Documentation/networking/ip-sysctl.txt

   1 /proc/sys/net/ipv4/* Variables:
   2
   3 ip_forward - BOOLEAN
   4         0 - disabled (default)
   5         not 0 - enabled
   6
   7         Forward Packets between interfaces.
   8
   9         This variable is special, its change resets all configuration
  10         parameters to their default state (RFC1122 for hosts, RFC1812
  11         for routers)
  12
  13 ip_default_ttl - INTEGER
  14         default 64
  15
  16 ip_no_pmtu_disc - BOOLEAN
  17         Disable Path MTU Discovery.
  18         default FALSE
  19
  20 min_pmtu - INTEGER
  21         default 562 - minimum discovered Path MTU
  22
  23 mtu_expires - INTEGER
  24         Time, in seconds, that cached PMTU information is kept.
  25
  26 min_adv_mss - INTEGER
  27         The advertised MSS depends on the first hop route MTU, but will
  28         never be lower than this setting.
  29
  30 IP Fragmentation:
  31
  32 ipfrag_high_thresh - INTEGER
  33         Maximum memory used to reassemble IP fragments. When
  34         ipfrag_high_thresh bytes of memory is allocated for this purpose,
  35         the fragment handler will toss packets until ipfrag_low_thresh
  36         is reached.
  37
  38 ipfrag_low_thresh - INTEGER
  39         See ipfrag_high_thresh
  40
  41 ipfrag_time - INTEGER
  42         Time in seconds to keep an IP fragment in memory.
  43
  44 ipfrag_secret_interval - INTEGER
  45         Regeneration interval (in seconds) of the hash secret (or lifetime
  46         for the hash secret) for IP fragments.
  47         Default: 600
  48
  49 INET peer storage:
  50
  51 inet_peer_threshold - INTEGER
  52         The approximate size of the storage.  Starting from this threshold
  53         entries will be thrown aggressively.  This threshold also determines
  54         entries' time-to-live and time intervals between garbage collection
  55         passes.  More entries, less time-to-live, less GC interval.
  56
  57 inet_peer_minttl - INTEGER
  58         Minimum time-to-live of entries.  Should be enough to cover fragment
  59         time-to-live on the reassembling side.  This minimum time-to-live  is
  60         guaranteed if the pool size is less than inet_peer_threshold.
  61         Measured in jiffies(1).
  62
  63 inet_peer_maxttl - INTEGER
  64         Maximum time-to-live of entries.  Unused entries will expire after
  65         this period of time if there is no memory pressure on the pool (i.e.
  66         when the number of entries in the pool is very small).
  67         Measured in jiffies(1).
  68
  69 inet_peer_gc_mintime - INTEGER
  70         Minimum interval between garbage collection passes.  This interval is
  71         in effect under high memory pressure on the pool.
  72         Measured in jiffies(1).
  73
  74 inet_peer_gc_maxtime - INTEGER
  75         Minimum interval between garbage collection passes.  This interval is
  76         in effect under low (or absent) memory pressure on the pool.
  77         Measured in jiffies(1).
  78
  79 TCP variables:
  80
  81 tcp_syn_retries - INTEGER
  82         Number of times initial SYNs for an active TCP connection attempt
  83         will be retransmitted. Should not be higher than 255. Default value
  84         is 5, which corresponds to ~180seconds.
  85
  86 tcp_synack_retries - INTEGER
  87         Number of times SYNACKs for a passive TCP connection attempt will
  88         be retransmitted. Should not be higher than 255. Default value
  89         is 5, which corresponds to ~180seconds.
  90
  91 tcp_keepalive_time - INTEGER
  92         How often TCP sends out keepalive messages when keepalive is enabled.
  93         Default: 2hours.
  94
  95 tcp_keepalive_probes - INTEGER
  96         How many keepalive probes TCP sends out, until it decides that the
  97         connection is broken. Default value: 9.
  98
  99 tcp_keepalive_intvl - INTEGER
 100         How frequently the probes are send out. Multiplied by
 101         tcp_keepalive_probes it is time to kill not responding connection,
 102         after probes started. Default value: 75sec i.e. connection
 103         will be aborted after ~11 minutes of retries.
 104
 105 tcp_retries1 - INTEGER
 106         How many times to retry before deciding that something is wrong
 107         and it is necessary to report this suspicion to network layer.
 108         Minimal RFC value is 3, it is default, which corresponds
 109         to ~3sec-8min depending on RTO.
