ip-sysctl.txt 35 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023
  1. /proc/sys/net/ipv4/* Variables:
  2. ip_forward - BOOLEAN
  3. 0 - disabled (default)
  4. not 0 - enabled
  5. Forward Packets between interfaces.
  6. This variable is special, its change resets all configuration
  7. parameters to their default state (RFC1122 for hosts, RFC1812
  8. for routers)
  9. ip_default_ttl - INTEGER
  10. default 64
  11. ip_no_pmtu_disc - BOOLEAN
  12. Disable Path MTU Discovery.
  13. default FALSE
  14. min_pmtu - INTEGER
  15. default 562 - minimum discovered Path MTU
  16. mtu_expires - INTEGER
  17. Time, in seconds, that cached PMTU information is kept.
  18. min_adv_mss - INTEGER
  19. The advertised MSS depends on the first hop route MTU, but will
  20. never be lower than this setting.
  21. IP Fragmentation:
  22. ipfrag_high_thresh - INTEGER
  23. Maximum memory used to reassemble IP fragments. When
  24. ipfrag_high_thresh bytes of memory is allocated for this purpose,
  25. the fragment handler will toss packets until ipfrag_low_thresh
  26. is reached.
  27. ipfrag_low_thresh - INTEGER
  28. See ipfrag_high_thresh
  29. ipfrag_time - INTEGER
  30. Time in seconds to keep an IP fragment in memory.
  31. ipfrag_secret_interval - INTEGER
  32. Regeneration interval (in seconds) of the hash secret (or lifetime
  33. for the hash secret) for IP fragments.
  34. Default: 600
  35. ipfrag_max_dist - INTEGER
  36. ipfrag_max_dist is a non-negative integer value which defines the
  37. maximum "disorder" which is allowed among fragments which share a
  38. common IP source address. Note that reordering of packets is
  39. not unusual, but if a large number of fragments arrive from a source
  40. IP address while a particular fragment queue remains incomplete, it
  41. probably indicates that one or more fragments belonging to that queue
  42. have been lost. When ipfrag_max_dist is positive, an additional check
  43. is done on fragments before they are added to a reassembly queue - if
  44. ipfrag_max_dist (or more) fragments have arrived from a particular IP
  45. address between additions to any IP fragment queue using that source
  46. address, it's presumed that one or more fragments in the queue are
  47. lost. The existing fragment queue will be dropped, and a new one
  48. started. An ipfrag_max_dist value of zero disables this check.
  49. Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can
  50. result in unnecessarily dropping fragment queues when normal
  51. reordering of packets occurs, which could lead to poor application
  52. performance. Using a very large value, e.g. 50000, increases the
  53. likelihood of incorrectly reassembling IP fragments that originate
  54. from different IP datagrams, which could result in data corruption.
  55. Default: 64
  56. INET peer storage:
  57. inet_peer_threshold - INTEGER
  58. The approximate size of the storage. Starting from this threshold
  59. entries will be thrown aggressively. This threshold also determines
  60. entries' time-to-live and time intervals between garbage collection
  61. passes. More entries, less time-to-live, less GC interval.
  62. inet_peer_minttl - INTEGER
  63. Minimum time-to-live of entries. Should be enough to cover fragment
  64. time-to-live on the reassembling side. This minimum time-to-live is
  65. guaranteed if the pool size is less than inet_peer_threshold.
  66. Measured in jiffies(1).
  67. inet_peer_maxttl - INTEGER
  68. Maximum time-to-live of entries. Unused entries will expire after
  69. this period of time if there is no memory pressure on the pool (i.e.
  70. when the number of entries in the pool is very small).
  71. Measured in jiffies(1).
  72. inet_peer_gc_mintime - INTEGER
  73. Minimum interval between garbage collection passes. This interval is
  74. in effect under high memory pressure on the pool.
  75. Measured in jiffies(1).
  76. inet_peer_gc_maxtime - INTEGER
  77. Minimum interval between garbage collection passes. This interval is
  78. in effect under low (or absent) memory pressure on the pool.
  79. Measured in jiffies(1).
  80. TCP variables:
  81. somaxconn - INTEGER
  82. Limit of socket listen() backlog, known in userspace as SOMAXCONN.
  83. Defaults to 128. See also tcp_max_syn_backlog for additional tuning
  84. for TCP sockets.
  85. tcp_abc - INTEGER
  86. Controls Appropriate Byte Count (ABC) defined in RFC3465.
  87. ABC is a way of increasing congestion window (cwnd) more slowly
  88. in response to partial acknowledgments.
  89. Possible values are:
  90. 0 increase cwnd once per acknowledgment (no ABC)
  91. 1 increase cwnd once per acknowledgment of full sized segment
  92. 2 allow increase cwnd by two if acknowledgment is
  93. of two segments to compensate for delayed acknowledgments.
