.5) 1833 1834force_igmp_version - INTEGER 1835 - 0 - (default) No enforcement of a IGMP version, IGMPv1/v2 fallback 1836 allowed. Will back to IGMPv3 mode again if all IGMPv1/v2 Querier 1837 Present timer expires. 1838 - 1 - Enforce to use IGMP version 1. Will also reply IGMPv1 report if 1839 receive IGMPv2/v3 query. 1840 - 2 - Enforce to use IGMP version 2. Will fallback to IGMPv1 if receive 1841 IGMPv1 query message. Will reply report if receive IGMPv3 query. 1842 - 3 - Enforce to use IGMP version 3. The same react with default 0. 1843 1844 .. note:: 1845 1846 this is not the same with force_mld_version because IGMPv3 RFC3376 1847 Security Considerations does not have clear description that we could 1848 ignore other version messages completely as MLDv2 RFC3810. So make 1849 this value as default 0 is recommended. 1850 1851``conf/interface/*`` 1852 changes special settings per interface (where 1853 interface" is the name of your network interface) 1854 1855``conf/all/*`` 1856 is special, changes the settings for all interfaces 1857 1858log_martians - BOOLEAN 1859 Log packets with impossible addresses to kernel log. 1860 log_martians for the interface will be enabled if at least one of 1861 conf/{all,interface}/log_martians is set to TRUE, 1862 it will be disabled otherwise 1863 1864accept_redirects - BOOLEAN 1865 Accept ICMP redirect messages. 1866 accept_redirects for the interface will be enabled if: 1867 1868 - both conf/{all,interface}/accept_redirects are TRUE in the case 1869 forwarding for the interface is enabled 1870 1871 or 1872 1873 - at least one of conf/{all,interface}/accept_redirects is TRUE in the 1874 case forwarding for the interface is disabled 1875 1876 accept_redirects for the interface will be disabled otherwise 1877 1878 default: 1879 1880 - TRUE (host) 1881 - FALSE (router) 1882 1883forwarding - BOOLEAN 1884 Enable IP forwarding on this interface. This controls whether packets 1885 received _on_ this interface can be forwarded. 1886 1887mc_forwarding - BOOLEAN 1888 Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE 1889 and a multicast routing daemon is required. 1890 conf/all/mc_forwarding must also be set to TRUE to enable multicast 1891 routing for the interface 1892 1893medium_id - INTEGER 1894 Integer value used to differentiate the devices by the medium they 1895 are attached to. Two devices can have different id values when 1896 the broadcast packets are received only on one of them. 1897 The default value 0 means that the device is the only interface 1898 to its medium, value of -1 means that medium is not known. 1899 1900 Currently, it is used to change the proxy_arp behavior: 1901 the proxy_arp feature is enabled for packets forwarded between 1902 two devices attached to different media. 1903 1904proxy_arp - BOOLEAN 1905 Do proxy arp. 1906 1907 proxy_arp for the interface will be enabled if at least one of 1908 conf/{all,interface}/proxy_arp is set to TRUE, 1909 it will be disabled otherwise 1910 1911proxy_arp_pvlan - BOOLEAN 1912 Private VLAN proxy arp. 1913 1914 Basically allow proxy arp replies back to the same interface 1915 (from which the ARP request/solicitation was received). 1916 1917 This is done to support (ethernet) switch features, like RFC 1918 3069, where the individual ports are NOT allowed to 1919 communicate with each other, but they are allowed to talk to 1920 the upstream router. As described in RFC 3069, it is possible 1921 to allow these hosts to communicate through the upstream 1922 router by proxy_arp'ing. Don't need to be used together with 1923 proxy_arp. 1924 1925 This technology is known by different names: 1926 1927 - In RFC 3069 it is called VLAN Aggregation. 1928 - Cisco and Allied Telesyn call it Private VLAN. 1929 - Hewlett-Packard call it Source-Port filtering or port-isolation. 1930 - Ericsson call it MAC-Forced Forwarding (RFC Draft). 1931 1932proxy_delay - INTEGER 1933 Delay proxy response. 1934 1935 Delay response to a neighbor solicitation when proxy_arp 1936 or proxy_ndp is enabled. A random value between [0, proxy_delay) 1937 will be chosen, setting to zero means reply with no delay. 1938 Value in jiffies. Defaults to 80. 1939 1940shared_media - BOOLEAN 1941 Send(router) or accept(host) RFC1620 shared media redirects. 1942 Overrides secure_redirects. 1943 1944 shared_media for the interface will be enabled if at least one of 1945 conf/{all,interface}/shared_media is set to TRUE, 1946 it will be disabled otherwise 1947 1948 default TRUE 1949 1950secure_redirects - BOOLEAN 1951 Accept ICMP redirect messages only to gateways listed in the 1952 interface's current gateway list. Even if disabled, RFC1122 redirect 1953 rules still apply. 1954 1955 Overridden by shared_media. 1956 1957 secure_redirects for the interface will be enabled if at least one of 1958 conf/{all,interface}/secure_redirects is set to TRUE, 1959 it will be disabled otherwise 1960 1961 default TRUE 1962 1963send_redirects - BOOLEAN 1964 Send redirects, if router. 1965 1966 send_redirects for the interface will be enabled if at least one of 1967 conf/{all,interface}/send_redirects is set to TRUE, 1968 it will be disabled otherwise 1969 1970 Default: TRUE 1971 1972bootp_relay - BOOLEAN 1973 Accept packets with source address 0.b.c.d destined 1974 not to this host as local ones. It is supposed, that 1975 BOOTP relay daemon will catch and forward such packets. 1976 conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay 1977 for the interface 1978 1979 default FALSE 1980 1981 Not Implemented Yet. 1982 1983accept_source_route - BOOLEAN 1984 Accept packets with SRR option. 1985 conf/all/accept_source_route must also be set to TRUE to accept packets 1986 with SRR option on the interface 1987 1988 default 1989 1990 - TRUE (router) 1991 - FALSE (host) 1992 1993accept_local - BOOLEAN 1994 Accept packets with local source addresses. In combination with 1995 suitable routing, this can be used to direct packets between two 1996 local interfaces over the wire and have them accepted properly. 1997 default FALSE 1998 1999route_localnet - BOOLEAN 2000 Do not consider loopback addresses as martian source or destination 2001 while routing. This enables the use of 127/8 for local routing purposes. 2002 2003 default FALSE 2004 2005rp_filter - INTEGER 2006 - 0 - No source validation. 2007 - 1 - Strict mode as defined in RFC3704 Strict Reverse Path 2008 Each incoming packet is tested against the FIB and if the interface 2009 is not the best reverse path the packet check will fail. 2010 By default failed packets are discarded. 2011 - 2 - Loose mode as defined in RFC3704 Loose Reverse Path 2012 Each incoming packet's source address is also tested against the FIB 2013 and if the source address is not reachable via any interface 2014 the packet check will fail. 2015 2016 Current recommended practice in RFC3704 is to enable strict mode 2017 to prevent IP spoofing from DDos attacks. If using asymmetric routing 2018 or other complicated routing, then loose mode is recommended. 2019 2020 The max value from conf/{all,interface}/rp_filter is used 2021 when doing source validation on the {interface}. 2022 2023 Default value is 0. Note that some distributions enable it 2024 in startup scripts. 2025 2026src_valid_mark - BOOLEAN 2027 - 0 - The fwmark of the packet is not included in reverse path 2028 route lookup. This allows for asymmetric routing configurations 2029 utilizing the fwmark in only one direction, e.g., transparent 2030 proxying. 2031 2032 - 1 - The fwmark of the packet is included in reverse path route 2033 lookup. This permits rp_filter to function when the fwmark is 2034 used for routing traffic in both directions. 2035 2036 This setting also affects the utilization of fmwark when 2037 performing source address selection for ICMP replies, or 2038 determining addresses stored for the IPOPT_TS_TSANDADDR and 2039 IPOPT_RR IP options. 2040 2041 The max value from conf/{all,interface}/src_valid_mark is used. 2042 2043 Default value is 0. 2044 2045arp_filter - BOOLEAN 2046 - 1 - Allows you to have multiple network interfaces on the same 2047 subnet, and have the ARPs for each interface be answered 2048 based on whether or not the kernel would route a packet from 2049 the ARP'd IP out that interface (therefore you must use source 2050 based routing for this to work). In other words it allows control 2051 of which cards (usually 1) will respond to an arp request. 