 110
 111 tcp_retries2 - INTEGER
 112         How may times to retry before killing alive TCP connection.
 113         RFC1122 says that the limit should be longer than 100 sec.
 114         It is too small number. Default value 15 corresponds to ~13-30min
 115         depending on RTO.
 116
 117 tcp_orphan_retries - INTEGER
 118         How may times to retry before killing TCP connection, closed
 119         by our side. Default value 7 corresponds to ~50sec-16min
 120         depending on RTO. If you machine is loaded WEB server,
 121         you should think about lowering this value, such sockets
 122         may consume significant resources. Cf. tcp_max_orphans.
 123
 124 tcp_fin_timeout - INTEGER
 125         Time to hold socket in state FIN-WAIT-2, if it was closed
 126         by our side. Peer can be broken and never close its side,
 127         or even died unexpectedly. Default value is 60sec.
 128         Usual value used in 2.2 was 180 seconds, you may restore
 129         it, but remember that if your machine is even underloaded WEB server,
 130         you risk to overflow memory with kilotons of dead sockets,
 131         FIN-WAIT-2 sockets are less dangerous than FIN-WAIT-1,
 132         because they eat maximum 1.5K of memory, but they tend
 133         to live longer. Cf. tcp_max_orphans.
 134
 135 tcp_max_tw_buckets - INTEGER
 136         Maximal number of timewait sockets held by system simultaneously.
 137         If this number is exceeded time-wait socket is immediately destroyed
 138         and warning is printed. This limit exists only to prevent
 139         simple DoS attacks, you _must_ not lower the limit artificially,
 140         but rather increase it (probably, after increasing installed memory),
 141         if network conditions require more than default value.
 142
 143 tcp_tw_recycle - BOOLEAN
 144         Enable fast recycling TIME-WAIT sockets. Default value is 0.
 145         It should not be changed without advice/request of technical
 146         experts.
 147
 148 tcp_tw_reuse - BOOLEAN
 149         Allow to reuse TIME-WAIT sockets for new connections when it is
 150         safe from protocol viewpoint. Default value is 0.
 151         It should not be changed without advice/request of technical
 152         experts.
 153
 154 tcp_max_orphans - INTEGER
 155         Maximal number of TCP sockets not attached to any user file handle,
 156         held by system. If this number is exceeded orphaned connections are
 157         reset immediately and warning is printed. This limit exists
 158         only to prevent simple DoS attacks, you _must_ not rely on this
 159         or lower the limit artificially, but rather increase it
 160         (probably, after increasing installed memory),
 161         if network conditions require more than default value,
 162         and tune network services to linger and kill such states
 163         more aggressively. Let me to remind again: each orphan eats
 164         up to ~64K of unswappable memory.
 165
 166 tcp_abort_on_overflow - BOOLEAN
 167         If listening service is too slow to accept new connections,
 168         reset them. Default state is FALSE. It means that if overflow
 169         occurred due to a burst, connection will recover. Enable this
 170         option _only_ if you are really sure that listening daemon
 171         cannot be tuned to accept connections faster. Enabling this
 172         option can harm clients of your server.
 173
 174 tcp_syncookies - BOOLEAN
 175         Only valid when the kernel was compiled with CONFIG_SYNCOOKIES
 176         Send out syncookies when the syn backlog queue of a socket
 177         overflows. This is to prevent against the common 'syn flood attack'
 178         Default: FALSE
 179
 180         Note, that syncookies is fallback facility.
 181         It MUST NOT be used to help highly loaded servers to stand
 182         against legal connection rate. If you see synflood warnings
 183         in your logs, but investigation shows that they occur
 184         because of overload with legal connections, you should tune
 185         another parameters until this warning disappear.
 186         See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
 187
 188         syncookies seriously violate TCP protocol, do not allow
 189         to use TCP extensions, can result in serious degradation
 190         of some services (f.e. SMTP relaying), visible not by you,
 191         but your clients and relays, contacting you. While you see
 192         synflood warnings in logs not being really flooded, your server
 193         is seriously misconfigured.
 194
 195 tcp_stdurg - BOOLEAN
 196         Use the Host requirements interpretation of the TCP urg pointer field.
 197         Most hosts use the older BSD interpretation, so if you turn this on
 198         Linux might not communicate correctly with them.
 199         Default: FALSE
 200
 201 tcp_max_syn_backlog - INTEGER
 202         Maximal number of remembered connection requests, which are
 203         still did not receive an acknowledgment from connecting client.