  94. Default: 0 (off)
  95. tcp_abort_on_overflow - BOOLEAN
  96. If listening service is too slow to accept new connections,
  97. reset them. Default state is FALSE. It means that if overflow
  98. occurred due to a burst, connection will recover. Enable this
  99. option _only_ if you are really sure that listening daemon
  100. cannot be tuned to accept connections faster. Enabling this
  101. option can harm clients of your server.
  102. tcp_adv_win_scale - INTEGER
  103. Count buffering overhead as bytes/2^tcp_adv_win_scale
  104. (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale),
  105. if it is <= 0.
  106. Default: 2
  107. tcp_allowed_congestion_control - STRING
  108. Show/set the congestion control choices available to non-privileged
  109. processes. The list is a subset of those listed in
  110. tcp_available_congestion_control.
  111. Default is "reno" and the default setting (tcp_congestion_control).
  112. tcp_app_win - INTEGER
  113. Reserve max(window/2^tcp_app_win, mss) of window for application
  114. buffer. Value 0 is special, it means that nothing is reserved.
  115. Default: 31
  116. tcp_available_congestion_control - STRING
  117. Shows the available congestion control choices that are registered.
  118. More congestion control algorithms may be available as modules,
  119. but not loaded.
  120. tcp_base_mss - INTEGER
  121. The initial value of search_low to be used by Packetization Layer
  122. Path MTU Discovery (MTU probing). If MTU probing is enabled,
  123. this is the inital MSS used by the connection.
  124. tcp_congestion_control - STRING
  125. Set the congestion control algorithm to be used for new
  126. connections. The algorithm "reno" is always available, but
  127. additional choices may be available based on kernel configuration.
  128. Default is set as part of kernel configuration.
  129. tcp_dsack - BOOLEAN
  130. Allows TCP to send "duplicate" SACKs.
  131. tcp_ecn - BOOLEAN
  132. Enable Explicit Congestion Notification in TCP.
  133. tcp_fack - BOOLEAN
  134. Enable FACK congestion avoidance and fast retransmission.
  135. The value is not used, if tcp_sack is not enabled.
  136. tcp_fin_timeout - INTEGER
  137. Time to hold socket in state FIN-WAIT-2, if it was closed
  138. by our side. Peer can be broken and never close its side,
  139. or even died unexpectedly. Default value is 60sec.
  140. Usual value used in 2.2 was 180 seconds, you may restore
  141. it, but remember that if your machine is even underloaded WEB server,
  142. you risk to overflow memory with kilotons of dead sockets,
  143. FIN-WAIT-2 sockets are less dangerous than FIN-WAIT-1,
  144. because they eat maximum 1.5K of memory, but they tend
  145. to live longer. Cf. tcp_max_orphans.
  146. tcp_frto - BOOLEAN
  147. Enables F-RTO, an enhanced recovery algorithm for TCP retransmission
  148. timeouts. It is particularly beneficial in wireless environments
  149. where packet loss is typically due to random radio interference
  150. rather than intermediate router congestion.
  151. tcp_keepalive_time - INTEGER
  152. How often TCP sends out keepalive messages when keepalive is enabled.
  153. Default: 2hours.
  154. tcp_keepalive_probes - INTEGER
  155. How many keepalive probes TCP sends out, until it decides that the
  156. connection is broken. Default value: 9.
  157. tcp_keepalive_intvl - INTEGER
  158. How frequently the probes are send out. Multiplied by
  159. tcp_keepalive_probes it is time to kill not responding connection,
  160. after probes started. Default value: 75sec i.e. connection
  161. will be aborted after ~11 minutes of retries.
  162. tcp_low_latency - BOOLEAN
  163. If set, the TCP stack makes decisions that prefer lower
  164. latency as opposed to higher throughput. By default, this
  165. option is not set meaning that higher throughput is preferred.
  166. An example of an application where this default should be
  167. changed would be a Beowulf compute cluster.
  168. Default: 0
  169. tcp_max_orphans - INTEGER
  170. Maximal number of TCP sockets not attached to any user file handle,
  171. held by system. If this number is exceeded orphaned connections are
  172. reset immediately and warning is printed. This limit exists
  173. only to prevent simple DoS attacks, you _must_ not rely on this
  174. or lower the limit artificially, but rather increase it
  175. (probably, after increasing installed memory),
  176. if network conditions require more than default value,
  177. and tune network services to linger and kill such states
  178. more aggressively. Let me to remind again: each orphan eats
  179. up to ~64K of unswappable memory.
  180. tcp_max_syn_backlog - INTEGER
  181. Maximal number of remembered connection requests, which are
  182. still did not receive an acknowledgment from connecting client.
  183. Default value is 1024 for systems with more than 128Mb of memory,
  184. and 128 for low memory machines. If server suffers of overload,
  185. try to increase this number.
  186. tcp_max_tw_buckets - INTEGER
  187. Maximal number of timewait sockets held by system simultaneously.
  188. If this number is exceeded time-wait socket is immediately destroyed
  189. and warning is printed. This limit exists only to prevent
  190. simple DoS attacks, you _must_ not lower the limit artificially,
  191. but rather increase it (probably, after increasing installed memory),
  192. if network conditions require more than default value.