2052 2053 - 0 - (default) The kernel can respond to arp requests with addresses 2054 from other interfaces. This may seem wrong but it usually makes 2055 sense, because it increases the chance of successful communication. 2056 IP addresses are owned by the complete host on Linux, not by 2057 particular interfaces. Only for more complex setups like load- 2058 balancing, does this behaviour cause problems. 2059 2060 arp_filter for the interface will be enabled if at least one of 2061 conf/{all,interface}/arp_filter is set to TRUE, 2062 it will be disabled otherwise 2063 2064arp_announce - INTEGER 2065 Define different restriction levels for announcing the local 2066 source IP address from IP packets in ARP requests sent on 2067 interface: 2068 2069 - 0 - (default) Use any local address, configured on any interface 2070 - 1 - Try to avoid local addresses that are not in the target's 2071 subnet for this interface. This mode is useful when target 2072 hosts reachable via this interface require the source IP 2073 address in ARP requests to be part of their logical network 2074 configured on the receiving interface. When we generate the 2075 request we will check all our subnets that include the 2076 target IP and will preserve the source address if it is from 2077 such subnet. If there is no such subnet we select source 2078 address according to the rules for level 2. 2079 - 2 - Always use the best local address for this target. 2080 In this mode we ignore the source address in the IP packet 2081 and try to select local address that we prefer for talks with 2082 the target host. Such local address is selected by looking 2083 for primary IP addresses on all our subnets on the outgoing 2084 interface that include the target IP address. If no suitable 2085 local address is found we select the first local address 2086 we have on the outgoing interface or on all other interfaces, 2087 with the hope we will receive reply for our request and 2088 even sometimes no matter the source IP address we announce. 2089 2090 The max value from conf/{all,interface}/arp_announce is used. 2091 2092 Increasing the restriction level gives more chance for 2093 receiving answer from the resolved target while decreasing 2094 the level announces more valid sender's information. 2095 2096arp_ignore - INTEGER 2097 Define different modes for sending replies in response to 2098 received ARP requests that resolve local target IP addresses: 2099 2100 - 0 - (default): reply for any local target IP address, configured 2101 on any interface 2102 - 1 - reply only if the target IP address is local address 2103 configured on the incoming interface 2104 - 2 - reply only if the target IP address is local address 2105 configured on the incoming interface and both with the 2106 sender's IP address are part from same subnet on this interface 2107 - 3 - do not reply for local addresses configured with scope host, 2108 only resolutions for global and link addresses are replied 2109 - 4-7 - reserved 2110 - 8 - do not reply for all local addresses 2111 2112 The max value from conf/{all,interface}/arp_ignore is used 2113 when ARP request is received on the {interface} 2114 2115arp_notify - BOOLEAN 2116 Define mode for notification of address and device changes. 2117 2118 == ========================================================== 2119 0 (default): do nothing 2120 1 Generate gratuitous arp requests when device is brought up 2121 or hardware address changes. 2122 == ========================================================== 2123 2124arp_accept - INTEGER 2125 Define behavior for accepting gratuitous ARP (garp) frames from devices 2126 that are not already present in the ARP table: 2127 2128 - 0 - don't create new entries in the ARP table 2129 - 1 - create new entries in the ARP table 2130 - 2 - create new entries only if the source IP address is in the same 2131 subnet as an address configured on the interface that received the 2132 garp message. 2133 2134 Both replies and requests type gratuitous arp will trigger the 2135 ARP table to be updated, if this setting is on. 2136 2137 If the ARP table already contains the IP address of the 2138 gratuitous arp frame, the arp table will be updated regardless 2139 if this setting is on or off. 2140 2141arp_evict_nocarrier - BOOLEAN 2142 Clears the ARP cache on NOCARRIER events. This option is important for 2143 wireless devices where the ARP cache should not be cleared when roaming 2144 between access points on the same network. In most cases this should 2145 remain as the default (1). 2146 2147 Possible values: 2148 2149 - 0 (disabled) - Do not clear ARP cache on NOCARRIER events 2150 - 1 (enabled) - Clear the ARP cache on NOCARRIER events 2151 2152 Default: 1 (enabled) 2153 2154mcast_solicit - INTEGER 2155 The maximum number of multicast probes in INCOMPLETE state, 2156 when the associated hardware address is unknown. Defaults 2157 to 3. 2158 2159ucast_solicit - INTEGER 2160 The maximum number of unicast probes in PROBE state, when 2161 the hardware address is being reconfirmed. Defaults to 3. 2162 2163app_solicit - INTEGER 2164 The maximum number of probes to send to the user space ARP daemon 2165 via netlink before dropping back to multicast probes (see 2166 mcast_resolicit). Defaults to 0. 2167 2168mcast_resolicit - INTEGER 2169 The maximum number of multicast probes after unicast and 2170 app probes in PROBE state. Defaults to 0. 2171 2172disable_policy - BOOLEAN 2173 Disable IPSEC policy (SPD) for this interface 2174 2175 Possible values: 2176 2177 - 0 (disabled) 2178 - 1 (enabled) 2179 2180 Default: 0 (disabled) 2181 2182disable_xfrm - BOOLEAN 2183 Disable IPSEC encryption on this interface, whatever the policy 2184 2185 Possible values: 2186 2187 - 0 (disabled) 2188 - 1 (enabled) 2189 2190 Default: 0 (disabled) 2191 2192igmpv2_unsolicited_report_interval - INTEGER 2193 The interval in milliseconds in which the next unsolicited 2194 IGMPv1 or IGMPv2 report retransmit will take place. 2195 2196 Default: 10000 (10 seconds) 2197 2198igmpv3_unsolicited_report_interval - INTEGER 2199 The interval in milliseconds in which the next unsolicited 2200 IGMPv3 report retransmit will take place. 2201 2202 Default: 1000 (1 seconds) 2203 2204ignore_routes_with_linkdown - BOOLEAN 2205 Ignore routes whose link is down when performing a FIB lookup. 2206 2207 Possible values: 2208 2209 - 0 (disabled) 2210 - 1 (enabled) 2211 2212 Default: 0 (disabled) 2213 2214promote_secondaries - BOOLEAN 2215 When a primary IP address is removed from this interface 2216 promote a corresponding secondary IP address instead of 2217 removing all the corresponding secondary IP addresses. 2218 2219 Possible values: 2220 2221 - 0 (disabled) 2222 - 1 (enabled) 2223 2224 Default: 0 (disabled) 2225 2226drop_unicast_in_l2_multicast - BOOLEAN 2227 Drop any unicast IP packets that are received in link-layer 2228 multicast (or broadcast) frames. 2229 2230 This behavior (for multicast) is actually a SHOULD in RFC 2231 1122, but is disabled by default for compatibility reasons. 2232 2233 Possible values: 2234 2235 - 0 (disabled) 2236 - 1 (enabled) 2237 2238 Default: 0 (disabled) 2239 2240drop_gratuitous_arp - BOOLEAN 2241 Drop all gratuitous ARP frames, for example if there's a known 2242 good ARP proxy on the network and such frames need not be used 2243 (or in the case of 802.11, must not be used to prevent attacks.) 2244 2245 Possible values: 2246 2247 - 0 (disabled) 2248 - 1 (enabled) 2249 2250 Default: 0 (disabled) 2251 2252 2253tag - INTEGER 2254 Allows you to write a number, which can be used as required. 2255 2256 Default value is 0. 2257 2258xfrm4_gc_thresh - INTEGER 2259 (Obsolete since linux-4.14) 2260 The threshold at which we will start garbage collecting for IPv4 2261 destination cache entries. At twice this value the system will 2262 refuse new allocations. 2263 2264igmp_link_local_mcast_reports - BOOLEAN 2265 Enable IGMP reports for link local multicast groups in the 2266 224.0.0.X range. 2267 2268 Default TRUE 2269 2270Alexey Kuznetsov. 2271kuznet@ms2.inr.ac.ru 2272 2273Updated by: 2274 2275- Andi Kleen 2276 ak@muc.de 2277- Nicolas Delon 2278 delon.nicolas@wanadoo.fr 2279 2280 2281 2282 2283/proc/sys/net/ipv6/* Variables 2284============================== 2285 2286IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also 2287apply to IPv6 [XXX?]