 204         Default value is 1024 for systems with more than 128Mb of memory,
 205         and 128 for low memory machines. If server suffers of overload,
 206         try to increase this number.
 207
 208 tcp_window_scaling - BOOLEAN
 209         Enable window scaling as defined in RFC1323.
 210
 211 tcp_timestamps - BOOLEAN
 212         Enable timestamps as defined in RFC1323.
 213
 214 tcp_sack - BOOLEAN
 215         Enable select acknowledgments (SACKS).
 216
 217 tcp_fack - BOOLEAN
 218         Enable FACK congestion avoidance and fast retransmission.
 219         The value is not used, if tcp_sack is not enabled.
 220
 221 tcp_dsack - BOOLEAN
 222         Allows TCP to send "duplicate" SACKs.
 223
 224 tcp_ecn - BOOLEAN
 225         Enable Explicit Congestion Notification in TCP.
 226
 227 tcp_reordering - INTEGER
 228         Maximal reordering of packets in a TCP stream.
 229         Default: 3
 230
 231 tcp_retrans_collapse - BOOLEAN
 232         Bug-to-bug compatibility with some broken printers.
 233         On retransmit try to send bigger packets to work around bugs in
 234         certain TCP stacks.
 235
 236 tcp_wmem - vector of 3 INTEGERs: min, default, max
 237         min: Amount of memory reserved for send buffers for TCP socket.
 238         Each TCP socket has rights to use it due to fact of its birth.
 239         Default: 4K
 240
 241         default: Amount of memory allowed for send buffers for TCP socket
 242         by default. This value overrides net.core.wmem_default used
 243         by other protocols, it is usually lower than net.core.wmem_default.
 244         Default: 16K
 245
 246         max: Maximal amount of memory allowed for automatically selected
 247         send buffers for TCP socket. This value does not override
 248         net.core.wmem_max, "static" selection via SO_SNDBUF does not use this.
 249         Default: 128K
 250
 251 tcp_rmem - vector of 3 INTEGERs: min, default, max
 252         min: Minimal size of receive buffer used by TCP sockets.
 253         It is guaranteed to each TCP socket, even under moderate memory
 254         pressure.
 255         Default: 8K
 256
 257         default: default size of receive buffer used by TCP sockets.
 258         This value overrides net.core.rmem_default used by other protocols.
 259         Default: 87380 bytes. This value results in window of 65535 with
 260         default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
 261         less for default tcp_app_win. See below about these variables.
 262
 263         max: maximal size of receive buffer allowed for automatically
 264         selected receiver buffers for TCP socket. This value does not override
 265         net.core.rmem_max, "static" selection via SO_RCVBUF does not use this.
 266         Default: 87380*2 bytes.
 267
 268 tcp_mem - vector of 3 INTEGERs: min, pressure, max
 269         low: below this number of pages TCP is not bothered about its
 270         memory appetite.
 271
 272         pressure: when amount of memory allocated by TCP exceeds this number
 273         of pages, TCP moderates its memory consumption and enters memory
 274         pressure mode, which is exited when memory consumption falls
 275         under "low".
 276
 277         high: number of pages allowed for queueing by all TCP sockets.
 278
 279         Defaults are calculated at boot time from amount of available
 280         memory.
 281
 282 tcp_app_win - INTEGER
 283         Reserve max(window/2^tcp_app_win, mss) of window for application
 284         buffer. Value 0 is special, it means that nothing is reserved.
 285         Default: 31
 286
 287 tcp_adv_win_scale - INTEGER
 288         Count buffering overhead as bytes/2^tcp_adv_win_scale
 289         (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
 290         if it is <= 0.
 291         Default: 2
 292
 293 tcp_rfc1337 - BOOLEAN
 294         If set, the TCP stack behaves conforming to RFC1337. If unset,
 295         we are not conforming to RFC, but prevent TCP TIME_WAIT
 296         assassination.
 297         Default: 0
 298
 299 tcp_low_latency - BOOLEAN
 300         If set, the TCP stack makes decisions that prefer lower
 301         latency as opposed to higher throughput.  By default, this
 302         option is not set meaning that higher throughput is preferred.
 303         An example of an application where this default should be
 304         changed would be a Beowulf compute cluster.