  193. tcp_mem - vector of 3 INTEGERs: min, pressure, max
  194. min: below this number of pages TCP is not bothered about its
  195. memory appetite.
  196. pressure: when amount of memory allocated by TCP exceeds this number
  197. of pages, TCP moderates its memory consumption and enters memory
  198. pressure mode, which is exited when memory consumption falls
  199. under "min".
  200. max: number of pages allowed for queueing by all TCP sockets.
  201. Defaults are calculated at boot time from amount of available
  202. memory.
  203. tcp_moderate_rcvbuf - BOOLEAN
  204. If set, TCP performs receive buffer autotuning, attempting to
  205. automatically size the buffer (no greater than tcp_rmem[2]) to
  206. match the size required by the path for full throughput. Enabled by
  207. default.
  208. tcp_mtu_probing - INTEGER
  209. Controls TCP Packetization-Layer Path MTU Discovery. Takes three
  210. values:
  211. 0 - Disabled
  212. 1 - Disabled by default, enabled when an ICMP black hole detected
  213. 2 - Always enabled, use initial MSS of tcp_base_mss.
  214. tcp_no_metrics_save - BOOLEAN
  215. By default, TCP saves various connection metrics in the route cache
  216. when the connection closes, so that connections established in the
  217. near future can use these to set initial conditions. Usually, this
  218. increases overall performance, but may sometimes cause performance
  219. degredation. If set, TCP will not cache metrics on closing
  220. connections.
  221. tcp_orphan_retries - INTEGER
  222. How may times to retry before killing TCP connection, closed
  223. by our side. Default value 7 corresponds to ~50sec-16min
  224. depending on RTO. If you machine is loaded WEB server,
  225. you should think about lowering this value, such sockets
  226. may consume significant resources. Cf. tcp_max_orphans.
  227. tcp_reordering - INTEGER
  228. Maximal reordering of packets in a TCP stream.
  229. Default: 3
  230. tcp_retrans_collapse - BOOLEAN
  231. Bug-to-bug compatibility with some broken printers.
  232. On retransmit try to send bigger packets to work around bugs in
  233. certain TCP stacks.
  234. tcp_retries1 - INTEGER
  235. How many times to retry before deciding that something is wrong
  236. and it is necessary to report this suspicion to network layer.
  237. Minimal RFC value is 3, it is default, which corresponds
  238. to ~3sec-8min depending on RTO.
  239. tcp_retries2 - INTEGER
  240. How may times to retry before killing alive TCP connection.
  241. RFC1122 says that the limit should be longer than 100 sec.
  242. It is too small number. Default value 15 corresponds to ~13-30min
  243. depending on RTO.
  244. tcp_rfc1337 - BOOLEAN
  245. If set, the TCP stack behaves conforming to RFC1337. If unset,
  246. we are not conforming to RFC, but prevent TCP TIME_WAIT
  247. assassination.
  248. Default: 0
  249. tcp_rmem - vector of 3 INTEGERs: min, default, max
  250. min: Minimal size of receive buffer used by TCP sockets.
  251. It is guaranteed to each TCP socket, even under moderate memory
  252. pressure.
  253. Default: 8K
  254. default: default size of receive buffer used by TCP sockets.
  255. This value overrides net.core.rmem_default used by other protocols.
  256. Default: 87380 bytes. This value results in window of 65535 with
  257. default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit
  258. less for default tcp_app_win. See below about these variables.
  259. max: maximal size of receive buffer allowed for automatically
  260. selected receiver buffers for TCP socket. This value does not override
  261. net.core.rmem_max, "static" selection via SO_RCVBUF does not use this.
  262. Default: 87380*2 bytes.
  263. tcp_sack - BOOLEAN
  264. Enable select acknowledgments (SACKS).
  265. tcp_slow_start_after_idle - BOOLEAN
  266. If set, provide RFC2861 behavior and time out the congestion
  267. window after an idle period. An idle period is defined at
  268. the current RTO. If unset, the congestion window will not
  269. be timed out after an idle period.
  270. Default: 1
  271. tcp_stdurg - BOOLEAN
  272. Use the Host requirements interpretation of the TCP urg pointer field.
  273. Most hosts use the older BSD interpretation, so if you turn this on
  274. Linux might not communicate correctly with them.
  275. Default: FALSE
  276. tcp_synack_retries - INTEGER
  277. Number of times SYNACKs for a passive TCP connection attempt will
  278. be retransmitted. Should not be higher than 255. Default value
  279. is 5, which corresponds to ~180seconds.
  280. tcp_syncookies - BOOLEAN
  281. Only valid when the kernel was compiled with CONFIG_SYNCOOKIES
  282. Send out syncookies when the syn backlog queue of a socket
  283. overflows. This is to prevent against the common 'syn flood attack'
  284. Default: FALSE
  285. Note, that syncookies is fallback facility.
  286. It MUST NOT be used to help highly loaded servers to stand
  287. against legal connection rate. If you see synflood warnings
  288. in your logs, but investigation shows that they occur
  289. because of overload with legal connections, you should tune
  290. another parameters until this warning disappear.