. 2288 2289bindv6only - BOOLEAN 2290 Default value for IPV6_V6ONLY socket option, 2291 which restricts use of the IPv6 socket to IPv6 communication 2292 only. 2293 2294 Possible values: 2295 2296 - 0 (disabled) - enable IPv4-mapped address feature 2297 - 1 (enabled) - disable IPv4-mapped address feature 2298 2299 Default: 0 (disabled) 2300 2301flowlabel_consistency - BOOLEAN 2302 Protect the consistency (and unicity) of flow label. 2303 You have to disable it to use IPV6_FL_F_REFLECT flag on the 2304 flow label manager. 2305 2306 Possible values: 2307 2308 - 0 (disabled) 2309 - 1 (enabled) 2310 2311 Default: 1 (enabled) 2312 2313auto_flowlabels - INTEGER 2314 Automatically generate flow labels based on a flow hash of the 2315 packet. This allows intermediate devices, such as routers, to 2316 identify packet flows for mechanisms like Equal Cost Multipath 2317 Routing (see RFC 6438). 2318 2319 = =========================================================== 2320 0 automatic flow labels are completely disabled 2321 1 automatic flow labels are enabled by default, they can be 2322 disabled on a per socket basis using the IPV6_AUTOFLOWLABEL 2323 socket option 2324 2 automatic flow labels are allowed, they may be enabled on a 2325 per socket basis using the IPV6_AUTOFLOWLABEL socket option 2326 3 automatic flow labels are enabled and enforced, they cannot 2327 be disabled by the socket option 2328 = =========================================================== 2329 2330 Default: 1 2331 2332flowlabel_state_ranges - BOOLEAN 2333 Split the flow label number space into two ranges. 0-0x7FFFF is 2334 reserved for the IPv6 flow manager facility, 0x80000-0xFFFFF 2335 is reserved for stateless flow labels as described in RFC6437. 2336 2337 Possible values: 2338 2339 - 0 (disabled) 2340 - 1 (enabled) 2341 2342 Default: 1 (enabled) 2343 2344 2345flowlabel_reflect - INTEGER 2346 Control flow label reflection. Needed for Path MTU 2347 Discovery to work with Equal Cost Multipath Routing in anycast 2348 environments. See RFC 7690 and: 2349 https://tools.ietf.org/html/draft-wang-6man-flow-label-reflection-01 2350 2351 This is a bitmask. 2352 2353 - 1: enabled for established flows 2354 2355 Note that this prevents automatic flowlabel changes, as done 2356 in "tcp: change IPv6 flow-label upon receiving spurious retransmission" 2357 and "tcp: Change txhash on every SYN and RTO retransmit" 2358 2359 - 2: enabled for TCP RESET packets (no active listener) 2360 If set, a RST packet sent in response to a SYN packet on a closed 2361 port will reflect the incoming flow label. 2362 2363 - 4: enabled for ICMPv6 echo reply messages. 2364 2365 Default: 0 2366 2367fib_multipath_hash_policy - INTEGER 2368 Controls which hash policy to use for multipath routes. 2369 2370 Default: 0 (Layer 3) 2371 2372 Possible values: 2373 2374 - 0 - Layer 3 (source and destination addresses plus flow label) 2375 - 1 - Layer 4 (standard 5-tuple) 2376 - 2 - Layer 3 or inner Layer 3 if present 2377 - 3 - Custom multipath hash. Fields used for multipath hash calculation 2378 are determined by fib_multipath_hash_fields sysctl 2379 2380fib_multipath_hash_fields - UNSIGNED INTEGER 2381 When fib_multipath_hash_policy is set to 3 (custom multipath hash), the 2382 fields used for multipath hash calculation are determined by this 2383 sysctl. 2384 2385 This value is a bitmask which enables various fields for multipath hash 2386 calculation. 2387 2388 Possible fields are: 2389 2390 ====== ============================ 2391 0x0001 Source IP address 2392 0x0002 Destination IP address 2393 0x0004 IP protocol 2394 0x0008 Flow Label 2395 0x0010 Source port 2396 0x0020 Destination port 2397 0x0040 Inner source IP address 2398 0x0080 Inner destination IP address 2399 0x0100 Inner IP protocol 2400 0x0200 Inner Flow Label 2401 0x0400 Inner source port 2402 0x0800 Inner destination port 2403 ====== ============================ 2404 2405 Default: 0x0007 (source IP, destination IP and IP protocol) 2406 2407anycast_src_echo_reply - BOOLEAN 2408 Controls the use of anycast addresses as source addresses for ICMPv6 2409 echo reply 2410 2411 Possible values: 2412 2413 - 0 (disabled) 2414 - 1 (enabled) 2415 2416 Default: 0 (disabled) 2417 2418 2419idgen_delay - INTEGER 2420 Controls the delay in seconds after which time to retry 2421 privacy stable address generation if a DAD conflict is 2422 detected. 2423 2424 Default: 1 (as specified in RFC7217) 2425 2426idgen_retries - INTEGER 2427 Controls the number of retries to generate a stable privacy 2428 address if a DAD conflict is detected. 2429 2430 Default: 3 (as specified in RFC7217) 2431 2432mld_qrv - INTEGER 2433 Controls the MLD query robustness variable (see RFC3810 9.1). 2434 2435 Default: 2 (as specified by RFC3810 9.1) 2436 2437 Minimum: 1 (as specified by RFC6636 4.5) 2438 2439max_dst_opts_number - INTEGER 2440 Maximum number of non-padding TLVs allowed in a Destination 2441 options extension header. If this value is less than zero 2442 then unknown options are disallowed and the number of known 2443 TLVs allowed is the absolute value of this number. 2444 2445 Default: 8 2446 2447max_hbh_opts_number - INTEGER 2448 Maximum number of non-padding TLVs allowed in a Hop-by-Hop 2449 options extension header. If this value is less than zero 2450 then unknown options are disallowed and the number of known 2451 TLVs allowed is the absolute value of this number. 2452 2453 Default: 8 2454 2455max_dst_opts_length - INTEGER 2456 Maximum length allowed for a Destination options extension 2457 header. 2458 2459 Default: INT_MAX (unlimited) 2460 2461max_hbh_length - INTEGER 2462 Maximum length allowed for a Hop-by-Hop options extension 2463 header. 2464 2465 Default: INT_MAX (unlimited) 2466 2467skip_notify_on_dev_down - BOOLEAN 2468 Controls whether an RTM_DELROUTE message is generated for routes 2469 removed when a device is taken down or deleted. IPv4 does not 2470 generate this message; IPv6 does by default. Setting this sysctl 2471 to true skips the message, making IPv4 and IPv6 on par in relying 2472 on userspace caches to track link events and evict routes. 2473 2474 Possible values: 2475 2476 - 0 (disabled) - generate the message 2477 - 1 (enabled) - skip generating the message 2478 2479 Default: 0 (disabled) 2480 2481nexthop_compat_mode - BOOLEAN 2482 New nexthop API provides a means for managing nexthops independent of 2483 prefixes. Backwards compatibility with old route format is enabled by 2484 default which means route dumps and notifications contain the new 2485 nexthop attribute but also the full, expanded nexthop definition. 2486 Further, updates or deletes of a nexthop configuration generate route 2487 notifications for each fib entry using the nexthop. Once a system 2488 understands the new API, this sysctl can be disabled to achieve full 2489 performance benefits of the new API by disabling the nexthop expansion 2490 and extraneous notifications. 2491 2492 Note that as a backward-compatible mode, dumping of modern features 2493 might be incomplete or wrong. For example, resilient groups will not be 2494 shown as such, but rather as just a list of next hops. Also weights that 2495 do not fit into 8 bits will show incorrectly. 2496 2497 Default: true (backward compat mode) 2498 2499fib_notify_on_flag_change - INTEGER 2500 Whether to emit RTM_NEWROUTE notifications whenever RTM_F_OFFLOAD/ 2501 RTM_F_TRAP/RTM_F_OFFLOAD_FAILED flags are changed. 2502 2503 After installing a route to the kernel, user space receives an 2504 acknowledgment, which means the route was installed in the kernel, 2505 but not necessarily in hardware. 2506 It is also possible for a route already installed in hardware to change 2507 its action and therefore its flags. For example, a host route that is 2508 trapping packets can be "promoted" to perform decapsulation following 2509 the installation of an IPinIP/VXLAN tunnel. 2510 The notifications will indicate to user-space the state of the route. 2511 2512 Default: 0 (Do not emit notifications.) 2513 2514 Possible values: 2515 2516 - 0 - Do not emit notifications. 2517 - 1 - Emit notifications. 2518 - 2 - Emit notifications only for RTM_F_OFFLOAD_FAILED flag change. 2519 2520ioam6_id - INTEGER 2521 Define the IOAM id of this node. Uses only 24 bits out of 32 in total. 