linux/Documentation/networking/ip-sysctl.txt
<<
>>
Prefs
   1/proc/sys/net/ipv4/* Variables:
   2
   3ip_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
  13ip_default_ttl - INTEGER
  14        Default value of TTL field (Time To Live) for outgoing (but not
  15        forwarded) IP packets. Should be between 1 and 255 inclusive.
  16        Default: 64 (as recommended by RFC1700)
  17
  18ip_no_pmtu_disc - BOOLEAN
  19        Disable Path MTU Discovery.
  20        default FALSE
  21
  22min_pmtu - INTEGER
  23        default 552 - minimum discovered Path MTU
  24
  25route/max_size - INTEGER
  26        Maximum number of routes allowed in the kernel.  Increase
  27        this when using large numbers of interfaces and/or routes.
  28
  29neigh/default/gc_thresh1 - INTEGER
  30        Minimum number of entries to keep.  Garbage collector will not
  31        purge entries if there are fewer than this number.
  32        Default: 256
  33
  34neigh/default/gc_thresh3 - INTEGER
  35        Maximum number of neighbor entries allowed.  Increase this
  36        when using large numbers of interfaces and when communicating
  37        with large numbers of directly-connected peers.
  38        Default: 1024
  39
  40neigh/default/unres_qlen_bytes - INTEGER
  41        The maximum number of bytes which may be used by packets
  42        queued for each unresolved address by other network layers.
  43        (added in linux 3.3)
  44        Setting negative value is meaningless and will return error.
  45        Default: 65536 Bytes(64KB)
  46
  47neigh/default/unres_qlen - INTEGER
  48        The maximum number of packets which may be queued for each
  49        unresolved address by other network layers.
  50        (deprecated in linux 3.3) : use unres_qlen_bytes instead.
  51        Prior to linux 3.3, the default value is 3 which may cause
  52        unexpected packet loss. The current default value is calculated
  53        according to default value of unres_qlen_bytes and true size of
  54        packet.
  55        Default: 31
  56
  57mtu_expires - INTEGER
  58        Time, in seconds, that cached PMTU information is kept.
  59
  60min_adv_mss - INTEGER
  61        The advertised MSS depends on the first hop route MTU, but will
  62        never be lower than this setting.
  63
  64IP Fragmentation:
  65
  66ipfrag_high_thresh - INTEGER
  67        Maximum memory used to reassemble IP fragments. When
  68        ipfrag_high_thresh bytes of memory is allocated for this purpose,
  69        the fragment handler will toss packets until ipfrag_low_thresh
  70        is reached.
  71
  72ipfrag_low_thresh - INTEGER
  73        See ipfrag_high_thresh
  74
  75ipfrag_time - INTEGER
  76        Time in seconds to keep an IP fragment in memory.
  77
  78ipfrag_secret_interval - INTEGER
  79        Regeneration interval (in seconds) of the hash secret (or lifetime
  80        for the hash secret) for IP fragments.
  81        Default: 600
  82
  83ipfrag_max_dist - INTEGER
  84        ipfrag_max_dist is a non-negative integer value which defines the
  85        maximum "disorder" which is allowed among fragments which share a
  86        common IP source address. Note that reordering of packets is
  87        not unusual, but if a large number of fragments arrive from a source
  88        IP address while a particular fragment queue remains incomplete, it
  89        probably indicates that one or more fragments belonging to that queue
  90        have been lost. When ipfrag_max_dist is positive, an additional check
  91        is done on fragments before they are added to a reassembly queue - if
  92        ipfrag_max_dist (or more) fragments have arrived from a particular IP
  93        address between additions to any IP fragment queue using that source
  94        address, it's presumed that one or more fragments in the queue are
  95        lost. The existing fragment queue will be dropped, and a new one
  96        started. An ipfrag_max_dist value of zero disables this check.
  97
  98        Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
  99        result in unnecessarily dropping fragment queues when normal
 100        reordering of packets occurs, which could lead to poor application
 101        performance. Using a very large value, e.g. 50000, increases the
 102        likelihood of incorrectly reassembling IP fragments that originate
 103        from different IP datagrams, which could result in data corruption.
 104        Default: 64
 105
 106INET peer storage:
 107
 108inet_peer_threshold - INTEGER
 109        The approximate size of the storage.  Starting from this threshold
 110        entries will be thrown aggressively.  This threshold also determines
 111        entries' time-to-live and time intervals between garbage collection
 112        passes.  More entries, less time-to-live, less GC interval.
 113
 114inet_peer_minttl - INTEGER
 115        Minimum time-to-live of entries.  Should be enough to cover fragment
 116        time-to-live on the reassembling side.  This minimum time-to-live  is
 117        guaranteed if the pool size is less than inet_peer_threshold.
 118        Measured in seconds.
 119
 120inet_peer_maxttl - INTEGER
 121        Maximum time-to-live of entries.  Unused entries will expire after
 122        this period of time if there is no memory pressure on the pool (i.e.
 123        when the number of entries in the pool is very small).
 124        Measured in seconds.
 125
 126TCP variables:
 127
 128somaxconn - INTEGER
 129        Limit of socket listen() backlog, known in userspace as SOMAXCONN.
 130        Defaults to 128.  See also tcp_max_syn_backlog for additional tuning
 131        for TCP sockets.
 132
 133tcp_abort_on_overflow - BOOLEAN
 134        If listening service is too slow to accept new connections,
 135        reset them. Default state is FALSE. It means that if overflow
 136        occurred due to a burst, connection will recover. Enable this
 137        option _only_ if you are really sure that listening daemon
 138        cannot be tuned to accept connections faster. Enabling this
 139        option can harm clients of your server.
 140
 141tcp_adv_win_scale - INTEGER
 142        Count buffering overhead as bytes/2^tcp_adv_win_scale
 143        (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
 144        if it is <= 0.
 145        Possible values are [-31, 31], inclusive.
 146        Default: 1
 147
 148tcp_allowed_congestion_control - STRING
 149        Show/set the congestion control choices available to non-privileged
 150        processes. The list is a subset of those listed in
 151        tcp_available_congestion_control.
 152        Default is "reno" and the default setting (tcp_congestion_control).
 153
 154tcp_app_win - INTEGER
 155        Reserve max(window/2^tcp_app_win, mss) of window for application
 156        buffer. Value 0 is special, it means that nothing is reserved.
 157        Default: 31
 158
 159tcp_available_congestion_control - STRING
 160        Shows the available congestion control choices that are registered.
 161        More congestion control algorithms may be available as modules,
 162        but not loaded.
 163
 164tcp_base_mss - INTEGER
 165        The initial value of search_low to be used by the packetization layer
 166        Path MTU discovery (MTU probing).  If MTU probing is enabled,
 167        this is the initial MSS used by the connection.
 168
 169tcp_congestion_control - STRING
 170        Set the congestion control algorithm to be used for new
 171        connections. The algorithm "reno" is always available, but
 172        additional choices may be available based on kernel configuration.
 173        Default is set as part of kernel configuration.
 174        For passive connections, the listener congestion control choice
 175        is inherited.
 176        [see setsockopt(listenfd, SOL_TCP, TCP_CONGESTION, "name" ...) ]
 177
 178tcp_cookie_size - INTEGER
 179        Default size of TCP Cookie Transactions (TCPCT) option, that may be
 180        overridden on a per socket basis by the TCPCT socket option.
 181        Values greater than the maximum (16) are interpreted as the maximum.
 182        Values greater than zero and less than the minimum (8) are interpreted
 183        as the minimum.  Odd values are interpreted as the next even value.
 184        Default: 0 (off).
 185
 186tcp_dsack - BOOLEAN
 187        Allows TCP to send "duplicate" SACKs.
 188
 189tcp_early_retrans - INTEGER
 190        Enable Early Retransmit (ER), per RFC 5827. ER lowers the threshold
 191        for triggering fast retransmit when the amount of outstanding data is
 192        small and when no previously unsent data can be transmitted (such
 193        that limited transmit could be used).
 194        Possible values:
 195                0 disables ER
 196                1 enables ER
 197                2 enables ER but delays fast recovery and fast retransmit
 198                  by a fourth of RTT. This mitigates connection falsely
 199                  recovers when network has a small degree of reordering
 200                  (less than 3 packets).