  291. See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow.
  292. syncookies seriously violate TCP protocol, do not allow
  293. to use TCP extensions, can result in serious degradation
  294. of some services (f.e. SMTP relaying), visible not by you,
  295. but your clients and relays, contacting you. While you see
  296. synflood warnings in logs not being really flooded, your server
  297. is seriously misconfigured.
  298. tcp_syn_retries - INTEGER
  299. Number of times initial SYNs for an active TCP connection attempt
  300. will be retransmitted. Should not be higher than 255. Default value
  301. is 5, which corresponds to ~180seconds.
  302. tcp_timestamps - BOOLEAN
  303. Enable timestamps as defined in RFC1323.
  304. tcp_tso_win_divisor - INTEGER
  305. This allows control over what percentage of the congestion window
  306. can be consumed by a single TSO frame.
  307. The setting of this parameter is a choice between burstiness and
  308. building larger TSO frames.
  309. Default: 3
  310. tcp_tw_recycle - BOOLEAN
  311. Enable fast recycling TIME-WAIT sockets. Default value is 0.
  312. It should not be changed without advice/request of technical
  313. experts.
  314. tcp_tw_reuse - BOOLEAN
  315. Allow to reuse TIME-WAIT sockets for new connections when it is
  316. safe from protocol viewpoint. Default value is 0.
  317. It should not be changed without advice/request of technical
  318. experts.
  319. tcp_window_scaling - BOOLEAN
  320. Enable window scaling as defined in RFC1323.
  321. tcp_wmem - vector of 3 INTEGERs: min, default, max
  322. min: Amount of memory reserved for send buffers for TCP socket.
  323. Each TCP socket has rights to use it due to fact of its birth.
  324. Default: 4K
  325. default: Amount of memory allowed for send buffers for TCP socket
  326. by default. This value overrides net.core.wmem_default used
  327. by other protocols, it is usually lower than net.core.wmem_default.
  328. Default: 16K
  329. max: Maximal amount of memory allowed for automatically selected
  330. send buffers for TCP socket. This value does not override
  331. net.core.wmem_max, "static" selection via SO_SNDBUF does not use this.
  332. Default: 128K
  333. tcp_workaround_signed_windows - BOOLEAN
  334. If set, assume no receipt of a window scaling option means the
  335. remote TCP is broken and treats the window as a signed quantity.
  336. If unset, assume the remote TCP is not broken even if we do
  337. not receive a window scaling option from them.
  338. Default: 0
  339. CIPSOv4 Variables:
  340. cipso_cache_enable - BOOLEAN
  341. If set, enable additions to and lookups from the CIPSO label mapping
  342. cache. If unset, additions are ignored and lookups always result in a
  343. miss. However, regardless of the setting the cache is still
  344. invalidated when required when means you can safely toggle this on and
  345. off and the cache will always be "safe".
  346. Default: 1
  347. cipso_cache_bucket_size - INTEGER
  348. The CIPSO label cache consists of a fixed size hash table with each
  349. hash bucket containing a number of cache entries. This variable limits
  350. the number of entries in each hash bucket; the larger the value the
  351. more CIPSO label mappings that can be cached. When the number of
  352. entries in a given hash bucket reaches this limit adding new entries
  353. causes the oldest entry in the bucket to be removed to make room.
  354. Default: 10
  355. cipso_rbm_optfmt - BOOLEAN
  356. Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of
  357. the CIPSO draft specification (see Documentation/netlabel for details).
  358. This means that when set the CIPSO tag will be padded with empty
  359. categories in order to make the packet data 32-bit aligned.
  360. Default: 0
  361. cipso_rbm_structvalid - BOOLEAN
  362. If set, do a very strict check of the CIPSO option when
  363. ip_options_compile() is called. If unset, relax the checks done during
  364. ip_options_compile(). Either way is "safe" as errors are caught else
  365. where in the CIPSO processing code but setting this to 0 (False) should
  366. result in less work (i.e. it should be faster) but could cause problems
  367. with other implementations that require strict checking.
  368. Default: 0
  369. IP Variables:
  370. ip_local_port_range - 2 INTEGERS
  371. Defines the local port range that is used by TCP and UDP to
  372. choose the local port. The first number is the first, the
  373. second the last local port number. Default value depends on
  374. amount of memory available on the system:
  375. > 128Mb 32768-61000
  376. < 128Mb 1024-4999 or even less.
  377. This number defines number of active connections, which this
  378. system can issue simultaneously to systems not supporting
  379. TCP extensions (timestamps). With tcp_tw_recycle enabled
  380. (i.e. by default) range 1024-4999 is enough to issue up to
  381. 2000 connections per second to systems supporting timestamps.
  382. ip_nonlocal_bind - BOOLEAN
  383. If set, allows processes to bind() to non-local IP addresses,
  384. which can be quite useful - but may break some applications.