2522 2523 Possible value range: 2524 2525 - Min: 0 2526 - Max: 0xFFFFFF 2527 2528 Default: 0xFFFFFF 2529 2530ioam6_id_wide - LONG INTEGER 2531 Define the wide IOAM id of this node. Uses only 56 bits out of 64 in 2532 total. Can be different from ioam6_id. 2533 2534 Possible value range: 2535 2536 - Min: 0 2537 - Max: 0xFFFFFFFFFFFFFF 2538 2539 Default: 0xFFFFFFFFFFFFFF 2540 2541IPv6 Fragmentation: 2542 2543ip6frag_high_thresh - INTEGER 2544 Maximum memory used to reassemble IPv6 fragments. When 2545 ip6frag_high_thresh bytes of memory is allocated for this purpose, 2546 the fragment handler will toss packets until ip6frag_low_thresh 2547 is reached. 2548 2549ip6frag_low_thresh - INTEGER 2550 See ip6frag_high_thresh 2551 2552ip6frag_time - INTEGER 2553 Time in seconds to keep an IPv6 fragment in memory. 2554 2555``conf/default/*``: 2556 Change the interface-specific default settings. 2557 2558 These settings would be used during creating new interfaces. 2559 2560 2561``conf/all/*``: 2562 Change all the interface-specific settings. 2563 2564 [XXX: Other special features than forwarding?] 2565 2566conf/all/disable_ipv6 - BOOLEAN 2567 Changing this value is same as changing ``conf/default/disable_ipv6`` 2568 setting and also all per-interface ``disable_ipv6`` settings to the same 2569 value. 2570 2571 Reading this value does not have any particular meaning. It does not say 2572 whether IPv6 support is enabled or disabled. Returned value can be 1 2573 also in the case when some interface has ``disable_ipv6`` set to 0 and 2574 has configured IPv6 addresses. 2575 2576conf/all/forwarding - BOOLEAN 2577 Enable global IPv6 forwarding between all interfaces. 2578 2579 IPv4 and IPv6 work differently here; the ``force_forwarding`` flag must 2580 be used to control which interfaces may forward packets. 2581 2582 This also sets all interfaces' Host/Router setting 2583 'forwarding' to the specified value. See below for details. 2584 2585 This referred to as global forwarding. 2586 2587proxy_ndp - BOOLEAN 2588 Do proxy ndp. 2589 2590 Possible values: 2591 2592 - 0 (disabled) 2593 - 1 (enabled) 2594 2595 Default: 0 (disabled) 2596 2597force_forwarding - BOOLEAN 2598 Enable forwarding on this interface only -- regardless of the setting on 2599 ``conf/all/forwarding``. When setting ``conf.all.forwarding`` to 0, 2600 the ``force_forwarding`` flag will be reset on all interfaces. 2601 2602fwmark_reflect - BOOLEAN 2603 Controls the fwmark of kernel-generated IPv6 reply packets that are not 2604 associated with a socket for example, TCP RSTs or ICMPv6 echo replies). 2605 If disabled, these packets have a fwmark of zero. If enabled, they have the 2606 fwmark of the packet they are replying to. 2607 2608 Possible values: 2609 2610 - 0 (disabled) 2611 - 1 (enabled) 2612 2613 Default: 0 (disabled) 2614 2615``conf/interface/*``: 2616 Change special settings per interface. 2617 2618 The functional behaviour for certain settings is different 2619 depending on whether local forwarding is enabled or not. 2620 2621accept_ra - INTEGER 2622 Accept Router Advertisements; autoconfigure using them. 2623 2624 It also determines whether or not to transmit Router 2625 Solicitations. If and only if the functional setting is to 2626 accept Router Advertisements, Router Solicitations will be 2627 transmitted. 2628 2629 Possible values are: 2630 2631 == =========================================================== 2632 0 Do not accept Router Advertisements. 2633 1 Accept Router Advertisements if forwarding is disabled. 2634 2 Overrule forwarding behaviour. Accept Router Advertisements 2635 even if forwarding is enabled. 2636 == =========================================================== 2637 2638 Functional default: 2639 2640 - enabled if local forwarding is disabled. 2641 - disabled if local forwarding is enabled. 2642 2643accept_ra_defrtr - BOOLEAN 2644 Learn default router in Router Advertisement. 2645 2646 Functional default: 2647 2648 - enabled if accept_ra is enabled. 2649 - disabled if accept_ra is disabled. 2650 2651ra_defrtr_metric - UNSIGNED INTEGER 2652 Route metric for default route learned in Router Advertisement. This value 2653 will be assigned as metric for the default route learned via IPv6 Router 2654 Advertisement. Takes affect only if accept_ra_defrtr is enabled. 2655 2656 Possible values: 2657 1 to 0xFFFFFFFF 2658 2659 Default: IP6_RT_PRIO_USER i.e. 1024. 2660 2661accept_ra_from_local - BOOLEAN 2662 Accept RA with source-address that is found on local machine 2663 if the RA is otherwise proper and able to be accepted. 2664 2665 Default is to NOT accept these as it may be an un-intended 2666 network loop. 2667 2668 Functional default: 2669 2670 - enabled if accept_ra_from_local is enabled 2671 on a specific interface. 2672 - disabled if accept_ra_from_local is disabled 2673 on a specific interface. 2674 2675accept_ra_min_hop_limit - INTEGER 2676 Minimum hop limit Information in Router Advertisement. 2677 2678 Hop limit Information in Router Advertisement less than this 2679 variable shall be ignored. 2680 2681 Default: 1 2682 2683accept_ra_min_lft - INTEGER 2684 Minimum acceptable lifetime value in Router Advertisement. 2685 2686 RA sections with a lifetime less than this value shall be 2687 ignored. Zero lifetimes stay unaffected. 2688 2689 Default: 0 2690 2691accept_ra_pinfo - BOOLEAN 2692 Learn Prefix Information in Router Advertisement. 2693 2694 Functional default: 2695 2696 - enabled if accept_ra is enabled. 2697 - disabled if accept_ra is disabled. 2698 2699ra_honor_pio_life - BOOLEAN 2700 Whether to use RFC4862 Section 5.5.3e to determine the valid 2701 lifetime of an address matching a prefix sent in a Router 2702 Advertisement Prefix Information Option. 2703 2704 Possible values: 2705 2706 - 0 (disabled) - RFC4862 section 5.5.3e is used to determine 2707 the valid lifetime of the address. 2708 - 1 (enabled) - the PIO valid lifetime will always be honored. 2709 2710 Default: 0 (disabled) 2711 2712ra_honor_pio_pflag - BOOLEAN 2713 The Prefix Information Option P-flag indicates the network can 2714 allocate a unique IPv6 prefix per client using DHCPv6-PD. 2715 This sysctl can be enabled when a userspace DHCPv6-PD client 2716 is running to cause the P-flag to take effect: i.e. the 2717 P-flag suppresses any effects of the A-flag within the same 2718 PIO. For a given PIO, P=1 and A=1 is treated as A=0. 2719 2720 Possible values: 2721 2722 - 0 (disabled) - the P-flag is ignored. 2723 - 1 (enabled) - the P-flag will disable SLAAC autoconfiguration 2724 for the given Prefix Information Option. 2725 2726 Default: 0 (disabled) 2727 2728accept_ra_rt_info_min_plen - INTEGER 2729 Minimum prefix length of Route Information in RA. 2730 2731 Route Information w/ prefix smaller than this variable shall 2732 be ignored. 2733 2734 Functional default: 2735 2736 * 0 if accept_ra_rtr_pref is enabled. 2737 * -1 if accept_ra_rtr_pref is disabled. 2738 2739accept_ra_rt_info_max_plen - INTEGER 2740 Maximum prefix length of Route Information in RA. 2741 2742 Route Information w/ prefix larger than this variable shall 2743 be ignored. 2744 2745 Functional default: 2746 2747 * 0 if accept_ra_rtr_pref is enabled. 2748 * -1 if accept_ra_rtr_pref is disabled. 2749 2750accept_ra_rtr_pref - BOOLEAN 2751 Accept Router Preference in RA. 2752 2753 Functional default: 2754 2755 - enabled if accept_ra is enabled. 2756 - disabled if accept_ra is disabled. 2757 2758accept_ra_mtu - BOOLEAN 2759 Apply the MTU value specified in RA option 5 (RFC4861). If 2760 disabled, the MTU specified in the RA will be ignored. 2761 2762 Functional default: 2763 2764 - enabled if accept_ra is enabled. 2765 - disabled if accept_ra is disabled. 2766 2767accept_redirects - BOOLEAN 2768 Accept Redirects. 2769 2770 Functional default: 2771 2772 - enabled if local forwarding is disabled. 2773 - disabled if local forwarding is enabled. 2774 2775accept_source_route - INTEGER 2776 Accept source routing (routing extension header). 2777 2778 - >= 0: Accept only routing header type 2. 2779 - < 0: Do not accept routing header. 2780 2781 Default: 0 2782 2783autoconf - BOOLEAN 2784 Autoconfigure addresses using Prefix Information in Router 2785 Advertisements. 2786 2787 Functional default: 2788 2789 - enabled if accept_ra_pinfo is enabled. 2790 - disabled if accept_ra_pinfo is disabled. 