 201        Default: 2
 202
 203tcp_ecn - INTEGER
 204        Control use of Explicit Congestion Notification (ECN) by TCP.
 205        ECN is used only when both ends of the TCP connection indicate
 206        support for it.  This feature is useful in avoiding losses due
 207        to congestion by allowing supporting routers to signal
 208        congestion before having to drop packets.
 209        Possible values are:
 210                0 Disable ECN.  Neither initiate nor accept ECN.
 211                1 Enable ECN when requested by incoming connections and
 212                  also request ECN on outgoing connection attempts.
 213                2 Enable ECN when requested by incoming connections
 214                  but do not request ECN on outgoing connections.
 215        Default: 2
 216
 217tcp_fack - BOOLEAN
 218        Enable FACK congestion avoidance and fast retransmission.
 219        The value is not used, if tcp_sack is not enabled.
 220
 221tcp_fin_timeout - INTEGER
 222        The length of time an orphaned (no longer referenced by any
 223        application) connection will remain in the FIN_WAIT_2 state
 224        before it is aborted at the local end.  While a perfectly
 225        valid "receive only" state for an un-orphaned connection, an
 226        orphaned connection in FIN_WAIT_2 state could otherwise wait
 227        forever for the remote to close its end of the connection.
 228        Cf. tcp_max_orphans
 229        Default: 60 seconds
 230
 231tcp_frto - INTEGER
 232        Enables Forward RTO-Recovery (F-RTO) defined in RFC4138.
 233        F-RTO is an enhanced recovery algorithm for TCP retransmission
 234        timeouts.  It is particularly beneficial in wireless environments
 235        where packet loss is typically due to random radio interference
 236        rather than intermediate router congestion.  F-RTO is sender-side
 237        only modification. Therefore it does not require any support from
 238        the peer.
 239
 240        If set to 1, basic version is enabled.  2 enables SACK enhanced
 241        F-RTO if flow uses SACK.  The basic version can be used also when
 242        SACK is in use though scenario(s) with it exists where F-RTO
 243        interacts badly with the packet counting of the SACK enabled TCP
 244        flow.
 245
 246tcp_frto_response - INTEGER
 247        When F-RTO has detected that a TCP retransmission timeout was
 248        spurious (i.e, the timeout would have been avoided had TCP set a
 249        longer retransmission timeout), TCP has several options what to do
 250        next. Possible values are:
 251                0 Rate halving based; a smooth and conservative response,
 252                  results in halved cwnd and ssthresh after one RTT
 253                1 Very conservative response; not recommended because even
 254                  though being valid, it interacts poorly with the rest of
 255                  Linux TCP, halves cwnd and ssthresh immediately
 256                2 Aggressive response; undoes congestion control measures
 257                  that are now known to be unnecessary (ignoring the
 258                  possibility of a lost retransmission that would require
 259                  TCP to be more cautious), cwnd and ssthresh are restored
 260                  to the values prior timeout
 261        Default: 0 (rate halving based)
 262
 263tcp_keepalive_time - INTEGER
 264        How often TCP sends out keepalive messages when keepalive is enabled.
 265        Default: 2hours.
 266
 267tcp_keepalive_probes - INTEGER
 268        How many keepalive probes TCP sends out, until it decides that the
 269        connection is broken. Default value: 9.
 270
 271tcp_keepalive_intvl - INTEGER
 272        How frequently the probes are send out. Multiplied by
 273        tcp_keepalive_probes it is time to kill not responding connection,
 274        after probes started. Default value: 75sec i.e. connection
 275        will be aborted after ~11 minutes of retries.
 276
 277tcp_low_latency - BOOLEAN
 278        If set, the TCP stack makes decisions that prefer lower
 279        latency as opposed to higher throughput.  By default, this
 280        option is not set meaning that higher throughput is preferred.
 281        An example of an application where this default should be
 282        changed would be a Beowulf compute cluster.
 283        Default: 0
 284
 285tcp_max_orphans - INTEGER
 286        Maximal number of TCP sockets not attached to any user file handle,
 287        held by system. If this number is exceeded orphaned connections are
 288        reset immediately and warning is printed. This limit exists
 289        only to prevent simple DoS attacks, you _must_ not rely on this
 290        or lower the limit artificially, but rather increase it
 291        (probably, after increasing installed memory),
 292        if network conditions require more than default value,
 293        and tune network services to linger and kill such states
 294        more aggressively. Let me to remind again: each orphan eats
 295        up to ~64K of unswappable memory.
 296
 297tcp_max_ssthresh - INTEGER
 298        Limited Slow-Start for TCP with large congestion windows (cwnd) defined in
 299        RFC3742. Limited slow-start is a mechanism to limit growth of the cwnd
 300        on the region where cwnd is larger than tcp_max_ssthresh. TCP increases cwnd
 301        by at most tcp_max_ssthresh segments, and by at least tcp_max_ssthresh/2
 302        segments per RTT when the cwnd is above tcp_max_ssthresh.
 303        If TCP connection increased cwnd to thousands (or tens of thousands) segments,
 304        and thousands of packets were being dropped during slow-start, you can set
 305        tcp_max_ssthresh to improve performance for new TCP connection.
 306        Default: 0 (off)
 307
 308tcp_max_syn_backlog - INTEGER
 309        Maximal number of remembered connection requests, which have not
 310        received an acknowledgment from connecting client.
 311        The minimal value is 128 for low memory machines, and it will
 312        increase in proportion to the memory of machine.
 313        If server suffers from overload, try increasing this number.
 314
 315tcp_max_tw_buckets - INTEGER
 316        Maximal number of timewait sockets held by system simultaneously.
 317        If this number is exceeded time-wait socket is immediately destroyed
 318        and warning is printed. This limit exists only to prevent
 319        simple DoS attacks, you _must_ not lower the limit artificially,
 320        but rather increase it (probably, after increasing installed memory),
 321        if network conditions require more than default value.
 322
 323tcp_mem - vector of 3 INTEGERs: min, pressure, max
 324        min: below this number of pages TCP is not bothered about its
 325        memory appetite.
 326
 327        pressure: when amount of memory allocated by TCP exceeds this number
 328        of pages, TCP moderates its memory consumption and enters memory
 329        pressure mode, which is exited when memory consumption falls
 330        under "min".
 331
 332        max: number of pages allowed for queueing by all TCP sockets.
 333
 334        Defaults are calculated at boot time from amount of available
 335        memory.
 336
 337tcp_moderate_rcvbuf - BOOLEAN
 338        If set, TCP performs receive buffer auto-tuning, attempting to
 339        automatically size the buffer (no greater than tcp_rmem[2]) to
 340        match the size required by the path for full throughput.  Enabled by
 341        default.
 342
 343tcp_mtu_probing - INTEGER
 344        Controls TCP Packetization-Layer Path MTU Discovery.  Takes three
 345        values:
 346          0 - Disabled
 347          1 - Disabled by default, enabled when an ICMP black hole detected
 348          2 - Always enabled, use initial MSS of tcp_base_mss.
 349
 350tcp_no_metrics_save - BOOLEAN
 351        By default, TCP saves various connection metrics in the route cache
 352        when the connection closes, so that connections established in the
 353        near future can use these to set initial conditions.  Usually, this
 354        increases overall performance, but may sometimes cause performance
 355        degradation.  If set, TCP will not cache metrics on closing
 356        connections.
 357
 358tcp_orphan_retries - INTEGER
 359        This value influences the timeout of a locally closed TCP connection,
 360        when RTO retransmissions remain unacknowledged.
 361        See tcp_retries2 for more details.
 362
 363        The default value is 8.
 364        If your machine is a loaded WEB server,
 365        you should think about lowering this value, such sockets
 366        may consume significant resources. Cf. tcp_max_orphans.
 367
 368tcp_reordering - INTEGER
 369        Maximal reordering of packets in a TCP stream.
 370        Default: 3
 371
 372tcp_retrans_collapse - BOOLEAN
 373        Bug-to-bug compatibility with some broken printers.
 374        On retransmit try to send bigger packets to work around bugs in
 375        certain TCP stacks.