  385. Default: 0
  386. ip_dynaddr - BOOLEAN
  387. If set non-zero, enables support for dynamic addresses.
  388. If set to a non-zero value larger than 1, a kernel log
  389. message will be printed when dynamic address rewriting
  390. occurs.
  391. Default: 0
  392. icmp_echo_ignore_all - BOOLEAN
  393. If set non-zero, then the kernel will ignore all ICMP ECHO
  394. requests sent to it.
  395. Default: 0
  396. icmp_echo_ignore_broadcasts - BOOLEAN
  397. If set non-zero, then the kernel will ignore all ICMP ECHO and
  398. TIMESTAMP requests sent to it via broadcast/multicast.
  399. Default: 1
  400. icmp_ratelimit - INTEGER
  401. Limit the maximal rates for sending ICMP packets whose type matches
  402. icmp_ratemask (see below) to specific targets.
  403. 0 to disable any limiting, otherwise the maximal rate in jiffies(1)
  404. Default: 100
  405. icmp_ratemask - INTEGER
  406. Mask made of ICMP types for which rates are being limited.
  407. Significant bits: IHGFEDCBA9876543210
  408. Default mask: 0000001100000011000 (6168)
  409. Bit definitions (see include/linux/icmp.h):
  410. 0 Echo Reply
  411. 3 Destination Unreachable *
  412. 4 Source Quench *
  413. 5 Redirect
  414. 8 Echo Request
  415. B Time Exceeded *
  416. C Parameter Problem *
  417. D Timestamp Request
  418. E Timestamp Reply
  419. F Info Request
  420. G Info Reply
  421. H Address Mask Request
  422. I Address Mask Reply
  423. * These are rate limited by default (see default mask above)
  424. icmp_ignore_bogus_error_responses - BOOLEAN
  425. Some routers violate RFC1122 by sending bogus responses to broadcast
  426. frames. Such violations are normally logged via a kernel warning.
  427. If this is set to TRUE, the kernel will not give such warnings, which
  428. will avoid log file clutter.
  429. Default: FALSE
  430. icmp_errors_use_inbound_ifaddr - BOOLEAN
  431. If zero, icmp error messages are sent with the primary address of
  432. the exiting interface.
  433. If non-zero, the message will be sent with the primary address of
  434. the interface that received the packet that caused the icmp error.
  435. This is the behaviour network many administrators will expect from
  436. a router. And it can make debugging complicated network layouts
  437. much easier.
  438. Note that if no primary address exists for the interface selected,
  439. then the primary address of the first non-loopback interface that
  440. has one will be used regardless of this setting.
  441. Default: 0
  442. igmp_max_memberships - INTEGER
  443. Change the maximum number of multicast groups we can subscribe to.
  444. Default: 20
  445. conf/interface/* changes special settings per interface (where "interface" is
  446. the name of your network interface)
  447. conf/all/* is special, changes the settings for all interfaces
  448. log_martians - BOOLEAN
  449. Log packets with impossible addresses to kernel log.
  450. log_martians for the interface will be enabled if at least one of
  451. conf/{all,interface}/log_martians is set to TRUE,
  452. it will be disabled otherwise
  453. accept_redirects - BOOLEAN
  454. Accept ICMP redirect messages.
  455. accept_redirects for the interface will be enabled if:
  456. - both conf/{all,interface}/accept_redirects are TRUE in the case forwarding
  457. for the interface is enabled
  458. or
  459. - at least one of conf/{all,interface}/accept_redirects is TRUE in the case
  460. forwarding for the interface is disabled
  461. accept_redirects for the interface will be disabled otherwise
  462. default TRUE (host)
  463. FALSE (router)
  464. forwarding - BOOLEAN
  465. Enable IP forwarding on this interface.
  466. mc_forwarding - BOOLEAN
  467. Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE
  468. and a multicast routing daemon is required.
  469. conf/all/mc_forwarding must also be set to TRUE to enable multicast routing
  470. for the interface
  471. medium_id - INTEGER
  472. Integer value used to differentiate the devices by the medium they
  473. are attached to. Two devices can have different id values when
  474. the broadcast packets are received only on one of them.
  475. The default value 0 means that the device is the only interface
  476. to its medium, value of -1 means that medium is not known.
  477. Currently, it is used to change the proxy_arp behavior:
  478. the proxy_arp feature is enabled for packets forwarded between
  479. two devices attached to different media.
  480. proxy_arp - BOOLEAN
  481. Do proxy arp.
  482. proxy_arp for the interface will be enabled if at least one of
  483. conf/{all,interface}/proxy_arp is set to TRUE,
  484. it will be disabled otherwise
  485. shared_media - BOOLEAN
  486. Send(router) or accept(host) RFC1620 shared media redirects.
  487. Overrides ip_secure_redirects.
  488. shared_media for the interface will be enabled if at least one of
  489. conf/{all,interface}/shared_media is set to TRUE,
  490. it will be disabled otherwise
  491. default TRUE
  492. secure_redirects - BOOLEAN
  493. Accept ICMP redirect messages only for gateways,
  494. listed in default gateway list.