 305         Default: 0
 306
 307 tcp_westwood - BOOLEAN
 308         Enable TCP Westwood+ congestion control algorithm.
 309         TCP Westwood+ is a sender-side only modification of the TCP Reno
 310         protocol stack that optimizes the performance of TCP congestion
 311         control. It is based on end-to-end bandwidth estimation to set
 312         congestion window and slow start threshold after a congestion
 313         episode. Using this estimation, TCP Westwood+ adaptively sets a
 314         slow start threshold and a congestion window which takes into
 315         account the bandwidth used  at the time congestion is experienced.
 316         TCP Westwood+ significantly increases fairness wrt TCP Reno in
 317         wired networks and throughput over wireless links.
 318         Default: 0
 319
 320 tcp_vegas_cong_avoid - BOOLEAN
 321         Enable TCP Vegas congestion avoidance algorithm.
 322         TCP Vegas is a sender-side only change to TCP that anticipates
 323         the onset of congestion by estimating the bandwidth. TCP Vegas
 324         adjusts the sending rate by modifying the congestion
 325         window. TCP Vegas should provide less packet loss, but it is
 326         not as aggressive as TCP Reno.
 327         Default:0
 328
 329 tcp_bic - BOOLEAN
 330         Enable BIC TCP congestion control algorithm.
 331         BIC-TCP is a sender-side only change that ensures a linear RTT
 332         fairness under large windows while offering both scalability and
 333         bounded TCP-friendliness. The protocol combines two schemes
 334         called additive increase and binary search increase. When the
 335         congestion window is large, additive increase with a large
 336         increment ensures linear RTT fairness as well as good
 337         scalability. Under small congestion windows, binary search
 338         increase provides TCP friendliness.
 339         Default: 0
 340
 341 tcp_bic_low_window - INTEGER
 342         Sets the threshold window (in packets) where BIC TCP starts to
 343         adjust the congestion window. Below this threshold BIC TCP behaves
 344         the same as the default TCP Reno.
 345         Default: 14
 346
 347 tcp_bic_fast_convergence - BOOLEAN
 348         Forces BIC TCP to more quickly respond to changes in congestion
 349         window. Allows two flows sharing the same connection to converge
 350         more rapidly.
 351         Default: 1
 352
 353 tcp_default_win_scale - INTEGER
 354         Sets the minimum window scale TCP will negotiate for on all
 355         conections.
 356         Default: 7
 357
 358 tcp_tso_win_divisor - INTEGER
 359        This allows control over what percentage of the congestion window
 360        can be consumed by a single TSO frame.
 361        The setting of this parameter is a choice between burstiness and
 362        building larger TSO frames.
 363        Default: 8
 364
 365 tcp_frto - BOOLEAN
 366         Enables F-RTO, an enhanced recovery algorithm for TCP retransmission
 367         timeouts.  It is particularly beneficial in wireless environments
 368         where packet loss is typically due to random radio interference
 369         rather than intermediate router congestion.
 370
 371 somaxconn - INTEGER
 372         Limit of socket listen() backlog, known in userspace as SOMAXCONN.
 373         Defaults to 128.  See also tcp_max_syn_backlog for additional tuning
 374         for TCP sockets.
 375
 376 IP Variables:
 377
 378 ip_local_port_range - 2 INTEGERS
 379         Defines the local port range that is used by TCP and UDP to
 380         choose the local port. The first number is the first, the
 381         second the last local port number. Default value depends on
 382         amount of memory available on the system:
 383         > 128Mb 32768-61000
 384         < 128Mb 1024-4999 or even less.
 385         This number defines number of active connections, which this
 386         system can issue simultaneously to systems not supporting
 387         TCP extensions (timestamps). With tcp_tw_recycle enabled
 388         (i.e. by default) range 1024-4999 is enough to issue up to
 389         2000 connections per second to systems supporting timestamps.
 390
 391 ip_nonlocal_bind - BOOLEAN
 392         If set, allows processes to bind() to non-local IP addresses,
 393         which can be quite useful - but may break some applications.
 394         Default: 0
 395
 396 ip_dynaddr - BOOLEAN
 397         If set non-zero, enables support for dynamic addresses.
 398         If set to a non-zero value larger than 1, a kernel log
 399         message will be printed when dynamic address rewriting
 400         occurs.
 401         Default: 0
 402
 403 icmp_echo_ignore_all - BOOLEAN
 404 icmp_echo_ignore_broadcasts - BOOLEAN
 405         If either is set to true, then the kernel will ignore either all
 406         ICMP ECHO requests sent to it or just those to broadcast/multicast
 407         addresses, respectively.
 408
 409 icmp_ratelimit - INTEGER
 410         Limit the maximal rates for sending ICMP packets whose type matches
 411         icmp_ratemask (see below) to specific targets.