2791 2792dad_transmits - INTEGER 2793 The amount of Duplicate Address Detection probes to send. 2794 2795 Default: 1 2796 2797forwarding - INTEGER 2798 Configure interface-specific Host/Router behaviour. 2799 2800 .. note:: 2801 2802 It is recommended to have the same setting on all 2803 interfaces; mixed router/host scenarios are rather uncommon. 2804 2805 Possible values are: 2806 2807 - 0 Forwarding disabled 2808 - 1 Forwarding enabled 2809 2810 **FALSE (0)**: 2811 2812 By default, Host behaviour is assumed. This means: 2813 2814 1. IsRouter flag is not set in Neighbour Advertisements. 2815 2. If accept_ra is TRUE (default), transmit Router 2816 Solicitations. 2817 3. If accept_ra is TRUE (default), accept Router 2818 Advertisements (and do autoconfiguration). 2819 4. If accept_redirects is TRUE (default), accept Redirects. 2820 2821 **TRUE (1)**: 2822 2823 If local forwarding is enabled, Router behaviour is assumed. 2824 This means exactly the reverse from the above: 2825 2826 1. IsRouter flag is set in Neighbour Advertisements. 2827 2. Router Solicitations are not sent unless accept_ra is 2. 2828 3. Router Advertisements are ignored unless accept_ra is 2. 2829 4. Redirects are ignored. 2830 2831 Default: 0 (disabled) if global forwarding is disabled (default), 2832 otherwise 1 (enabled). 2833 2834hop_limit - INTEGER 2835 Default Hop Limit to set. 2836 2837 Default: 64 2838 2839mtu - INTEGER 2840 Default Maximum Transfer Unit 2841 2842 Default: 1280 (IPv6 required minimum) 2843 2844ip_nonlocal_bind - BOOLEAN 2845 If enabled, allows processes to bind() to non-local IPv6 addresses, 2846 which can be quite useful - but may break some applications. 2847 2848 Possible values: 2849 2850 - 0 (disabled) 2851 - 1 (enabled) 2852 2853 Default: 0 (disabled) 2854 2855router_probe_interval - INTEGER 2856 Minimum interval (in seconds) between Router Probing described 2857 in RFC4191. 2858 2859 Default: 60 2860 2861router_solicitation_delay - INTEGER 2862 Number of seconds to wait after interface is brought up 2863 before sending Router Solicitations. 2864 2865 Default: 1 2866 2867router_solicitation_interval - INTEGER 2868 Number of seconds to wait between Router Solicitations. 2869 2870 Default: 4 2871 2872router_solicitations - INTEGER 2873 Number of Router Solicitations to send until assuming no 2874 routers are present. 2875 2876 Default: 3 2877 2878use_oif_addrs_only - BOOLEAN 2879 When enabled, the candidate source addresses for destinations 2880 routed via this interface are restricted to the set of addresses 2881 configured on this interface (vis. RFC 6724, section 4). 2882 2883 Possible values: 2884 2885 - 0 (disabled) 2886 - 1 (enabled) 2887 2888 Default: 0 (disabled) 2889 2890use_tempaddr - INTEGER 2891 Preference for Privacy Extensions (RFC3041). 2892 2893 * <= 0 : disable Privacy Extensions 2894 * == 1 : enable Privacy Extensions, but prefer public 2895 addresses over temporary addresses. 2896 * > 1 : enable Privacy Extensions and prefer temporary 2897 addresses over public addresses. 2898 2899 Default: 2900 2901 * 0 (for most devices) 2902 * -1 (for point-to-point devices and loopback devices) 2903 2904temp_valid_lft - INTEGER 2905 valid lifetime (in seconds) for temporary addresses. If less than the 2906 minimum required lifetime (typically 5-7 seconds), temporary addresses 2907 will not be created. 2908 2909 Default: 172800 (2 days) 2910 2911temp_prefered_lft - INTEGER 2912 Preferred lifetime (in seconds) for temporary addresses. If 2913 temp_prefered_lft is less than the minimum required lifetime (typically 2914 5-7 seconds), the preferred lifetime is the minimum required. If 2915 temp_prefered_lft is greater than temp_valid_lft, the preferred lifetime 2916 is temp_valid_lft. 2917 2918 Default: 86400 (1 day) 2919 2920keep_addr_on_down - INTEGER 2921 Keep all IPv6 addresses on an interface down event. If set static 2922 global addresses with no expiration time are not flushed. 2923 2924 * >0 : enabled 2925 * 0 : system default 2926 * <0 : disabled 2927 2928 Default: 0 (addresses are removed) 2929 2930max_desync_factor - INTEGER 2931 Maximum value for DESYNC_FACTOR, which is a random value 2932 that ensures that clients don't synchronize with each 2933 other and generate new addresses at exactly the same time. 2934 value is in seconds. 2935 2936 Default: 600 2937 2938regen_min_advance - INTEGER 2939 How far in advance (in seconds), at minimum, to create a new temporary 2940 address before the current one is deprecated. This value is added to 2941 the amount of time that may be required for duplicate address detection 2942 to determine when to create a new address. Linux permits setting this 2943 value to less than the default of 2 seconds, but a value less than 2 2944 does not conform to RFC 8981. 2945 2946 Default: 2 2947 2948regen_max_retry - INTEGER 2949 Number of attempts before give up attempting to generate 2950 valid temporary addresses. 2951 2952 Default: 5 2953 2954max_addresses - INTEGER 2955 Maximum number of autoconfigured addresses per interface. Setting 2956 to zero disables the limitation. It is not recommended to set this 2957 value too large (or to zero) because it would be an easy way to 2958 crash the kernel by allowing too many addresses to be created. 2959 2960 Default: 16 2961 2962disable_ipv6 - BOOLEAN 2963 Disable IPv6 operation. If accept_dad is set to 2, this value 2964 will be dynamically set to TRUE if DAD fails for the link-local 2965 address. 2966 2967 Default: FALSE (enable IPv6 operation) 2968 2969 When this value is changed from 1 to 0 (IPv6 is being enabled), 2970 it will dynamically create a link-local address on the given 2971 interface and start Duplicate Address Detection, if necessary. 2972 2973 When this value is changed from 0 to 1 (IPv6 is being disabled), 2974 it will dynamically delete all addresses and routes on the given 2975 interface. From now on it will not possible to add addresses/routes 2976 to the selected interface. 2977 2978accept_dad - INTEGER 2979 Whether to accept DAD (Duplicate Address Detection). 2980 2981 == ============================================================== 2982 0 Disable DAD 2983 1 Enable DAD (default) 2984 2 Enable DAD, and disable IPv6 operation if MAC-based duplicate 2985 link-local address has been found. 2986 == ============================================================== 2987 2988 DAD operation and mode on a given interface will be selected according 2989 to the maximum value of conf/{all,interface}/accept_dad. 2990 2991force_tllao - BOOLEAN 2992 Enable sending the target link-layer address option even when 2993 responding to a unicast neighbor solicitation. 2994 2995 Default: FALSE 2996 2997 Quoting from RFC 2461, section 4.4, Target link-layer address: 2998 2999 "The option MUST be included for multicast solicitations in order to 3000 avoid infinite Neighbor Solicitation "recursion" when the peer node 3001 does not have a cache entry to return a Neighbor Advertisements 3002 message. When responding to unicast solicitations, the option can be 3003 omitted since the sender of the solicitation has the correct link- 3004 layer address; otherwise it would not have be able to send the unicast 3005 solicitation in the first place. However, including the link-layer 3006 address in this case adds little overhead and eliminates a potential 3007 race condition where the sender deletes the cached link-layer address 3008 prior to receiving a response to a previous solicitation." 3009 3010ndisc_notify - BOOLEAN 3011 Define mode for notification of address and device changes. 3012 3013 Possible values: 3014 3015 - 0 (disabled) - do nothing 3016 - 1 (enabled) - Generate unsolicited neighbour advertisements when device is brought 3017 up or hardware address changes. 3018 3019 Default: 0 (disabled) 3020 3021ndisc_tclass - INTEGER 3022 The IPv6 Traffic Class to use by default when sending IPv6 Neighbor 3023 Discovery (Router Solicitation, Router Advertisement, Neighbor 3024 Solicitation, Neighbor Advertisement, Redirect) messages. 3025 These 8 bits can be interpreted as 6 high order bits holding the DSCP 3026 value and 2 low order bits representing ECN (which you probably want 3027 to leave cleared). 3028 3029 * 0 - (default) 3030 3031ndisc_evict_nocarrier - BOOLEAN 3032 Clears the neighbor discovery table on NOCARRIER events. This option is 3033 important for wireless devices where the neighbor discovery cache should 3034 not be cleared when roaming between access points on the same network. 