  495. secure_redirects for the interface will be enabled if at least one of
  496. conf/{all,interface}/secure_redirects is set to TRUE,
  497. it will be disabled otherwise
  498. default TRUE
  499. send_redirects - BOOLEAN
  500. Send redirects, if router.
  501. send_redirects for the interface will be enabled if at least one of
  502. conf/{all,interface}/send_redirects is set to TRUE,
  503. it will be disabled otherwise
  504. Default: TRUE
  505. bootp_relay - BOOLEAN
  506. Accept packets with source address 0.b.c.d destined
  507. not to this host as local ones. It is supposed, that
  508. BOOTP relay daemon will catch and forward such packets.
  509. conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay
  510. for the interface
  511. default FALSE
  512. Not Implemented Yet.
  513. accept_source_route - BOOLEAN
  514. Accept packets with SRR option.
  515. conf/all/accept_source_route must also be set to TRUE to accept packets
  516. with SRR option on the interface
  517. default TRUE (router)
  518. FALSE (host)
  519. rp_filter - BOOLEAN
  520. 1 - do source validation by reversed path, as specified in RFC1812
  521. Recommended option for single homed hosts and stub network
  522. routers. Could cause troubles for complicated (not loop free)
  523. networks running a slow unreliable protocol (sort of RIP),
  524. or using static routes.
  525. 0 - No source validation.
  526. conf/all/rp_filter must also be set to TRUE to do source validation
  527. on the interface
  528. Default value is 0. Note that some distributions enable it
  529. in startup scripts.
  530. arp_filter - BOOLEAN
  531. 1 - Allows you to have multiple network interfaces on the same
  532. subnet, and have the ARPs for each interface be answered
  533. based on whether or not the kernel would route a packet from
  534. the ARP'd IP out that interface (therefore you must use source
  535. based routing for this to work). In other words it allows control
  536. of which cards (usually 1) will respond to an arp request.
  537. 0 - (default) The kernel can respond to arp requests with addresses
  538. from other interfaces. This may seem wrong but it usually makes
  539. sense, because it increases the chance of successful communication.
  540. IP addresses are owned by the complete host on Linux, not by
  541. particular interfaces. Only for more complex setups like load-
  542. balancing, does this behaviour cause problems.
  543. arp_filter for the interface will be enabled if at least one of
  544. conf/{all,interface}/arp_filter is set to TRUE,
  545. it will be disabled otherwise
  546. arp_announce - INTEGER
  547. Define different restriction levels for announcing the local
  548. source IP address from IP packets in ARP requests sent on
  549. interface:
  550. 0 - (default) Use any local address, configured on any interface
  551. 1 - Try to avoid local addresses that are not in the target's
  552. subnet for this interface. This mode is useful when target
  553. hosts reachable via this interface require the source IP
  554. address in ARP requests to be part of their logical network
  555. configured on the receiving interface. When we generate the
  556. request we will check all our subnets that include the
  557. target IP and will preserve the source address if it is from
  558. such subnet. If there is no such subnet we select source
  559. address according to the rules for level 2.
  560. 2 - Always use the best local address for this target.
  561. In this mode we ignore the source address in the IP packet
  562. and try to select local address that we prefer for talks with
  563. the target host. Such local address is selected by looking
  564. for primary IP addresses on all our subnets on the outgoing
  565. interface that include the target IP address. If no suitable
  566. local address is found we select the first local address
  567. we have on the outgoing interface or on all other interfaces,
  568. with the hope we will receive reply for our request and
  569. even sometimes no matter the source IP address we announce.
  570. The max value from conf/{all,interface}/arp_announce is used.
  571. Increasing the restriction level gives more chance for
  572. receiving answer from the resolved target while decreasing
  573. the level announces more valid sender's information.
  574. arp_ignore - INTEGER
  575. Define different modes for sending replies in response to
  576. received ARP requests that resolve local target IP addresses:
  577. 0 - (default): reply for any local target IP address, configured
  578. on any interface
  579. 1 - reply only if the target IP address is local address
  580. configured on the incoming interface
  581. 2 - reply only if the target IP address is local address
  582. configured on the incoming interface and both with the
  583. sender's IP address are part from same subnet on this interface
  584. 3 - do not reply for local addresses configured with scope host,
  585. only resolutions for global and link addresses are replied
  586. 4-7 - reserved
  587. 8 - do not reply for all local addresses
  588. The max value from conf/{all,interface}/arp_ignore is used
  589. when ARP request is received on the {interface}
  590. arp_accept - BOOLEAN
  591. Define behavior when gratuitous arp replies are received:
  592. 0 - drop gratuitous arp frames
  593. 1 - accept gratuitous arp frames
  594. app_solicit - INTEGER
  595. The maximum number of probes to send to the user space ARP daemon
  596. via netlink before dropping back to multicast probes (see
  597. mcast_solicit). Defaults to 0.