 412         0 to disable any limiting, otherwise the maximal rate in jiffies(1)
 413         Default: 100
 414
 415 icmp_ratemask - INTEGER
 416         Mask made of ICMP types for which rates are being limited.
 417         Significant bits: IHGFEDCBA9876543210
 418         Default mask:     0000001100000011000 (6168)
 419
 420         Bit definitions (see include/linux/icmp.h):
 421                 0 Echo Reply
 422                 3 Destination Unreachable *
 423                 4 Source Quench *
 424                 5 Redirect
 425                 8 Echo Request
 426                 B Time Exceeded *
 427                 C Parameter Problem *
 428                 D Timestamp Request
 429                 E Timestamp Reply
 430                 F Info Request
 431                 G Info Reply
 432                 H Address Mask Request
 433                 I Address Mask Reply
 434
 435         * These are rate limited by default (see default mask above)
 436
 437 icmp_ignore_bogus_error_responses - BOOLEAN
 438         Some routers violate RFC1122 by sending bogus responses to broadcast
 439         frames.  Such violations are normally logged via a kernel warning.
 440         If this is set to TRUE, the kernel will not give such warnings, which
 441         will avoid log file clutter.
 442         Default: FALSE
 443
 444 igmp_max_memberships - INTEGER
 445         Change the maximum number of multicast groups we can subscribe to.
 446         Default: 20
 447
 448 conf/interface/*  changes special settings per interface (where "interface" is
 449                   the name of your network interface)
 450 conf/all/*        is special, changes the settings for all interfaces
 451
 452
 453 log_martians - BOOLEAN
 454         Log packets with impossible addresses to kernel log.
 455         log_martians for the interface will be enabled if at least one of
 456         conf/{all,interface}/log_martians is set to TRUE,
 457         it will be disabled otherwise
 458
 459 accept_redirects - BOOLEAN
 460         Accept ICMP redirect messages.
 461         accept_redirects for the interface will be enabled if:
 462         - both conf/{all,interface}/accept_redirects are TRUE in the case forwarding
 463           for the interface is enabled
 464         or
 465         - at least one of conf/{all,interface}/accept_redirects is TRUE in the case
 466           forwarding for the interface is disabled
 467         accept_redirects for the interface will be disabled otherwise
 468         default TRUE (host)
 469                 FALSE (router)
 470
 471 forwarding - BOOLEAN
 472         Enable IP forwarding on this interface.
 473
 474 mc_forwarding - BOOLEAN
 475         Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
 476         and a multicast routing daemon is required.
 477         conf/all/mc_forwarding must also be set to TRUE to enable multicast routing
 478         for the interface
 479
 480 medium_id - INTEGER
 481         Integer value used to differentiate the devices by the medium they
 482         are attached to. Two devices can have different id values when
 483         the broadcast packets are received only on one of them.
 484         The default value 0 means that the device is the only interface
 485         to its medium, value of -1 means that medium is not known.
 486
 487         Currently, it is used to change the proxy_arp behavior:
 488         the proxy_arp feature is enabled for packets forwarded between
 489         two devices attached to different media.
 490
 491 proxy_arp - BOOLEAN
 492         Do proxy arp.
 493         proxy_arp for the interface will be enabled if at least one of
 494         conf/{all,interface}/proxy_arp is set to TRUE,
 495         it will be disabled otherwise
 496
 497 shared_media - BOOLEAN
 498         Send(router) or accept(host) RFC1620 shared media redirects.
 499         Overrides ip_secure_redirects.
 500         shared_media for the interface will be enabled if at least one of
 501         conf/{all,interface}/shared_media is set to TRUE,
 502         it will be disabled otherwise
 503         default TRUE
 504
 505 secure_redirects - BOOLEAN
 506         Accept ICMP redirect messages only for gateways,
 507         listed in default gateway list.