 376
 377tcp_retries1 - INTEGER
 378        This value influences the time, after which TCP decides, that
 379        something is wrong due to unacknowledged RTO retransmissions,
 380        and reports this suspicion to the network layer.
 381        See tcp_retries2 for more details.
 382
 383        RFC 1122 recommends at least 3 retransmissions, which is the
 384        default.
 385
 386tcp_retries2 - INTEGER
 387        This value influences the timeout of an alive TCP connection,
 388        when RTO retransmissions remain unacknowledged.
 389        Given a value of N, a hypothetical TCP connection following
 390        exponential backoff with an initial RTO of TCP_RTO_MIN would
 391        retransmit N times before killing the connection at the (N+1)th RTO.
 392
 393        The default value of 15 yields a hypothetical timeout of 924.6
 394        seconds and is a lower bound for the effective timeout.
 395        TCP will effectively time out at the first RTO which exceeds the
 396        hypothetical timeout.
 397
 398        RFC 1122 recommends at least 100 seconds for the timeout,
 399        which corresponds to a value of at least 8.
 400
 401tcp_rfc1337 - BOOLEAN
 402        If set, the TCP stack behaves conforming to RFC1337. If unset,
 403        we are not conforming to RFC, but prevent TCP TIME_WAIT
 404        assassination.
 405        Default: 0
 406
 407tcp_rmem - vector of 3 INTEGERs: min, default, max
 408        min: Minimal size of receive buffer used by TCP sockets.
 409        It is guaranteed to each TCP socket, even under moderate memory
 410        pressure.
 411        Default: 1 page
 412
 413        default: initial size of receive buffer used by TCP sockets.
 414        This value overrides net.core.rmem_default used by other protocols.
 415        Default: 87380 bytes. This value results in window of 65535 with
 416        default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
 417        less for default tcp_app_win. See below about these variables.
 418
 419        max: maximal size of receive buffer allowed for automatically
 420        selected receiver buffers for TCP socket. This value does not override
 421        net.core.rmem_max.  Calling setsockopt() with SO_RCVBUF disables
 422        automatic tuning of that socket's receive buffer size, in which
 423        case this value is ignored.
 424        Default: between 87380B and 6MB, depending on RAM size.
 425
 426tcp_sack - BOOLEAN
 427        Enable select acknowledgments (SACKS).
 428
 429tcp_slow_start_after_idle - BOOLEAN
 430        If set, provide RFC2861 behavior and time out the congestion
 431        window after an idle period.  An idle period is defined at
 432        the current RTO.  If unset, the congestion window will not
 433        be timed out after an idle period.
 434        Default: 1
 435
 436tcp_stdurg - BOOLEAN
 437        Use the Host requirements interpretation of the TCP urgent pointer field.
 438        Most hosts use the older BSD interpretation, so if you turn this on
 439        Linux might not communicate correctly with them.
 440        Default: FALSE
 441
 442tcp_synack_retries - INTEGER
 443        Number of times SYNACKs for a passive TCP connection attempt will
 444        be retransmitted. Should not be higher than 255. Default value
 445        is 5, which corresponds to 31seconds till the last retransmission
 446        with the current initial RTO of 1second. With this the final timeout
 447        for a passive TCP connection will happen after 63seconds.
 448
 449tcp_syncookies - BOOLEAN
 450        Only valid when the kernel was compiled with CONFIG_SYNCOOKIES
 451        Send out syncookies when the syn backlog queue of a socket
 452        overflows. This is to prevent against the common 'SYN flood attack'
 453        Default: FALSE
 454
 455        Note, that syncookies is fallback facility.
 456        It MUST NOT be used to help highly loaded servers to stand
 457        against legal connection rate. If you see SYN flood warnings
 458        in your logs, but investigation shows that they occur
 459        because of overload with legal connections, you should tune
 460        another parameters until this warning disappear.
 461        See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
 462
 463        syncookies seriously violate TCP protocol, do not allow
 464        to use TCP extensions, can result in serious degradation
 465        of some services (f.e. SMTP relaying), visible not by you,
 466        but your clients and relays, contacting you. While you see
 467        SYN flood warnings in logs not being really flooded, your server
 468        is seriously misconfigured.
 469
 470tcp_fastopen - INTEGER
 471        Enable TCP Fast Open feature (draft-ietf-tcpm-fastopen) to send data
 472        in the opening SYN packet. To use this feature, the client application
 473        must use sendmsg() or sendto() with MSG_FASTOPEN flag rather than
 474        connect() to perform a TCP handshake automatically.
 475
 476        The values (bitmap) are
 477        1: Enables sending data in the opening SYN on the client.
 478        2: Enables TCP Fast Open on the server side, i.e., allowing data in
 479           a SYN packet to be accepted and passed to the application before
 480           3-way hand shake finishes.
 481        4: Send data in the opening SYN regardless of cookie availability and
 482           without a cookie option.
 483        0x100: Accept SYN data w/o validating the cookie.
 484        0x200: Accept data-in-SYN w/o any cookie option present.
 485        0x400/0x800: Enable Fast Open on all listeners regardless of the
 486           TCP_FASTOPEN socket option. The two different flags designate two
 487           different ways of setting max_qlen without the TCP_FASTOPEN socket
 488           option.
 489
 490        Default: 0
 491
 492        Note that the client & server side Fast Open flags (1 and 2
 493        respectively) must be also enabled before the rest of flags can take
 494        effect.
 495
 496        See include/net/tcp.h and the code for more details.
 497
 498tcp_syn_retries - INTEGER
 499        Number of times initial SYNs for an active TCP connection attempt
 500        will be retransmitted. Should not be higher than 255. Default value
 501        is 6, which corresponds to 63seconds till the last retransmission
 502        with the current initial RTO of 1second. With this the final timeout
 503        for an active TCP connection attempt will happen after 127seconds.
 504
 505tcp_timestamps - BOOLEAN
 506        Enable timestamps as defined in RFC1323.
 507
 508tcp_tso_win_divisor - INTEGER
 509        This allows control over what percentage of the congestion window
 510        can be consumed by a single TSO frame.
 511        The setting of this parameter is a choice between burstiness and
 512        building larger TSO frames.
 513        Default: 3
 514
 515tcp_tw_recycle - BOOLEAN
 516        Enable fast recycling TIME-WAIT sockets. Default value is 0.
 517        It should not be changed without advice/request of technical
 518        experts.
 519
 520tcp_tw_reuse - BOOLEAN
 521        Allow to reuse TIME-WAIT sockets for new connections when it is
 522        safe from protocol viewpoint. Default value is 0.
 523        It should not be changed without advice/request of technical
 524        experts.
 525
 526tcp_window_scaling - BOOLEAN
 527        Enable window scaling as defined in RFC1323.
 528
 529tcp_wmem - vector of 3 INTEGERs: min, default, max
 530        min: Amount of memory reserved for send buffers for TCP sockets.
 531        Each TCP socket has rights to use it due to fact of its birth.
 532        Default: 1 page
 533
 534        default: initial size of send buffer used by TCP sockets.  This
 535        value overrides net.core.wmem_default used by other protocols.
 536        It is usually lower than net.core.wmem_default.
 537        Default: 16K
 538
 539        max: Maximal amount of memory allowed for automatically tuned
 540        send buffers for TCP sockets. This value does not override
 541        net.core.wmem_max.  Calling setsockopt() with SO_SNDBUF disables
 542        automatic tuning of that socket's send buffer size, in which case
 543        this value is ignored.
 544        Default: between 64K and 4MB, depending on RAM size.
 545
 546tcp_workaround_signed_windows - BOOLEAN
 547        If set, assume no receipt of a window scaling option means the
 548        remote TCP is broken and treats the window as a signed quantity.
 549        If unset, assume the remote TCP is not broken even if we do
 550        not receive a window scaling option from them.
 551        Default: 0
 552
 553tcp_dma_copybreak - INTEGER
 554        Lower limit, in bytes, of the size of socket reads that will be
 555        offloaded to a DMA copy engine, if one is present in the system
 556        and CONFIG_NET_DMA is enabled.