  598. disable_policy - BOOLEAN
  599. Disable IPSEC policy (SPD) for this interface
  600. disable_xfrm - BOOLEAN
  601. Disable IPSEC encryption on this interface, whatever the policy
  602. tag - INTEGER
  603. Allows you to write a number, which can be used as required.
  604. Default value is 0.
  605. (1) Jiffie: internal timeunit for the kernel. On the i386 1/100s, on the
  606. Alpha 1/1024s. See the HZ define in /usr/include/asm/param.h for the exact
  607. value on your system.
  608. Alexey Kuznetsov.
  609. kuznet@ms2.inr.ac.ru
  610. Updated by:
  611. Andi Kleen
  612. ak@muc.de
  613. Nicolas Delon
  614. delon.nicolas@wanadoo.fr
  615. /proc/sys/net/ipv6/* Variables:
  616. IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also
  617. apply to IPv6 [XXX?].
  618. bindv6only - BOOLEAN
  619. Default value for IPV6_V6ONLY socket option,
  620. which restricts use of the IPv6 socket to IPv6 communication
  621. only.
  622. TRUE: disable IPv4-mapped address feature
  623. FALSE: enable IPv4-mapped address feature
  624. Default: FALSE (as specified in RFC2553bis)
  625. IPv6 Fragmentation:
  626. ip6frag_high_thresh - INTEGER
  627. Maximum memory used to reassemble IPv6 fragments. When
  628. ip6frag_high_thresh bytes of memory is allocated for this purpose,
  629. the fragment handler will toss packets until ip6frag_low_thresh
  630. is reached.
  631. ip6frag_low_thresh - INTEGER
  632. See ip6frag_high_thresh
  633. ip6frag_time - INTEGER
  634. Time in seconds to keep an IPv6 fragment in memory.
  635. ip6frag_secret_interval - INTEGER
  636. Regeneration interval (in seconds) of the hash secret (or lifetime
  637. for the hash secret) for IPv6 fragments.
  638. Default: 600
  639. conf/default/*:
  640. Change the interface-specific default settings.
  641. conf/all/*:
  642. Change all the interface-specific settings.
  643. [XXX: Other special features than forwarding?]
  644. conf/all/forwarding - BOOLEAN
  645. Enable global IPv6 forwarding between all interfaces.
  646. IPv4 and IPv6 work differently here; e.g. netfilter must be used
  647. to control which interfaces may forward packets and which not.
  648. This also sets all interfaces' Host/Router setting
  649. 'forwarding' to the specified value. See below for details.
  650. This referred to as global forwarding.
  651. proxy_ndp - BOOLEAN
  652. Do proxy ndp.
  653. conf/interface/*:
  654. Change special settings per interface.
  655. The functional behaviour for certain settings is different
  656. depending on whether local forwarding is enabled or not.
  657. accept_ra - BOOLEAN
  658. Accept Router Advertisements; autoconfigure using them.
  659. Functional default: enabled if local forwarding is disabled.
  660. disabled if local forwarding is enabled.
  661. accept_ra_defrtr - BOOLEAN
  662. Learn default router in Router Advertisement.
  663. Functional default: enabled if accept_ra is enabled.
  664. disabled if accept_ra is disabled.
  665. accept_ra_pinfo - BOOLEAN
  666. Learn Prefix Information in Router Advertisement.
  667. Functional default: enabled if accept_ra is enabled.
  668. disabled if accept_ra is disabled.
  669. accept_ra_rt_info_max_plen - INTEGER
  670. Maximum prefix length of Route Information in RA.
  671. Route Information w/ prefix larger than or equal to this
  672. variable shall be ignored.
  673. Functional default: 0 if accept_ra_rtr_pref is enabled.
  674. -1 if accept_ra_rtr_pref is disabled.
  675. accept_ra_rtr_pref - BOOLEAN
  676. Accept Router Preference in RA.
  677. Functional default: enabled if accept_ra is enabled.
  678. disabled if accept_ra is disabled.
  679. accept_redirects - BOOLEAN
  680. Accept Redirects.
  681. Functional default: enabled if local forwarding is disabled.
  682. disabled if local forwarding is enabled.
  683. accept_source_route - INTEGER
  684. Accept source routing (routing extension header).
  685. > 0: Accept routing header.
  686. = 0: Accept only routing header type 2.
  687. < 0: Do not accept routing header.
  688. Default: 0
  689. autoconf - BOOLEAN
  690. Autoconfigure addresses using Prefix Information in Router
  691. Advertisements.
  692. Functional default: enabled if accept_ra_pinfo is enabled.
  693. disabled if accept_ra_pinfo is disabled.
  694. dad_transmits - INTEGER
  695. The amount of Duplicate Address Detection probes to send.
  696. Default: 1
  697. forwarding - BOOLEAN
  698. Configure interface-specific Host/Router behaviour.
  699. Note: It is recommended to have the same setting on all
  700. interfaces; mixed router/host scenarios are rather uncommon.