 508         secure_redirects for the interface will be enabled if at least one of
 509         conf/{all,interface}/secure_redirects is set to TRUE,
 510         it will be disabled otherwise
 511         default TRUE
 512
 513 send_redirects - BOOLEAN
 514         Send redirects, if router.
 515         send_redirects for the interface will be enabled if at least one of
 516         conf/{all,interface}/send_redirects is set to TRUE,
 517         it will be disabled otherwise
 518         Default: TRUE
 519
 520 bootp_relay - BOOLEAN
 521         Accept packets with source address 0.b.c.d destined
 522         not to this host as local ones. It is supposed, that
 523         BOOTP relay daemon will catch and forward such packets.
 524         conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
 525         for the interface
 526         default FALSE
 527         Not Implemented Yet.
 528
 529 accept_source_route - BOOLEAN
 530         Accept packets with SRR option.
 531         conf/all/accept_source_route must also be set to TRUE to accept packets
 532         with SRR option on the interface
 533         default TRUE (router)
 534                 FALSE (host)
 535
 536 rp_filter - BOOLEAN
 537         1 - do source validation by reversed path, as specified in RFC1812
 538             Recommended option for single homed hosts and stub network
 539             routers. Could cause troubles for complicated (not loop free)
 540             networks running a slow unreliable protocol (sort of RIP),
 541             or using static routes.
 542
 543         0 - No source validation.
 544
 545         conf/all/rp_filter must also be set to TRUE to do source validation
 546         on the interface
 547
 548         Default value is 0. Note that some distributions enable it
 549         in startup scripts.
 550
 551 arp_filter - BOOLEAN
 552         1 - Allows you to have multiple network interfaces on the same
 553         subnet, and have the ARPs for each interface be answered
 554         based on whether or not the kernel would route a packet from
 555         the ARP'd IP out that interface (therefore you must use source
 556         based routing for this to work). In other words it allows control
 557         of which cards (usually 1) will respond to an arp request.
 558
 559         0 - (default) The kernel can respond to arp requests with addresses
 560         from other interfaces. This may seem wrong but it usually makes
 561         sense, because it increases the chance of successful communication.
 562         IP addresses are owned by the complete host on Linux, not by
 563         particular interfaces. Only for more complex setups like load-
 564         balancing, does this behaviour cause problems.
 565
 566         arp_filter for the interface will be enabled if at least one of
 567         conf/{all,interface}/arp_filter is set to TRUE,
 568         it will be disabled otherwise
 569
 570 arp_announce - INTEGER
 571         Define different restriction levels for announcing the local
 572         source IP address from IP packets in ARP requests sent on
 573         interface:
 574         0 - (default) Use any local address, configured on any interface
 575         1 - Try to avoid local addresses that are not in the target's
 576         subnet for this interface. This mode is useful when target
 577         hosts reachable via this interface require the source IP
 578         address in ARP requests to be part of their logical network
 579         configured on the receiving interface. When we generate the
 580         request we will check all our subnets that include the
 581         target IP and will preserve the source address if it is from
 582         such subnet. If there is no such subnet we select source
 583         address according to the rules for level 2.
 584         2 - Always use the best local address for this target.
 585         In this mode we ignore the source address in the IP packet
 586         and try to select local address that we prefer for talks with
 587         the target host. Such local address is selected by looking
 588         for primary IP addresses on all our subnets on the outgoing
 589         interface that include the target IP address. If no suitable
 590         local address is found we select the first local address
 591         we have on the outgoing interface or on all other interfaces,
 592         with the hope we will receive reply for our request and
 593         even sometimes no matter the source IP address we announce.
 594
 595         The max value from conf/{all,interface}/arp_announce is used.
 596
 597         Increasing the restriction level gives more chance for
 598         receiving answer from the resolved target while decreasing
 599         the level announces more valid sender's information.
 600
 601 arp_ignore - INTEGER
 602         Define different modes for sending replies in response to
 603         received ARP requests that resolve local target IP addresses:
 604         0 - (default): reply for any local target IP address, configured
 605         on any interface
 606         1 - reply only if the target IP address is local address
 607         configured on the incoming interface
 608         2 - reply only if the target IP address is local address
 609         configured on the incoming interface and both with the
 610         sender's IP address are part from same subnet on this interface
 611         3 - do not reply for local addresses configured with scope host,
 612         only resolutions for global and link addresses are replied
 613         4-7 - reserved
 614         8 - do not reply for all local addresses
 615
 616         The max value from conf/{all,interface}/arp_ignore is used
 617         when ARP request is received on the {interface}
 618
 619 app_solicit - INTEGER
 620         The maximum number of probes to send to the user space ARP daemon
 621         via netlink before dropping back to multicast probes (see
 622         mcast_solicit).  Defaults to 0.