 557        Default: 4096
 558
 559tcp_thin_linear_timeouts - BOOLEAN
 560        Enable dynamic triggering of linear timeouts for thin streams.
 561        If set, a check is performed upon retransmission by timeout to
 562        determine if the stream is thin (less than 4 packets in flight).
 563        As long as the stream is found to be thin, up to 6 linear
 564        timeouts may be performed before exponential backoff mode is
 565        initiated. This improves retransmission latency for
 566        non-aggressive thin streams, often found to be time-dependent.
 567        For more information on thin streams, see
 568        Documentation/networking/tcp-thin.txt
 569        Default: 0
 570
 571tcp_thin_dupack - BOOLEAN
 572        Enable dynamic triggering of retransmissions after one dupACK
 573        for thin streams. If set, a check is performed upon reception
 574        of a dupACK to determine if the stream is thin (less than 4
 575        packets in flight). As long as the stream is found to be thin,
 576        data is retransmitted on the first received dupACK. This
 577        improves retransmission latency for non-aggressive thin
 578        streams, often found to be time-dependent.
 579        For more information on thin streams, see
 580        Documentation/networking/tcp-thin.txt
 581        Default: 0
 582
 583tcp_limit_output_bytes - INTEGER
 584        Controls TCP Small Queue limit per tcp socket.
 585        TCP bulk sender tends to increase packets in flight until it
 586        gets losses notifications. With SNDBUF autotuning, this can
 587        result in a large amount of packets queued in qdisc/device
 588        on the local machine, hurting latency of other flows, for
 589        typical pfifo_fast qdiscs.
 590        tcp_limit_output_bytes limits the number of bytes on qdisc
 591        or device to reduce artificial RTT/cwnd and reduce bufferbloat.
 592        Note: For GSO/TSO enabled flows, we try to have at least two
 593        packets in flight. Reducing tcp_limit_output_bytes might also
 594        reduce the size of individual GSO packet (64KB being the max)
 595        Default: 131072
 596
 597tcp_challenge_ack_limit - INTEGER
 598        Limits number of Challenge ACK sent per second, as recommended
 599        in RFC 5961 (Improving TCP's Robustness to Blind In-Window Attacks)
 600        Default: 100
 601
 602UDP variables:
 603
 604udp_mem - vector of 3 INTEGERs: min, pressure, max
 605        Number of pages allowed for queueing by all UDP sockets.
 606
 607        min: Below this number of pages UDP is not bothered about its
 608        memory appetite. When amount of memory allocated by UDP exceeds
 609        this number, UDP starts to moderate memory usage.
 610
 611        pressure: This value was introduced to follow format of tcp_mem.
 612
 613        max: Number of pages allowed for queueing by all UDP sockets.
 614
 615        Default is calculated at boot time from amount of available memory.
 616
 617udp_rmem_min - INTEGER
 618        Minimal size of receive buffer used by UDP sockets in moderation.
 619        Each UDP socket is able to use the size for receiving data, even if
 620        total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
 621        Default: 1 page
 622
 623udp_wmem_min - INTEGER
 624        Minimal size of send buffer used by UDP sockets in moderation.
 625        Each UDP socket is able to use the size for sending data, even if
 626        total pages of UDP sockets exceed udp_mem pressure. The unit is byte.
 627        Default: 1 page
 628
 629CIPSOv4 Variables:
 630
 631cipso_cache_enable - BOOLEAN
 632        If set, enable additions to and lookups from the CIPSO label mapping
 633        cache.  If unset, additions are ignored and lookups always result in a
 634        miss.  However, regardless of the setting the cache is still
 635        invalidated when required when means you can safely toggle this on and
 636        off and the cache will always be "safe".
 637        Default: 1
 638
 639cipso_cache_bucket_size - INTEGER
 640        The CIPSO label cache consists of a fixed size hash table with each
 641        hash bucket containing a number of cache entries.  This variable limits
 642        the number of entries in each hash bucket; the larger the value the
 643        more CIPSO label mappings that can be cached.  When the number of
 644        entries in a given hash bucket reaches this limit adding new entries
 645        causes the oldest entry in the bucket to be removed to make room.
 646        Default: 10
 647
 648cipso_rbm_optfmt - BOOLEAN
 649        Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
 650        the CIPSO draft specification (see Documentation/netlabel for details).
 651        This means that when set the CIPSO tag will be padded with empty
 652        categories in order to make the packet data 32-bit aligned.
 653        Default: 0
 654
 655cipso_rbm_structvalid - BOOLEAN
 656        If set, do a very strict check of the CIPSO option when
 657        ip_options_compile() is called.  If unset, relax the checks done during
 658        ip_options_compile().  Either way is "safe" as errors are caught else
 659        where in the CIPSO processing code but setting this to 0 (False) should
 660        result in less work (i.e. it should be faster) but could cause problems
 661        with other implementations that require strict checking.
 662        Default: 0
 663
 664IP Variables:
 665
 666ip_local_port_range - 2 INTEGERS
 667        Defines the local port range that is used by TCP and UDP to
 668        choose the local port. The first number is the first, the
 669        second the last local port number. The default values are
 670        32768 and 61000 respectively.
 671
 672ip_local_reserved_ports - list of comma separated ranges
 673        Specify the ports which are reserved for known third-party
 674        applications. These ports will not be used by automatic port
 675        assignments (e.g. when calling connect() or bind() with port
 676        number 0). Explicit port allocation behavior is unchanged.
 677
 678        The format used for both input and output is a comma separated
 679        list of ranges (e.g. "1,2-4,10-10" for ports 1, 2, 3, 4 and
 680        10). Writing to the file will clear all previously reserved
 681        ports and update the current list with the one given in the
 682        input.
 683
 684        Note that ip_local_port_range and ip_local_reserved_ports
 685        settings are independent and both are considered by the kernel
 686        when determining which ports are available for automatic port
 687        assignments.
 688
 689        You can reserve ports which are not in the current
 690        ip_local_port_range, e.g.:
 691
 692        $ cat /proc/sys/net/ipv4/ip_local_port_range
 693        32000   61000
 694        $ cat /proc/sys/net/ipv4/ip_local_reserved_ports
 695        8080,9148
 696
 697        although this is redundant. However such a setting is useful
 698        if later the port range is changed to a value that will
 699        include the reserved ports.
 700
 701        Default: Empty
 702
 703ip_nonlocal_bind - BOOLEAN
 704        If set, allows processes to bind() to non-local IP addresses,
 705        which can be quite useful - but may break some applications.
 706        Default: 0
 707
 708ip_dynaddr - BOOLEAN
 709        If set non-zero, enables support for dynamic addresses.
 710        If set to a non-zero value larger than 1, a kernel log
 711        message will be printed when dynamic address rewriting
 712        occurs.
 713        Default: 0
 714
 715icmp_echo_ignore_all - BOOLEAN
 716        If set non-zero, then the kernel will ignore all ICMP ECHO
 717        requests sent to it.
 718        Default: 0
 719
 720icmp_echo_ignore_broadcasts - BOOLEAN
 721        If set non-zero, then the kernel will ignore all ICMP ECHO and
 722        TIMESTAMP requests sent to it via broadcast/multicast.
 723        Default: 1
 724
 725icmp_ratelimit - INTEGER
 726        Limit the maximal rates for sending ICMP packets whose type matches
 727        icmp_ratemask (see below) to specific targets.
 728        0 to disable any limiting,
 729        otherwise the minimal space between responses in milliseconds.
 730        Default: 1000
 731
 732icmp_ratemask - INTEGER
 733        Mask made of ICMP types for which rates are being limited.
 734        Significant bits: IHGFEDCBA9876543210
 735        Default mask:     0000001100000011000 (6168)
 736
 737        Bit definitions (see include/linux/icmp.h):
 738                0 Echo Reply
 739                3 Destination Unreachable *
 740                4 Source Quench *
 741                5 Redirect
 742                8 Echo Request
 743                B Time Exceeded *
 744                C Parameter Problem *
 745                D Timestamp Request
 746                E Timestamp Reply
 747                F Info Request
 748                G Info Reply
 749                H Address Mask Request
 750                I Address Mask Reply
 751
 752        * These are rate limited by default (see default mask above)
 753
 754icmp_ignore_bogus_error_responses - BOOLEAN
 755        Some routers violate RFC1122 by sending bogus responses to broadcast
 756        frames.  Such violations are normally logged via a kernel warning.