3035 In most cases this should remain as the default (1). 3036 3037 Possible values: 3038 3039 - 0 (disabled) - Do not clear neighbor discovery cache on NOCARRIER events. 3040 - 1 (enabled) - Clear neighbor discover cache on NOCARRIER events. 3041 3042 Default: 1 (enabled) 3043 3044mldv1_unsolicited_report_interval - INTEGER 3045 The interval in milliseconds in which the next unsolicited 3046 MLDv1 report retransmit will take place. 3047 3048 Default: 10000 (10 seconds) 3049 3050mldv2_unsolicited_report_interval - INTEGER 3051 The interval in milliseconds in which the next unsolicited 3052 MLDv2 report retransmit will take place. 3053 3054 Default: 1000 (1 second) 3055 3056force_mld_version - INTEGER 3057 * 0 - (default) No enforcement of a MLD version, MLDv1 fallback allowed 3058 * 1 - Enforce to use MLD version 1 3059 * 2 - Enforce to use MLD version 2 3060 3061suppress_frag_ndisc - INTEGER 3062 Control RFC 6980 (Security Implications of IPv6 Fragmentation 3063 with IPv6 Neighbor Discovery) behavior: 3064 3065 * 1 - (default) discard fragmented neighbor discovery packets 3066 * 0 - allow fragmented neighbor discovery packets 3067 3068optimistic_dad - BOOLEAN 3069 Whether to perform Optimistic Duplicate Address Detection (RFC 4429). 3070 3071 Optimistic Duplicate Address Detection for the interface will be enabled 3072 if at least one of conf/{all,interface}/optimistic_dad is set to 1, 3073 it will be disabled otherwise. 3074 3075 Possible values: 3076 3077 - 0 (disabled) 3078 - 1 (enabled) 3079 3080 Default: 0 (disabled) 3081 3082 3083use_optimistic - BOOLEAN 3084 If enabled, do not classify optimistic addresses as deprecated during 3085 source address selection. Preferred addresses will still be chosen 3086 before optimistic addresses, subject to other ranking in the source 3087 address selection algorithm. 3088 3089 This will be enabled if at least one of 3090 conf/{all,interface}/use_optimistic is set to 1, disabled otherwise. 3091 3092 Possible values: 3093 3094 - 0 (disabled) 3095 - 1 (enabled) 3096 3097 Default: 0 (disabled) 3098 3099stable_secret - IPv6 address 3100 This IPv6 address will be used as a secret to generate IPv6 3101 addresses for link-local addresses and autoconfigured 3102 ones. All addresses generated after setting this secret will 3103 be stable privacy ones by default. This can be changed via the 3104 addrgenmode ip-link. conf/default/stable_secret is used as the 3105 secret for the namespace, the interface specific ones can 3106 overwrite that. Writes to conf/all/stable_secret are refused. 3107 3108 It is recommended to generate this secret during installation 3109 of a system and keep it stable after that. 3110 3111 By default the stable secret is unset. 3112 3113addr_gen_mode - INTEGER 3114 Defines how link-local and autoconf addresses are generated. 3115 3116 = ================================================================= 3117 0 generate address based on EUI64 (default) 3118 1 do no generate a link-local address, use EUI64 for addresses 3119 generated from autoconf 3120 2 generate stable privacy addresses, using the secret from 3121 stable_secret (RFC7217) 3122 3 generate stable privacy addresses, using a random secret if unset 3123 = ================================================================= 3124 3125drop_unicast_in_l2_multicast - BOOLEAN 3126 Drop any unicast IPv6 packets that are received in link-layer 3127 multicast (or broadcast) frames. 3128 3129 Possible values: 3130 3131 - 0 (disabled) 3132 - 1 (enabled) 3133 3134 Default: 0 (disabled) 3135 3136drop_unsolicited_na - BOOLEAN 3137 Drop all unsolicited neighbor advertisements, for example if there's 3138 a known good NA proxy on the network and such frames need not be used 3139 (or in the case of 802.11, must not be used to prevent attacks.) 3140 3141 Possible values: 3142 3143 - 0 (disabled) 3144 - 1 (enabled) 3145 3146 Default: 0 (disabled). 3147 3148accept_untracked_na - INTEGER 3149 Define behavior for accepting neighbor advertisements from devices that 3150 are absent in the neighbor cache: 3151 3152 - 0 - (default) Do not accept unsolicited and untracked neighbor 3153 advertisements. 3154 3155 - 1 - Add a new neighbor cache entry in STALE state for routers on 3156 receiving a neighbor advertisement (either solicited or unsolicited) 3157 with target link-layer address option specified if no neighbor entry 3158 is already present for the advertised IPv6 address. Without this knob, 3159 NAs received for untracked addresses (absent in neighbor cache) are 3160 silently ignored. 3161 3162 This is as per router-side behavior documented in RFC9131. 3163 3164 This has lower precedence than drop_unsolicited_na. 3165 3166 This will optimize the return path for the initial off-link 3167 communication that is initiated by a directly connected host, by 3168 ensuring that the first-hop router which turns on this setting doesn't 3169 have to buffer the initial return packets to do neighbor-solicitation. 3170 The prerequisite is that the host is configured to send unsolicited 3171 neighbor advertisements on interface bringup. This setting should be 3172 used in conjunction with the ndisc_notify setting on the host to 3173 satisfy this prerequisite. 3174 3175 - 2 - Extend option (1) to add a new neighbor cache entry only if the 3176 source IP address is in the same subnet as an address configured on 3177 the interface that received the neighbor advertisement. 3178 3179enhanced_dad - BOOLEAN 3180 Include a nonce option in the IPv6 neighbor solicitation messages used for 3181 duplicate address detection per RFC7527. A received DAD NS will only signal 3182 a duplicate address if the nonce is different. This avoids any false 3183 detection of duplicates due to loopback of the NS messages that we send. 3184 The nonce option will be sent on an interface unless both of 3185 conf/{all,interface}/enhanced_dad are set to FALSE. 3186 3187 Possible values: 3188 3189 - 0 (disabled) 3190 - 1 (enabled) 3191 3192 Default: 1 (enabled) 3193 3194``icmp/*``: 3195=========== 3196 3197ratelimit - INTEGER 3198 Limit the maximal rates for sending ICMPv6 messages. 3199 3200 0 to disable any limiting, 3201 otherwise the minimal space between responses in milliseconds. 3202 3203 Default: 1000 3204 3205ratemask - list of comma separated ranges 3206 For ICMPv6 message types matching the ranges in the ratemask, limit 3207 the sending of the message according to ratelimit parameter. 3208 3209 The format used for both input and output is a comma separated 3210 list of ranges (e.g. "0-127,129" for ICMPv6 message type 0 to 127 and 3211 129). Writing to the file will clear all previous ranges of ICMPv6 3212 message types and update the current list with the input. 3213 3214 Refer to: https://www.iana.org/assignments/icmpv6-parameters/icmpv6-parameters.xhtml 3215 for numerical values of ICMPv6 message types, e.g. echo request is 128 3216 and echo reply is 129. 3217 3218 Default: 0-1,3-127 (rate limit ICMPv6 errors except Packet Too Big) 3219 3220echo_ignore_all - BOOLEAN 3221 If enabled, then the kernel will ignore all ICMP ECHO 3222 requests sent to it over the IPv6 protocol. 3223 3224 Possible values: 3225 3226 - 0 (disabled) 3227 - 1 (enabled) 3228 3229 Default: 0 (disabled) 3230 3231echo_ignore_multicast - BOOLEAN 3232 If enabled, then the kernel will ignore all ICMP ECHO 3233 requests sent to it over the IPv6 protocol via multicast. 3234 3235 Possible values: 3236 3237 - 0 (disabled) 3238 - 1 (enabled) 3239 3240 Default: 0 (disabled) 3241 3242echo_ignore_anycast - BOOLEAN 3243 If enabled, then the kernel will ignore all ICMP ECHO 3244 requests sent to it over the IPv6 protocol destined to anycast address. 3245 3246 Possible values: 3247 3248 - 0 (disabled) 3249 - 1 (enabled) 3250 3251 Default: 0 (disabled) 3252 3253error_anycast_as_unicast - BOOLEAN 3254 If enabled, then the kernel will respond with ICMP Errors 3255 resulting from requests sent to it over the IPv6 protocol destined 3256 to anycast address essentially treating anycast as unicast. 3257 3258 Possible values: 3259 3260 - 0 (disabled) 3261 - 1 (enabled) 3262 3263 Default: 0 (disabled) 3264 3265xfrm6_gc_thresh - INTEGER 3266 (Obsolete since linux-4.14) 3267 The threshold at which we will start garbage collecting for IPv6 3268 destination cache entries. At twice this value the system will 3269 refuse new allocations. 