 623
 624 disable_policy - BOOLEAN
 625         Disable IPSEC policy (SPD) for this interface
 626
 627 disable_xfrm - BOOLEAN
 628         Disable IPSEC encryption on this interface, whatever the policy
 629
 630
 631
 632 tag - INTEGER
 633         Allows you to write a number, which can be used as required.
 634         Default value is 0.
 635
 636 (1) Jiffie: internal timeunit for the kernel. On the i386 1/100s, on the
 637 Alpha 1/1024s. See the HZ define in /usr/include/asm/param.h for the exact
 638 value on your system.
 639
 640 Alexey Kuznetsov.
 641 kuznet@ms2.inr.ac.ru
 642
 643 Updated by:
 644 Andi Kleen
 645 ak@muc.de
 646 Nicolas Delon
 647 delon.nicolas@wanadoo.fr
 648
 649
 650
 651
 652 /proc/sys/net/ipv6/* Variables:
 653
 654 IPv6 has no global variables such as tcp_*.  tcp_* settings under ipv4/ also
 655 apply to IPv6 [XXX?].
 656
 657 bindv6only - BOOLEAN
 658         Default value for IPV6_V6ONLY socket option,
 659         which restricts use of the IPv6 socket to IPv6 communication
 660         only.
 661                 TRUE: disable IPv4-mapped address feature
 662                 FALSE: enable IPv4-mapped address feature
 663
 664         Default: FALSE (as specified in RFC2553bis)
 665
 666 IPv6 Fragmentation:
 667
 668 ip6frag_high_thresh - INTEGER
 669         Maximum memory used to reassemble IPv6 fragments. When
 670         ip6frag_high_thresh bytes of memory is allocated for this purpose,
 671         the fragment handler will toss packets until ip6frag_low_thresh
 672         is reached.
 673
 674 ip6frag_low_thresh - INTEGER
 675         See ip6frag_high_thresh
 676
 677 ip6frag_time - INTEGER
 678         Time in seconds to keep an IPv6 fragment in memory.
 679
 680 ip6frag_secret_interval - INTEGER
 681         Regeneration interval (in seconds) of the hash secret (or lifetime
 682         for the hash secret) for IPv6 fragments.
 683         Default: 600
 684
 685 conf/default/*:
 686         Change the interface-specific default settings.
 687
 688
 689 conf/all/*:
 690         Change all the interface-specific settings.
 691
 692         [XXX:  Other special features than forwarding?]
 693
 694 conf/all/forwarding - BOOLEAN
 695         Enable global IPv6 forwarding between all interfaces.
 696
 697         IPv4 and IPv6 work differently here; e.g. netfilter must be used
 698         to control which interfaces may forward packets and which not.
 699
 700         This also sets all interfaces' Host/Router setting
 701         'forwarding' to the specified value.  See below for details.
 702
 703         This referred to as global forwarding.
 704
 705 conf/interface/*:
 706         Change special settings per interface.
 707
 708         The functional behaviour for certain settings is different
 709         depending on whether local forwarding is enabled or not.
 710
 711 accept_ra - BOOLEAN
 712         Accept Router Advertisements; autoconfigure using them.
 713
 714         Functional default: enabled if local forwarding is disabled.
 715                             disabled if local forwarding is enabled.
 716
 717 accept_redirects - BOOLEAN
 718         Accept Redirects.
 719
 720         Functional default: enabled if local forwarding is disabled.
 721                             disabled if local forwarding is enabled.
 722
 723 autoconf - BOOLEAN
 724         Autoconfigure addresses using Prefix Information in Router
 725         Advertisements.
 726
 727         Functional default: enabled if accept_ra is enabled.
 728                             disabled if accept_ra is disabled.
 729
 730 dad_transmits - INTEGER
 731         The amount of Duplicate Address Detection probes to send.
 732         Default: 1
 733
 734 forwarding - BOOLEAN
 735         Configure interface-specific Host/Router behaviour.
 736
 737         Note: It is recommended to have the same setting on all
 738         interfaces; mixed router/host scenarios are rather uncommon.
 739
 740         FALSE:
 741
 742         By default, Host behaviour is assumed.  This means:
 743
 744         1. IsRouter flag is not set in Neighbour Advertisements.
 745         2. Router Solicitations are being sent when necessary.
 746         3. If accept_ra is TRUE (default), accept Router
 747            Advertisements (and do autoconfiguration).
 748         4. If accept_redirects is TRUE (default), accept Redirects.
 749
 750         TRUE:
 751
 752         If local forwarding is enabled, Router behaviour is assumed.
 753         This means exactly the reverse from the above:
 754
 755         1. IsRouter flag is set in Neighbour Advertisements.
 756         2. Router Solicitations are not sent.