 757        If this is set to TRUE, the kernel will not give such warnings, which
 758        will avoid log file clutter.
 759        Default: FALSE
 760
 761icmp_errors_use_inbound_ifaddr - BOOLEAN
 762
 763        If zero, icmp error messages are sent with the primary address of
 764        the exiting interface.
 765
 766        If non-zero, the message will be sent with the primary address of
 767        the interface that received the packet that caused the icmp error.
 768        This is the behaviour network many administrators will expect from
 769        a router. And it can make debugging complicated network layouts
 770        much easier.
 771
 772        Note that if no primary address exists for the interface selected,
 773        then the primary address of the first non-loopback interface that
 774        has one will be used regardless of this setting.
 775
 776        Default: 0
 777
 778igmp_max_memberships - INTEGER
 779        Change the maximum number of multicast groups we can subscribe to.
 780        Default: 20
 781
 782        Theoretical maximum value is bounded by having to send a membership
 783        report in a single datagram (i.e. the report can't span multiple
 784        datagrams, or risk confusing the switch and leaving groups you don't
 785        intend to).
 786
 787        The number of supported groups 'M' is bounded by the number of group
 788        report entries you can fit into a single datagram of 65535 bytes.
 789
 790        M = 65536-sizeof (ip header)/(sizeof(Group record))
 791
 792        Group records are variable length, with a minimum of 12 bytes.
 793        So net.ipv4.igmp_max_memberships should not be set higher than:
 794
 795        (65536-24) / 12 = 5459
 796
 797        The value 5459 assumes no IP header options, so in practice
 798        this number may be lower.
 799
 800        conf/interface/*  changes special settings per interface (where
 801        "interface" is the name of your network interface)
 802
 803        conf/all/*        is special, changes the settings for all interfaces
 804
 805log_martians - BOOLEAN
 806        Log packets with impossible addresses to kernel log.
 807        log_martians for the interface will be enabled if at least one of
 808        conf/{all,interface}/log_martians is set to TRUE,
 809        it will be disabled otherwise
 810
 811accept_redirects - BOOLEAN
 812        Accept ICMP redirect messages.
 813        accept_redirects for the interface will be enabled if:
 814        - both conf/{all,interface}/accept_redirects are TRUE in the case
 815          forwarding for the interface is enabled
 816        or
 817        - at least one of conf/{all,interface}/accept_redirects is TRUE in the
 818          case forwarding for the interface is disabled
 819        accept_redirects for the interface will be disabled otherwise
 820        default TRUE (host)
 821                FALSE (router)
 822
 823forwarding - BOOLEAN
 824        Enable IP forwarding on this interface.
 825
 826mc_forwarding - BOOLEAN
 827        Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
 828        and a multicast routing daemon is required.
 829        conf/all/mc_forwarding must also be set to TRUE to enable multicast
 830        routing for the interface
 831
 832medium_id - INTEGER
 833        Integer value used to differentiate the devices by the medium they
 834        are attached to. Two devices can have different id values when
 835        the broadcast packets are received only on one of them.
 836        The default value 0 means that the device is the only interface
 837        to its medium, value of -1 means that medium is not known.
 838
 839        Currently, it is used to change the proxy_arp behavior:
 840        the proxy_arp feature is enabled for packets forwarded between
 841        two devices attached to different media.
 842
 843proxy_arp - BOOLEAN
 844        Do proxy arp.
 845        proxy_arp for the interface will be enabled if at least one of
 846        conf/{all,interface}/proxy_arp is set to TRUE,
 847        it will be disabled otherwise
 848
 849proxy_arp_pvlan - BOOLEAN
 850        Private VLAN proxy arp.
 851        Basically allow proxy arp replies back to the same interface
 852        (from which the ARP request/solicitation was received).
 853
 854        This is done to support (ethernet) switch features, like RFC
 855        3069, where the individual ports are NOT allowed to
 856        communicate with each other, but they are allowed to talk to
 857        the upstream router.  As described in RFC 3069, it is possible
 858        to allow these hosts to communicate through the upstream
 859        router by proxy_arp'ing. Don't need to be used together with
 860        proxy_arp.
 861
 862        This technology is known by different names:
 863          In RFC 3069 it is called VLAN Aggregation.
 864          Cisco and Allied Telesyn call it Private VLAN.
 865          Hewlett-Packard call it Source-Port filtering or port-isolation.
 866          Ericsson call it MAC-Forced Forwarding (RFC Draft).
 867
 868shared_media - BOOLEAN
 869        Send(router) or accept(host) RFC1620 shared media redirects.
 870        Overrides ip_secure_redirects.
 871        shared_media for the interface will be enabled if at least one of
 872        conf/{all,interface}/shared_media is set to TRUE,
 873        it will be disabled otherwise
 874        default TRUE
 875
 876secure_redirects - BOOLEAN
 877        Accept ICMP redirect messages only for gateways,
 878        listed in default gateway list.
 879        secure_redirects for the interface will be enabled if at least one of
 880        conf/{all,interface}/secure_redirects is set to TRUE,
 881        it will be disabled otherwise
 882        default TRUE
 883
 884send_redirects - BOOLEAN
 885        Send redirects, if router.
 886        send_redirects for the interface will be enabled if at least one of
 887        conf/{all,interface}/send_redirects is set to TRUE,
 888        it will be disabled otherwise
 889        Default: TRUE
 890
 891bootp_relay - BOOLEAN
 892        Accept packets with source address 0.b.c.d destined
 893        not to this host as local ones. It is supposed, that
 894        BOOTP relay daemon will catch and forward such packets.
 895        conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
 896        for the interface
 897        default FALSE
 898        Not Implemented Yet.
 899
 900accept_source_route - BOOLEAN
 901        Accept packets with SRR option.
 902        conf/all/accept_source_route must also be set to TRUE to accept packets
 903        with SRR option on the interface
 904        default TRUE (router)
 905                FALSE (host)
 906
 907accept_local - BOOLEAN
 908        Accept packets with local source addresses. In combination
 909        with suitable routing, this can be used to direct packets
 910        between two local interfaces over the wire and have them
 911        accepted properly.
 912
 913        rp_filter must be set to a non-zero value in order for
 914        accept_local to have an effect.
 915
 916        default FALSE
 917
 918route_localnet - BOOLEAN
 919        Do not consider loopback addresses as martian source or destination
 920        while routing. This enables the use of 127/8 for local routing purposes.
 921        default FALSE
 922
 923rp_filter - INTEGER
 924        0 - No source validation.
 925        1 - Strict mode as defined in RFC3704 Strict Reverse Path
 926            Each incoming packet is tested against the FIB and if the interface
 927            is not the best reverse path the packet check will fail.
 928            By default failed packets are discarded.
 929        2 - Loose mode as defined in RFC3704 Loose Reverse Path
 930            Each incoming packet's source address is also tested against the FIB
 931            and if the source address is not reachable via any interface
 932            the packet check will fail.
 933
 934        Current recommended practice in RFC3704 is to enable strict mode
 935        to prevent IP spoofing from DDos attacks. If using asymmetric routing
 936        or other complicated routing, then loose mode is recommended.
 937
 938        The max value from conf/{all,interface}/rp_filter is used
 939        when doing source validation on the {interface}.
 940
 941        Default value is 0. Note that some distributions enable it
 942        in startup scripts.
 943
 944arp_filter - BOOLEAN
 945        1 - Allows you to have multiple network interfaces on the same
 946        subnet, and have the ARPs for each interface be answered
 947        based on whether or not the kernel would route a packet from
 948        the ARP'd IP out that interface (therefore you must use source
 949        based routing for this to work). In other words it allows control
 950        of which cards (usually 1) will respond to an arp request.
 951
 952        0 - (default) The kernel can respond to arp requests with addresses
 953        from other interfaces. This may seem wrong but it usually makes
 954        sense, because it increases the chance of successful communication.
 955        IP addresses are owned by the complete host on Linux, not by
 956        particular interfaces. Only for more complex setups like load-
 957        balancing, does this behaviour cause problems.