3270 3271 3272IPv6 Update by: 3273Pekka Savola <pekkas@netcore.fi> 3274YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org> 3275 3276 3277/proc/sys/net/bridge/* Variables: 3278================================= 3279 3280bridge-nf-call-arptables - BOOLEAN 3281 3282 Possible values: 3283 3284 - 0 (disabled) - disable this. 3285 - 1 (enabled) - pass bridged ARP traffic to arptables' FORWARD chain. 3286 3287 Default: 1 (enabled) 3288 3289bridge-nf-call-iptables - BOOLEAN 3290 3291 Possible values: 3292 3293 - 0 (disabled) - disable this. 3294 - 1 (enabled) - pass bridged IPv4 traffic to iptables' chains. 3295 3296 Default: 1 (enabled) 3297 3298bridge-nf-call-ip6tables - BOOLEAN 3299 3300 Possible values: 3301 3302 - 0 (disabled) - disable this. 3303 - 1 (enabled) - pass bridged IPv6 traffic to ip6tables' chains. 3304 3305 Default: 1 (enabled) 3306 3307bridge-nf-filter-vlan-tagged - BOOLEAN 3308 3309 Possible values: 3310 3311 - 0 (disabled) - disable this. 3312 - 1 (enabled) - pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables 3313 3314 Default: 0 (disabled) 3315 3316bridge-nf-filter-pppoe-tagged - BOOLEAN 3317 3318 Possible values: 3319 3320 - 0 (disabled) - disable this. 3321 - 1 (enabled) - pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables. 3322 3323 Default: 0 (disabled) 3324 3325bridge-nf-pass-vlan-input-dev - BOOLEAN 3326 - 1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan 3327 interface on the bridge and set the netfilter input device to the 3328 vlan. This allows use of e.g. "iptables -i br0.1" and makes the 3329 REDIRECT target work with vlan-on-top-of-bridge interfaces. When no 3330 matching vlan interface is found, or this switch is off, the input 3331 device is set to the bridge interface. 3332 3333 - 0: disable bridge netfilter vlan interface lookup. 3334 3335 Default: 0 3336 3337``proc/sys/net/sctp/*`` Variables: 3338================================== 3339 3340addip_enable - BOOLEAN 3341 Enable or disable extension of Dynamic Address Reconfiguration 3342 (ADD-IP) functionality specified in RFC5061. This extension provides 3343 the ability to dynamically add and remove new addresses for the SCTP 3344 associations. 3345 3346 Possible values: 3347 3348 - 0 (disabled) - disable extension. 3349 - 1 (enabled) - enable extension 3350 3351 Default: 0 (disabled) 3352 3353pf_enable - INTEGER 3354 Enable or disable pf (pf is short for potentially failed) state. A value 3355 of pf_retrans > path_max_retrans also disables pf state. That is, one of 3356 both pf_enable and pf_retrans > path_max_retrans can disable pf state. 3357 Since pf_retrans and path_max_retrans can be changed by userspace 3358 application, sometimes user expects to disable pf state by the value of 3359 pf_retrans > path_max_retrans, but occasionally the value of pf_retrans 3360 or path_max_retrans is changed by the user application, this pf state is 3361 enabled. As such, it is necessary to add this to dynamically enable 3362 and disable pf state. See: 3363 https://datatracker.ietf.org/doc/draft-ietf-tsvwg-sctp-failover for 3364 details. 3365 3366 Possible values: 3367 3368 - 1: Enable pf. 3369 - 0: Disable pf. 3370 3371 Default: 1 3372 3373pf_expose - INTEGER 3374 Unset or enable/disable pf (pf is short for potentially failed) state 3375 exposure. Applications can control the exposure of the PF path state 3376 in the SCTP_PEER_ADDR_CHANGE event and access of SCTP_PF-state 3377 transport info via SCTP_GET_PEER_ADDR_INFO sockopt. 3378 3379 Possible values: 3380 3381 - 0: Unset pf state exposure (compatible with old applications). No 3382 event will be sent but the transport info can be queried. 3383 - 1: Disable pf state exposure. No event will be sent and trying to 3384 obtain transport info will return -EACCESS. 3385 - 2: Enable pf state exposure. The event will be sent for a transport 3386 becoming SCTP_PF state and transport info can be obtained. 3387 3388 Default: 0 3389 3390addip_noauth_enable - BOOLEAN 3391 Dynamic Address Reconfiguration (ADD-IP) requires the use of 3392 authentication to protect the operations of adding or removing new 3393 addresses. This requirement is mandated so that unauthorized hosts 3394 would not be able to hijack associations. However, older 3395 implementations may not have implemented this requirement while 3396 allowing the ADD-IP extension. For reasons of interoperability, 3397 we provide this variable to control the enforcement of the 3398 authentication requirement. 3399 3400 == =============================================================== 3401 1 Allow ADD-IP extension to be used without authentication. This 3402 should only be set in a closed environment for interoperability 3403 with older implementations. 3404 3405 0 Enforce the authentication requirement 3406 == =============================================================== 3407 3408 Default: 0 3409 3410auth_enable - BOOLEAN 3411 Enable or disable Authenticated Chunks extension. This extension 3412 provides the ability to send and receive authenticated chunks and is 3413 required for secure operation of Dynamic Address Reconfiguration 3414 (ADD-IP) extension. 3415 3416 Possible values: 3417 3418 - 0 (disabled) - disable extension. 3419 - 1 (enabled) - enable extension 3420 3421 Default: 0 (disabled) 3422 3423prsctp_enable - BOOLEAN 3424 Enable or disable the Partial Reliability extension (RFC3758) which 3425 is used to notify peers that a given DATA should no longer be expected. 3426 3427 Possible values: 3428 3429 - 0 (disabled) - disable extension. 3430 - 1 (enabled) - enable extension 3431 3432 Default: 1 (enabled) 3433 3434max_burst - INTEGER 3435 The limit of the number of new packets that can be initially sent. It 3436 controls how bursty the generated traffic can be. 3437 3438 Default: 4 3439 3440association_max_retrans - INTEGER 3441 Set the maximum number for retransmissions that an association can 3442 attempt deciding that the remote end is unreachable. If this value 3443 is exceeded, the association is terminated. 3444 3445 Default: 10 3446 3447max_init_retransmits - INTEGER 3448 The maximum number of retransmissions of INIT and COOKIE-ECHO chunks 3449 that an association will attempt before declaring the destination 3450 unreachable and terminating. 3451 3452 Default: 8 3453 3454path_max_retrans - INTEGER 3455 The maximum number of retransmissions that will be attempted on a given 3456 path. Once this threshold is exceeded, the path is considered 3457 unreachable, and new traffic will use a different path when the 3458 association is multihomed. 3459 3460 Default: 5 3461 3462pf_retrans - INTEGER 3463 The number of retransmissions that will be attempted on a given path 3464 before traffic is redirected to an alternate transport (should one 3465 exist). Note this is distinct from path_max_retrans, as a path that 3466 passes the pf_retrans threshold can still be used. Its only 3467 deprioritized when a transmission path is selected by the stack. This 3468 setting is primarily used to enable fast failover mechanisms without 3469 having to reduce path_max_retrans to a very low value. See: 3470 http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt 3471 for details. Note also that a value of pf_retrans > path_max_retrans 3472 disables this feature. Since both pf_retrans and path_max_retrans can 3473 be changed by userspace application, a variable pf_enable is used to 3474 disable pf state. 3475 3476 Default: 0 3477 3478ps_retrans - INTEGER 3479 Primary.Switchover.Max.Retrans (PSMR), it's a tunable parameter coming 3480 from section-5 "Primary Path Switchover" in rfc7829. The primary path 3481 will be changed to another active path when the path error counter on 3482 the old primary path exceeds PSMR, so that "the SCTP sender is allowed 3483 to continue data transmission on a new working path even when the old 3484 primary destination address becomes active again". Note this feature 3485 is disabled by initializing 'ps_retrans' per netns as 0xffff by default, 3486 and its value can't be less than 'pf_retrans' when changing by sysctl. 3487 3488 Default: 0xffff 3489 3490rto_initial - INTEGER 3491 The initial round trip timeout value in milliseconds that will be used 3492 in calculating round trip times. This is the initial time interval 3493 for retransmissions. 