  701. FALSE:
  702. By default, Host behaviour is assumed. This means:
  703. 1. IsRouter flag is not set in Neighbour Advertisements.
  704. 2. Router Solicitations are being sent when necessary.
  705. 3. If accept_ra is TRUE (default), accept Router
  706. Advertisements (and do autoconfiguration).
  707. 4. If accept_redirects is TRUE (default), accept Redirects.
  708. TRUE:
  709. If local forwarding is enabled, Router behaviour is assumed.
  710. This means exactly the reverse from the above:
  711. 1. IsRouter flag is set in Neighbour Advertisements.
  712. 2. Router Solicitations are not sent.
  713. 3. Router Advertisements are ignored.
  714. 4. Redirects are ignored.
  715. Default: FALSE if global forwarding is disabled (default),
  716. otherwise TRUE.
  717. hop_limit - INTEGER
  718. Default Hop Limit to set.
  719. Default: 64
  720. mtu - INTEGER
  721. Default Maximum Transfer Unit
  722. Default: 1280 (IPv6 required minimum)
  723. router_probe_interval - INTEGER
  724. Minimum interval (in seconds) between Router Probing described
  725. in RFC4191.
  726. Default: 60
  727. router_solicitation_delay - INTEGER
  728. Number of seconds to wait after interface is brought up
  729. before sending Router Solicitations.
  730. Default: 1
  731. router_solicitation_interval - INTEGER
  732. Number of seconds to wait between Router Solicitations.
  733. Default: 4
  734. router_solicitations - INTEGER
  735. Number of Router Solicitations to send until assuming no
  736. routers are present.
  737. Default: 3
  738. use_tempaddr - INTEGER
  739. Preference for Privacy Extensions (RFC3041).
  740. <= 0 : disable Privacy Extensions
  741. == 1 : enable Privacy Extensions, but prefer public
  742. addresses over temporary addresses.
  743. > 1 : enable Privacy Extensions and prefer temporary
  744. addresses over public addresses.
  745. Default: 0 (for most devices)
  746. -1 (for point-to-point devices and loopback devices)
  747. temp_valid_lft - INTEGER
  748. valid lifetime (in seconds) for temporary addresses.
  749. Default: 604800 (7 days)
  750. temp_prefered_lft - INTEGER
  751. Preferred lifetime (in seconds) for temporary addresses.
  752. Default: 86400 (1 day)
  753. max_desync_factor - INTEGER
  754. Maximum value for DESYNC_FACTOR, which is a random value
  755. that ensures that clients don't synchronize with each
  756. other and generate new addresses at exactly the same time.
  757. value is in seconds.
  758. Default: 600
  759. regen_max_retry - INTEGER
  760. Number of attempts before give up attempting to generate
  761. valid temporary addresses.
  762. Default: 5
  763. max_addresses - INTEGER
  764. Number of maximum addresses per interface. 0 disables limitation.
  765. It is recommended not set too large value (or 0) because it would
  766. be too easy way to crash kernel to allow to create too much of
  767. autoconfigured addresses.
  768. Default: 16
  769. icmp/*:
  770. ratelimit - INTEGER
  771. Limit the maximal rates for sending ICMPv6 packets.
  772. 0 to disable any limiting, otherwise the maximal rate in jiffies(1)
  773. Default: 100
  774. IPv6 Update by:
  775. Pekka Savola <pekkas@netcore.fi>
  776. YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org>
  777. /proc/sys/net/bridge/* Variables:
  778. bridge-nf-call-arptables - BOOLEAN
  779. 1 : pass bridged ARP traffic to arptables' FORWARD chain.
  780. 0 : disable this.
  781. Default: 1
  782. bridge-nf-call-iptables - BOOLEAN
  783. 1 : pass bridged IPv4 traffic to iptables' chains.
  784. 0 : disable this.
  785. Default: 1
  786. bridge-nf-call-ip6tables - BOOLEAN
  787. 1 : pass bridged IPv6 traffic to ip6tables' chains.
  788. 0 : disable this.
  789. Default: 1
  790. bridge-nf-filter-vlan-tagged - BOOLEAN
  791. 1 : pass bridged vlan-tagged ARP/IP traffic to arptables/iptables.
  792. 0 : disable this.
  793. Default: 1
  794. UNDOCUMENTED:
  795. dev_weight FIXME
  796. discovery_slots FIXME
  797. discovery_timeout FIXME
  798. fast_poll_increase FIXME
  799. ip6_queue_maxlen FIXME
  800. lap_keepalive_time FIXME
  801. lo_cong FIXME
  802. max_baud_rate FIXME
  803. max_dgram_qlen FIXME
  804. max_noreply_time FIXME
  805. max_tx_data_size FIXME
  806. max_tx_window FIXME
  807. min_tx_turn_time FIXME
  808. mod_cong FIXME
  809. no_cong FIXME
  810. no_cong_thresh FIXME
  811. slot_timeout FIXME
  812. warn_noreply_time FIXME