 958
 959        arp_filter for the interface will be enabled if at least one of
 960        conf/{all,interface}/arp_filter is set to TRUE,
 961        it will be disabled otherwise
 962
 963arp_announce - INTEGER
 964        Define different restriction levels for announcing the local
 965        source IP address from IP packets in ARP requests sent on
 966        interface:
 967        0 - (default) Use any local address, configured on any interface
 968        1 - Try to avoid local addresses that are not in the target's
 969        subnet for this interface. This mode is useful when target
 970        hosts reachable via this interface require the source IP
 971        address in ARP requests to be part of their logical network
 972        configured on the receiving interface. When we generate the
 973        request we will check all our subnets that include the
 974        target IP and will preserve the source address if it is from
 975        such subnet. If there is no such subnet we select source
 976        address according to the rules for level 2.
 977        2 - Always use the best local address for this target.
 978        In this mode we ignore the source address in the IP packet
 979        and try to select local address that we prefer for talks with
 980        the target host. Such local address is selected by looking
 981        for primary IP addresses on all our subnets on the outgoing
 982        interface that include the target IP address. If no suitable
 983        local address is found we select the first local address
 984        we have on the outgoing interface or on all other interfaces,
 985        with the hope we will receive reply for our request and
 986        even sometimes no matter the source IP address we announce.
 987
 988        The max value from conf/{all,interface}/arp_announce is used.
 989
 990        Increasing the restriction level gives more chance for
 991        receiving answer from the resolved target while decreasing
 992        the level announces more valid sender's information.
 993
 994arp_ignore - INTEGER
 995        Define different modes for sending replies in response to
 996        received ARP requests that resolve local target IP addresses:
 997        0 - (default): reply for any local target IP address, configured
 998        on any interface
 999        1 - reply only if the target IP address is local address
1000        configured on the incoming interface
1001        2 - reply only if the target IP address is local address
1002        configured on the incoming interface and both with the
1003        sender's IP address are part from same subnet on this interface
1004        3 - do not reply for local addresses configured with scope host,
1005        only resolutions for global and link addresses are replied
1006        4-7 - reserved
1007        8 - do not reply for all local addresses
1008
1009        The max value from conf/{all,interface}/arp_ignore is used
1010        when ARP request is received on the {interface}
1011
1012arp_notify - BOOLEAN
1013        Define mode for notification of address and device changes.
1014        0 - (default): do nothing
1015        1 - Generate gratuitous arp requests when device is brought up
1016            or hardware address changes.
1017
1018arp_accept - BOOLEAN
1019        Define behavior for gratuitous ARP frames who's IP is not
1020        already present in the ARP table:
1021        0 - don't create new entries in the ARP table
1022        1 - create new entries in the ARP table
1023
1024        Both replies and requests type gratuitous arp will trigger the
1025        ARP table to be updated, if this setting is on.
1026
1027        If the ARP table already contains the IP address of the
1028        gratuitous arp frame, the arp table will be updated regardless
1029        if this setting is on or off.
1030
1031
1032app_solicit - INTEGER
1033        The maximum number of probes to send to the user space ARP daemon
1034        via netlink before dropping back to multicast probes (see
1035        mcast_solicit).  Defaults to 0.
1036
1037disable_policy - BOOLEAN
1038        Disable IPSEC policy (SPD) for this interface
1039
1040disable_xfrm - BOOLEAN
1041        Disable IPSEC encryption on this interface, whatever the policy
1042
1043
1044
1045tag - INTEGER
1046        Allows you to write a number, which can be used as required.
1047        Default value is 0.
1048
1049Alexey Kuznetsov.
1050kuznet@ms2.inr.ac.ru
1051
1052Updated by:
1053Andi Kleen
1054ak@muc.de
1055Nicolas Delon
1056delon.nicolas@wanadoo.fr
1057
1058
1059
1060
1061/proc/sys/net/ipv6/* Variables:
1062
1063IPv6 has no global variables such as tcp_*.  tcp_* settings under ipv4/ also
1064apply to IPv6 [XXX?].
1065
1066bindv6only - BOOLEAN
1067        Default value for IPV6_V6ONLY socket option,
1068        which restricts use of the IPv6 socket to IPv6 communication
1069        only.
1070                TRUE: disable IPv4-mapped address feature
1071                FALSE: enable IPv4-mapped address feature
1072
1073        Default: FALSE (as specified in RFC3493)
1074
1075IPv6 Fragmentation:
1076
1077ip6frag_high_thresh - INTEGER
1078        Maximum memory used to reassemble IPv6 fragments. When
1079        ip6frag_high_thresh bytes of memory is allocated for this purpose,
1080        the fragment handler will toss packets until ip6frag_low_thresh
1081        is reached.
1082
1083ip6frag_low_thresh - INTEGER
1084        See ip6frag_high_thresh
1085
1086ip6frag_time - INTEGER
1087        Time in seconds to keep an IPv6 fragment in memory.
1088
1089ip6frag_secret_interval - INTEGER
1090        Regeneration interval (in seconds) of the hash secret (or lifetime
1091        for the hash secret) for IPv6 fragments.
1092        Default: 600
1093
1094conf/default/*:
1095        Change the interface-specific default settings.
1096
1097
1098conf/all/*:
1099        Change all the interface-specific settings.
1100
1101        [XXX:  Other special features than forwarding?]
1102
1103conf/all/forwarding - BOOLEAN
1104        Enable global IPv6 forwarding between all interfaces.
1105
1106        IPv4 and IPv6 work differently here; e.g. netfilter must be used
1107        to control which interfaces may forward packets and which not.
1108
1109        This also sets all interfaces' Host/Router setting
1110        'forwarding' to the specified value.  See below for details.
1111
1112        This referred to as global forwarding.
1113
1114proxy_ndp - BOOLEAN
1115        Do proxy ndp.
1116
1117conf/interface/*:
1118        Change special settings per interface.
1119
1120        The functional behaviour for certain settings is different
1121        depending on whether local forwarding is enabled or not.
1122
1123accept_ra - INTEGER
1124        Accept Router Advertisements; autoconfigure using them.
1125
1126        It also determines whether or not to transmit Router
1127        Solicitations. If and only if the functional setting is to
1128        accept Router Advertisements, Router Solicitations will be
1129        transmitted.
1130
1131        Possible values are:
1132                0 Do not accept Router Advertisements.
1133                1 Accept Router Advertisements if forwarding is disabled.
1134                2 Overrule forwarding behaviour. Accept Router Advertisements
1135                  even if forwarding is enabled.
1136
1137        Functional default: enabled if local forwarding is disabled.
1138                            disabled if local forwarding is enabled.
1139
1140accept_ra_defrtr - BOOLEAN
1141        Learn default router in Router Advertisement.
1142
1143        Functional default: enabled if accept_ra is enabled.
1144                            disabled if accept_ra is disabled.
1145
1146accept_ra_pinfo - BOOLEAN
1147        Learn Prefix Information in Router Advertisement.
1148
1149        Functional default: enabled if accept_ra is enabled.
1150                            disabled if accept_ra is disabled.
1151
1152accept_ra_rt_info_max_plen - INTEGER
1153        Maximum prefix length of Route Information in RA.
1154
1155        Route Information w/ prefix larger than or equal to this
1156        variable shall be ignored.
1157
1158        Functional default: 0 if accept_ra_rtr_pref is enabled.
1159                            -1 if accept_ra_rtr_pref is disabled.
1160
1161accept_ra_rtr_pref - BOOLEAN
1162        Accept Router Preference in RA.
1163
1164        Functional default: enabled if accept_ra is enabled.
1165                            disabled if accept_ra is disabled.
1166
1167accept_redirects - BOOLEAN
1168        Accept Redirects.
1169
1170        Functional default: enabled if local forwarding is disabled.
1171                            disabled if local forwarding is enabled.
1172
1173accept_source_route - INTEGER
1174        Accept source routing (routing extension header).
1175
1176        >= 0: Accept only routing header type 2.
1177        < 0: Do not accept routing header.