 757         3. Router Advertisements are ignored.
 758         4. Redirects are ignored.
 759
 760         Default: FALSE if global forwarding is disabled (default),
 761                  otherwise TRUE.
 762
 763 hop_limit - INTEGER
 764         Default Hop Limit to set.
 765         Default: 64
 766
 767 mtu - INTEGER
 768         Default Maximum Transfer Unit
 769         Default: 1280 (IPv6 required minimum)
 770
 771 router_solicitation_delay - INTEGER
 772         Number of seconds to wait after interface is brought up
 773         before sending Router Solicitations.
 774         Default: 1
 775
 776 router_solicitation_interval - INTEGER
 777         Number of seconds to wait between Router Solicitations.
 778         Default: 4
 779
 780 router_solicitations - INTEGER
 781         Number of Router Solicitations to send until assuming no
 782         routers are present.
 783         Default: 3
 784
 785 use_tempaddr - INTEGER
 786         Preference for Privacy Extensions (RFC3041).
 787           <= 0 : disable Privacy Extensions
 788           == 1 : enable Privacy Extensions, but prefer public
 789                  addresses over temporary addresses.
 790           >  1 : enable Privacy Extensions and prefer temporary
 791                  addresses over public addresses.
 792         Default:  0 (for most devices)
 793                  -1 (for point-to-point devices and loopback devices)
 794
 795 temp_valid_lft - INTEGER
 796         valid lifetime (in seconds) for temporary addresses.
 797         Default: 604800 (7 days)
 798
 799 temp_prefered_lft - INTEGER
 800         Preferred lifetime (in seconds) for temporary addresses.
 801         Default: 86400 (1 day)
 802
 803 max_desync_factor - INTEGER
 804         Maximum value for DESYNC_FACTOR, which is a random value
 805         that ensures that clients don't synchronize with each
 806         other and generate new addresses at exactly the same time.
 807         value is in seconds.
 808         Default: 600
 809
 810 regen_max_retry - INTEGER
 811         Number of attempts before give up attempting to generate
 812         valid temporary addresses.
 813         Default: 5
 814
 815 max_addresses - INTEGER
 816         Number of maximum addresses per interface.  0 disables limitation.
 817         It is recommended not set too large value (or 0) because it would
 818         be too easy way to crash kernel to allow to create too much of
 819         autoconfigured addresses.
 820         Default: 16
 821
 822 icmp/*:
 823 ratelimit - INTEGER
 824         Limit the maximal rates for sending ICMPv6 packets.
 825         0 to disable any limiting, otherwise the maximal rate in jiffies(1)
 826         Default: 100
 827
 828
 829 IPv6 Update by:
 830 Pekka Savola <pekkas@netcore.fi>
 831 YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
 832
 833
 834 /proc/sys/net/bridge/* Variables:
 835
 836 bridge-nf-call-arptables - BOOLEAN
 837         1 : pass bridged ARP traffic to arptables' FORWARD chain.
 838         0 : disable this.
 839         Default: 1
 840
 841 bridge-nf-call-iptables - BOOLEAN
 842         1 : pass bridged IPv4 traffic to iptables' chains.
 843         0 : disable this.
 844         Default: 1
 845
 846 bridge-nf-call-ip6tables - BOOLEAN
 847         1 : pass bridged IPv6 traffic to ip6tables' chains.
 848         0 : disable this.
 849         Default: 1
 850
 851 bridge-nf-filter-vlan-tagged - BOOLEAN
 852         1 : pass bridged vlan-tagged ARP/IP traffic to arptables/iptables.
 853         0 : disable this.
 854         Default: 1
 855
 856
 857 UNDOCUMENTED:
 858
 859 dev_weight FIXME
 860 discovery_slots FIXME
 861 discovery_timeout FIXME
 862 fast_poll_increase FIXME
 863 ip6_queue_maxlen FIXME
 864 lap_keepalive_time FIXME
 865 lo_cong FIXME
 866 max_baud_rate FIXME
 867 max_dgram_qlen FIXME
 868 max_noreply_time FIXME
 869 max_tx_data_size FIXME
 870 max_tx_window FIXME
 871 min_tx_turn_time FIXME
 872 mod_cong FIXME
 873 no_cong FIXME
 874 no_cong_thresh FIXME
 875 slot_timeout FIXME
 876 warn_noreply_time FIXME
 877
 878 $Id: ip-sysctl.txt,v 1.20 2001/12/13 09:00:18 davem Exp $