3494 3495 Default: 3000 3496 3497rto_max - INTEGER 3498 The maximum value (in milliseconds) of the round trip timeout. This 3499 is the largest time interval that can elapse between retransmissions. 3500 3501 Default: 60000 3502 3503rto_min - INTEGER 3504 The minimum value (in milliseconds) of the round trip timeout. This 3505 is the smallest time interval the can elapse between retransmissions. 3506 3507 Default: 1000 3508 3509hb_interval - INTEGER 3510 The interval (in milliseconds) between HEARTBEAT chunks. These chunks 3511 are sent at the specified interval on idle paths to probe the state of 3512 a given path between 2 associations. 3513 3514 Default: 30000 3515 3516sack_timeout - INTEGER 3517 The amount of time (in milliseconds) that the implementation will wait 3518 to send a SACK. 3519 3520 Default: 200 3521 3522valid_cookie_life - INTEGER 3523 The default lifetime of the SCTP cookie (in milliseconds). The cookie 3524 is used during association establishment. 3525 3526 Default: 60000 3527 3528cookie_preserve_enable - BOOLEAN 3529 Enable or disable the ability to extend the lifetime of the SCTP cookie 3530 that is used during the establishment phase of SCTP association 3531 3532 Possible values: 3533 3534 - 0 (disabled) - disable. 3535 - 1 (enabled) - enable cookie lifetime extension. 3536 3537 Default: 1 (enabled) 3538 3539cookie_hmac_alg - STRING 3540 Select the hmac algorithm used when generating the cookie value sent by 3541 a listening sctp socket to a connecting client in the INIT-ACK chunk. 3542 Valid values are: 3543 3544 * sha256 3545 * none 3546 3547 Default: sha256 3548 3549rcvbuf_policy - INTEGER 3550 Determines if the receive buffer is attributed to the socket or to 3551 association. SCTP supports the capability to create multiple 3552 associations on a single socket. When using this capability, it is 3553 possible that a single stalled association that's buffering a lot 3554 of data may block other associations from delivering their data by 3555 consuming all of the receive buffer space. To work around this, 3556 the rcvbuf_policy could be set to attribute the receiver buffer space 3557 to each association instead of the socket. This prevents the described 3558 blocking. 3559 3560 - 1: rcvbuf space is per association 3561 - 0: rcvbuf space is per socket 3562 3563 Default: 0 3564 3565sndbuf_policy - INTEGER 3566 Similar to rcvbuf_policy above, this applies to send buffer space. 3567 3568 - 1: Send buffer is tracked per association 3569 - 0: Send buffer is tracked per socket. 3570 3571 Default: 0 3572 3573sctp_mem - vector of 3 INTEGERs: min, pressure, max 3574 Number of pages allowed for queueing by all SCTP sockets. 3575 3576 * min: Below this number of pages SCTP is not bothered about its 3577 memory usage. When amount of memory allocated by SCTP exceeds 3578 this number, SCTP starts to moderate memory usage. 3579 * pressure: This value was introduced to follow format of tcp_mem. 3580 * max: Maximum number of allowed pages. 3581 3582 Default is calculated at boot time from amount of available memory. 3583 3584sctp_rmem - vector of 3 INTEGERs: min, default, max 3585 Only the first value ("min") is used, "default" and "max" are 3586 ignored. 3587 3588 * min: Minimal size of receive buffer used by SCTP socket. 3589 It is guaranteed to each SCTP socket (but not association) even 3590 under moderate memory pressure. 3591 3592 Default: 4K 3593 3594sctp_wmem - vector of 3 INTEGERs: min, default, max 3595 Only the first value ("min") is used, "default" and "max" are 3596 ignored. 3597 3598 * min: Minimum size of send buffer that can be used by SCTP sockets. 3599 It is guaranteed to each SCTP socket (but not association) even 3600 under moderate memory pressure. 3601 3602 Default: 4K 3603 3604addr_scope_policy - INTEGER 3605 Control IPv4 address scoping (see 3606 https://datatracker.ietf.org/doc/draft-stewart-tsvwg-sctp-ipv4/00/ 3607 for details). 3608 3609 - 0 - Disable IPv4 address scoping 3610 - 1 - Enable IPv4 address scoping 3611 - 2 - Follow draft but allow IPv4 private addresses 3612 - 3 - Follow draft but allow IPv4 link local addresses 3613 3614 Default: 1 3615 3616udp_port - INTEGER 3617 The listening port for the local UDP tunneling sock. Normally it's 3618 using the IANA-assigned UDP port number 9899 (sctp-tunneling). 3619 3620 This UDP sock is used for processing the incoming UDP-encapsulated 3621 SCTP packets (from RFC6951), and shared by all applications in the 3622 same net namespace. This UDP sock will be closed when the value is 3623 set to 0. 3624 3625 The value will also be used to set the src port of the UDP header 3626 for the outgoing UDP-encapsulated SCTP packets. For the dest port, 3627 please refer to 'encap_port' below. 3628 3629 Default: 0 3630 3631encap_port - INTEGER 3632 The default remote UDP encapsulation port. 3633 3634 This value is used to set the dest port of the UDP header for the 3635 outgoing UDP-encapsulated SCTP packets by default. Users can also 3636 change the value for each sock/asoc/transport by using setsockopt. 3637 For further information, please refer to RFC6951. 3638 3639 Note that when connecting to a remote server, the client should set 3640 this to the port that the UDP tunneling sock on the peer server is 3641 listening to and the local UDP tunneling sock on the client also 3642 must be started. On the server, it would get the encap_port from 3643 the incoming packet's source port. 3644 3645 Default: 0 3646 3647plpmtud_probe_interval - INTEGER 3648 The time interval (in milliseconds) for the PLPMTUD probe timer, 3649 which is configured to expire after this period to receive an 3650 acknowledgment to a probe packet. This is also the time interval 3651 between the probes for the current pmtu when the probe search 3652 is done. 3653 3654 PLPMTUD will be disabled when 0 is set, and other values for it 3655 must be >= 5000. 3656 3657 Default: 0 3658 3659reconf_enable - BOOLEAN 3660 Enable or disable extension of Stream Reconfiguration functionality 3661 specified in RFC6525. This extension provides the ability to "reset" 3662 a stream, and it includes the Parameters of "Outgoing/Incoming SSN 3663 Reset", "SSN/TSN Reset" and "Add Outgoing/Incoming Streams". 3664 3665 Possible values: 3666 3667 - 0 (disabled) - Disable extension. 3668 - 1 (enabled) - Enable extension. 3669 3670 Default: 0 (disabled) 3671 3672intl_enable - BOOLEAN 3673 Enable or disable extension of User Message Interleaving functionality 3674 specified in RFC8260. This extension allows the interleaving of user 3675 messages sent on different streams. With this feature enabled, I-DATA 3676 chunk will replace DATA chunk to carry user messages if also supported 3677 by the peer. Note that to use this feature, one needs to set this option 3678 to 1 and also needs to set socket options SCTP_FRAGMENT_INTERLEAVE to 2 3679 and SCTP_INTERLEAVING_SUPPORTED to 1. 3680 3681 Possible values: 3682 3683 - 0 (disabled) - Disable extension. 3684 - 1 (enabled) - Enable extension. 3685 3686 Default: 0 (disabled) 3687 3688ecn_enable - BOOLEAN 3689 Control use of Explicit Congestion Notification (ECN) by SCTP. 3690 Like in TCP, ECN is used only when both ends of the SCTP connection 3691 indicate support for it. This feature is useful in avoiding losses 3692 due to congestion by allowing supporting routers to signal congestion 3693 before having to drop packets. 3694 3695 Possible values: 3696 3697 - 0 (disabled) - Disable ecn. 3698 - 1 (enabled) - Enable ecn. 3699 3700 Default: 1 (enabled) 3701 3702l3mdev_accept - BOOLEAN 3703 Enabling this option allows a "global" bound socket to work 3704 across L3 master domains (e.g., VRFs) with packets capable of 3705 being received regardless of the L3 domain in which they 3706 originated. Only valid when the kernel was compiled with 3707 CONFIG_NET_L3_MASTER_DEV. 3708 3709 Possible values: 3710 3711 - 0 (disabled) 3712 - 1 (enabled) 3713 3714 Default: 1 (enabled) 3715 3716 3717``/proc/sys/net/core/*`` 3718======================== 3719 3720 Please see: Documentation/admin-guide/sysctl/net.rst for descriptions of these entries. 3721 3722 3723``/proc/sys/net/unix/*`` 3724======================== 3725 3726max_dgram_qlen - INTEGER 3727 The maximum length of dgram socket receive queue 3728 3729 Default: 10 3730 3731