1178
1179        Default: 0
1180
1181autoconf - BOOLEAN
1182        Autoconfigure addresses using Prefix Information in Router
1183        Advertisements.
1184
1185        Functional default: enabled if accept_ra_pinfo is enabled.
1186                            disabled if accept_ra_pinfo is disabled.
1187
1188dad_transmits - INTEGER
1189        The amount of Duplicate Address Detection probes to send.
1190        Default: 1
1191
1192forwarding - INTEGER
1193        Configure interface-specific Host/Router behaviour.
1194
1195        Note: It is recommended to have the same setting on all
1196        interfaces; mixed router/host scenarios are rather uncommon.
1197
1198        Possible values are:
1199                0 Forwarding disabled
1200                1 Forwarding enabled
1201
1202        FALSE (0):
1203
1204        By default, Host behaviour is assumed.  This means:
1205
1206        1. IsRouter flag is not set in Neighbour Advertisements.
1207        2. If accept_ra is TRUE (default), transmit Router
1208           Solicitations.
1209        3. If accept_ra is TRUE (default), accept Router
1210           Advertisements (and do autoconfiguration).
1211        4. If accept_redirects is TRUE (default), accept Redirects.
1212
1213        TRUE (1):
1214
1215        If local forwarding is enabled, Router behaviour is assumed.
1216        This means exactly the reverse from the above:
1217
1218        1. IsRouter flag is set in Neighbour Advertisements.
1219        2. Router Solicitations are not sent unless accept_ra is 2.
1220        3. Router Advertisements are ignored unless accept_ra is 2.
1221        4. Redirects are ignored.
1222
1223        Default: 0 (disabled) if global forwarding is disabled (default),
1224                 otherwise 1 (enabled).
1225
1226hop_limit - INTEGER
1227        Default Hop Limit to set.
1228        Default: 64
1229
1230mtu - INTEGER
1231        Default Maximum Transfer Unit
1232        Default: 1280 (IPv6 required minimum)
1233
1234router_probe_interval - INTEGER
1235        Minimum interval (in seconds) between Router Probing described
1236        in RFC4191.
1237
1238        Default: 60
1239
1240router_solicitation_delay - INTEGER
1241        Number of seconds to wait after interface is brought up
1242        before sending Router Solicitations.
1243        Default: 1
1244
1245router_solicitation_interval - INTEGER
1246        Number of seconds to wait between Router Solicitations.
1247        Default: 4
1248
1249router_solicitations - INTEGER
1250        Number of Router Solicitations to send until assuming no
1251        routers are present.
1252        Default: 3
1253
1254use_tempaddr - INTEGER
1255        Preference for Privacy Extensions (RFC3041).
1256          <= 0 : disable Privacy Extensions
1257          == 1 : enable Privacy Extensions, but prefer public
1258                 addresses over temporary addresses.
1259          >  1 : enable Privacy Extensions and prefer temporary
1260                 addresses over public addresses.
1261        Default:  0 (for most devices)
1262                 -1 (for point-to-point devices and loopback devices)
1263
1264temp_valid_lft - INTEGER
1265        valid lifetime (in seconds) for temporary addresses.
1266        Default: 604800 (7 days)
1267
1268temp_prefered_lft - INTEGER
1269        Preferred lifetime (in seconds) for temporary addresses.
1270        Default: 86400 (1 day)
1271
1272max_desync_factor - INTEGER
1273        Maximum value for DESYNC_FACTOR, which is a random value
1274        that ensures that clients don't synchronize with each
1275        other and generate new addresses at exactly the same time.
1276        value is in seconds.
1277        Default: 600
1278
1279regen_max_retry - INTEGER
1280        Number of attempts before give up attempting to generate
1281        valid temporary addresses.
1282        Default: 5
1283
1284max_addresses - INTEGER
1285        Maximum number of autoconfigured addresses per interface.  Setting
1286        to zero disables the limitation.  It is not recommended to set this
1287        value too large (or to zero) because it would be an easy way to
1288        crash the kernel by allowing too many addresses to be created.
1289        Default: 16
1290
1291disable_ipv6 - BOOLEAN
1292        Disable IPv6 operation.  If accept_dad is set to 2, this value
1293        will be dynamically set to TRUE if DAD fails for the link-local
1294        address.
1295        Default: FALSE (enable IPv6 operation)
1296
1297        When this value is changed from 1 to 0 (IPv6 is being enabled),
1298        it will dynamically create a link-local address on the given
1299        interface and start Duplicate Address Detection, if necessary.
1300
1301        When this value is changed from 0 to 1 (IPv6 is being disabled),
1302        it will dynamically delete all address on the given interface.
1303
1304accept_dad - INTEGER
1305        Whether to accept DAD (Duplicate Address Detection).
1306        0: Disable DAD
1307        1: Enable DAD (default)
1308        2: Enable DAD, and disable IPv6 operation if MAC-based duplicate
1309           link-local address has been found.
1310
1311force_tllao - BOOLEAN
1312        Enable sending the target link-layer address option even when
1313        responding to a unicast neighbor solicitation.
1314        Default: FALSE
1315
1316        Quoting from RFC 2461, section 4.4, Target link-layer address:
1317
1318        "The option MUST be included for multicast solicitations in order to
1319        avoid infinite Neighbor Solicitation "recursion" when the peer node
1320        does not have a cache entry to return a Neighbor Advertisements
1321        message.  When responding to unicast solicitations, the option can be
1322        omitted since the sender of the solicitation has the correct link-
1323        layer address; otherwise it would not have be able to send the unicast
1324        solicitation in the first place. However, including the link-layer
1325        address in this case adds little overhead and eliminates a potential
1326        race condition where the sender deletes the cached link-layer address
1327        prior to receiving a response to a previous solicitation."
1328
1329ndisc_notify - BOOLEAN
1330        Define mode for notification of address and device changes.
1331        0 - (default): do nothing
1332        1 - Generate unsolicited neighbour advertisements when device is brought
1333            up or hardware address changes.
1334
1335icmp/*:
1336ratelimit - INTEGER
1337        Limit the maximal rates for sending ICMPv6 packets.
1338        0 to disable any limiting,
1339        otherwise the minimal space between responses in milliseconds.
1340        Default: 1000
1341
1342
1343IPv6 Update by:
1344Pekka Savola <pekkas@netcore.fi>
1345YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
1346
1347
1348/proc/sys/net/bridge/* Variables:
1349
1350bridge-nf-call-arptables - BOOLEAN
1351        1 : pass bridged ARP traffic to arptables' FORWARD chain.
1352        0 : disable this.
1353        Default: 1
1354
1355bridge-nf-call-iptables - BOOLEAN
1356        1 : pass bridged IPv4 traffic to iptables' chains.
1357        0 : disable this.
1358        Default: 1
1359
1360bridge-nf-call-ip6tables - BOOLEAN
1361        1 : pass bridged IPv6 traffic to ip6tables' chains.
1362        0 : disable this.
1363        Default: 1
1364
1365bridge-nf-filter-vlan-tagged - BOOLEAN
1366        1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables.
1367        0 : disable this.
1368        Default: 0
1369
1370bridge-nf-filter-pppoe-tagged - BOOLEAN
1371        1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables.
1372        0 : disable this.
1373        Default: 0
1374
1375bridge-nf-pass-vlan-input-dev - BOOLEAN
1376        1: if bridge-nf-filter-vlan-tagged is enabled, try to find a vlan
1377        interface on the bridge and set the netfilter input device to the vlan.
1378        This allows use of e.g. "iptables -i br0.1" and makes the REDIRECT
1379        target work with vlan-on-top-of-bridge interfaces.  When no matching
1380        vlan interface is found, or this switch is off, the input device is
1381        set to the bridge interface.
1382        0: disable bridge netfilter vlan interface lookup.
1383        Default: 0
1384
1385proc/sys/net/sctp/* Variables:
1386
1387addip_enable - BOOLEAN
1388        Enable or disable extension of  Dynamic Address Reconfiguration
1389        (ADD-IP) functionality specified in RFC5061.  This extension provides
1390        the ability to dynamically add and remove new addresses for the SCTP
1391        associations.
1392
1393        1: Enable extension.
1394
1395        0: Disable extension.
1396
1397        Default: 0
1398
1399addip_noauth_enable - BOOLEAN
1400        Dynamic Address Reconfiguration (ADD-IP) requires the use of
1401        authentication to protect the operations of adding or removing new
1402        addresses.  This requirement is mandated so that unauthorized hosts
1403        would not be able to hijack associations.  However, older
1404        implementations may not have implemented this requirement while
1405        allowing the ADD-IP extension.  For reasons of interoperability,
1406        we provide this variable to control the enforcement of the
1407        authentication requirement.
1408
1409        1: Allow ADD-IP extension to be used without authentication.  This
1410           should only be set in a closed environment for interoperability
1411           with older implementations.
1412
1413        0: Enforce the authentication requirement
1414
1415        Default: 0
1416
1417auth_enable - BOOLEAN
1418        Enable or disable Authenticated Chunks extension.  This extension
1419        provides the ability to send and receive authenticated chunks and is
1420        required for secure operation of Dynamic Address Reconfiguration
1421        (ADD-IP) extension.
1422
1423        1: Enable this extension.
1424        0: Disable this extension.
1425
1426        Default: 0
1427
1428prsctp_enable - BOOLEAN
1429        Enable or disable the Partial Reliability extension (RFC3758) which
1430        is used to notify peers that a given DATA should no longer be expected.
1431
1432        1: Enable extension
1433        0: Disable
1434
1435        Default: 1
1436
1437max_burst - INTEGER
1438        The limit of the number of new packets that can be initially sent.  It
1439        controls how bursty the generated traffic can be.
1440
1441        Default: 4
1442
1443association_max_retrans - INTEGER
1444        Set the maximum number for retransmissions that an association can
1445        attempt deciding that the remote end is unreachable.  If this value
1446        is exceeded, the association is terminated.
1447
1448        Default: 10
1449
1450max_init_retransmits - INTEGER
1451        The maximum number of retransmissions of INIT and COOKIE-ECHO chunks
1452        that an association will attempt before declaring the destination
1453        unreachable and terminating.
1454
1455        Default: 8
1456
1457path_max_retrans - INTEGER
1458        The maximum number of retransmissions that will be attempted on a given
1459        path.  Once this threshold is exceeded, the path is considered
1460        unreachable, and new traffic will use a different path when the
1461        association is multihomed.
1462
1463        Default: 5
1464
1465pf_retrans - INTEGER
1466        The number of retransmissions that will be attempted on a given path
1467        before traffic is redirected to an alternate transport (should one
1468        exist).  Note this is distinct from path_max_retrans, as a path that
1469        passes the pf_retrans threshold can still be used.  Its only
1470        deprioritized when a transmission path is selected by the stack.  This
1471        setting is primarily used to enable fast failover mechanisms without
1472        having to reduce path_max_retrans to a very low value.  See:
1473        http://www.ietf.org/id/draft-nishida-tsvwg-sctp-failover-05.txt
1474        for details.  Note also that a value of pf_retrans > path_max_retrans
1475        disables this feature
1476
1477        Default: 0
1478
1479rto_initial - INTEGER
1480        The initial round trip timeout value in milliseconds that will be used
1481        in calculating round trip times.  This is the initial time interval
1482        for retransmissions.
1483
1484        Default: 3000
1485
1486rto_max - INTEGER
1487        The maximum value (in milliseconds) of the round trip timeout.  This
1488        is the largest time interval that can elapse between retransmissions.
1489
1490        Default: 60000
1491
1492rto_min - INTEGER
1493        The minimum value (in milliseconds) of the round trip timeout.  This
1494        is the smallest time interval the can elapse between retransmissions.
1495
1496        Default: 1000
1497
1498hb_interval - INTEGER
1499        The interval (in milliseconds) between HEARTBEAT chunks.  These chunks
1500        are sent at the specified interval on idle paths to probe the state of
1501        a given path between 2 associations.
1502
1503        Default: 30000
1504
1505sack_timeout - INTEGER
1506        The amount of time (in milliseconds) that the implementation will wait
1507        to send a SACK.
1508
1509        Default: 200
1510
1511valid_cookie_life - INTEGER
1512        The default lifetime of the SCTP cookie (in milliseconds).  The cookie
1513        is used during association establishment.
1514
1515        Default: 60000
1516
1517cookie_preserve_enable - BOOLEAN
1518        Enable or disable the ability to extend the lifetime of the SCTP cookie
1519        that is used during the establishment phase of SCTP association
1520
1521        1: Enable cookie lifetime extension.
1522        0: Disable
1523
1524        Default: 1
1525
1526cookie_hmac_alg - STRING
1527        Select the hmac algorithm used when generating the cookie value sent by
1528        a listening sctp socket to a connecting client in the INIT-ACK chunk.
1529        Valid values are:
1530        * md5
1531        * sha1
1532        * none
1533        Ability to assign md5 or sha1 as the selected alg is predicated on the
1534        configuration of those algorithms at build time (CONFIG_CRYPTO_MD5 and
1535        CONFIG_CRYPTO_SHA1).
1536
1537        Default: Dependent on configuration.  MD5 if available, else SHA1 if
1538        available, else none.
1539
1540rcvbuf_policy - INTEGER
1541        Determines if the receive buffer is attributed to the socket or to
1542        association.   SCTP supports the capability to create multiple
1543        associations on a single socket.  When using this capability, it is
1544        possible that a single stalled association that's buffering a lot
1545        of data may block other associations from delivering their data by
1546        consuming all of the receive buffer space.  To work around this,
1547        the rcvbuf_policy could be set to attribute the receiver buffer space
1548        to each association instead of the socket.  This prevents the described
1549        blocking.
1550
1551        1: rcvbuf space is per association
1552        0: rcvbuf space is per socket
1553
1554        Default: 0
1555
1556sndbuf_policy - INTEGER
1557        Similar to rcvbuf_policy above, this applies to send buffer space.
1558
1559        1: Send buffer is tracked per association
1560        0: Send buffer is tracked per socket.
1561
1562        Default: 0
1563
1564sctp_mem - vector of 3 INTEGERs: min, pressure, max
1565        Number of pages allowed for queueing by all SCTP sockets.
1566
1567        min: Below this number of pages SCTP is not bothered about its
1568        memory appetite. When amount of memory allocated by SCTP exceeds
1569        this number, SCTP starts to moderate memory usage.
1570
1571        pressure: This value was introduced to follow format of tcp_mem.
1572
1573        max: Number of pages allowed for queueing by all SCTP sockets.
1574
1575        Default is calculated at boot time from amount of available memory.
1576
1577sctp_rmem - vector of 3 INTEGERs: min, default, max
1578        Only the first value ("min") is used, "default" and "max" are
1579        ignored.
1580
1581        min: Minimal size of receive buffer used by SCTP socket.
1582        It is guaranteed to each SCTP socket (but not association) even
1583        under moderate memory pressure.
1584
1585        Default: 1 page
1586
1587sctp_wmem  - vector of 3 INTEGERs: min, default, max
1588        Currently this tunable has no effect.
1589
1590addr_scope_policy - INTEGER
1591        Control IPv4 address scoping - draft-stewart-tsvwg-sctp-ipv4-00
1592
1593        0   - Disable IPv4 address scoping
1594        1   - Enable IPv4 address scoping
1595        2   - Follow draft but allow IPv4 private addresses
1596        3   - Follow draft but allow IPv4 link local addresses
1597
1598        Default: 1
1599
1600
1601/proc/sys/net/core/*
1602        Please see: Documentation/sysctl/net.txt for descriptions of these entries.
1603
1604
1605/proc/sys/net/unix/*
1606max_dgram_qlen - INTEGER
1607        The maximum length of dgram socket receive queue
1608
1609        Default: 10
1610
1611
1612UNDOCUMENTED:
1613
1614/proc/sys/net/irda/*
1615        fast_poll_increase FIXME
1616        warn_noreply_time FIXME
1617        discovery_slots FIXME
1618        slot_timeout FIXME
1619        max_baud_rate FIXME
1620        discovery_timeout FIXME
1621        lap_keepalive_time FIXME
1622        max_noreply_time FIXME
1623        max_tx_data_size FIXME
1624        max_tx_window FIXME
1625        min_tx_turn_time FIXME
1626
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.