Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | /proc/sys/net/ipv4/* Variables: |
| 2 | |
| 3 | ip_forward - BOOLEAN |
| 4 | 0 - disabled (default) |
| 5 | not 0 - enabled |
| 6 | |
| 7 | Forward Packets between interfaces. |
| 8 | |
| 9 | This variable is special, its change resets all configuration |
| 10 | parameters to their default state (RFC1122 for hosts, RFC1812 |
| 11 | for routers) |
| 12 | |
| 13 | ip_default_ttl - INTEGER |
| 14 | default 64 |
| 15 | |
| 16 | ip_no_pmtu_disc - BOOLEAN |
| 17 | Disable Path MTU Discovery. |
| 18 | default FALSE |
| 19 | |
| 20 | min_pmtu - INTEGER |
| 21 | default 562 - minimum discovered Path MTU |
| 22 | |
| 23 | mtu_expires - INTEGER |
| 24 | Time, in seconds, that cached PMTU information is kept. |
| 25 | |
| 26 | min_adv_mss - INTEGER |
| 27 | The advertised MSS depends on the first hop route MTU, but will |
| 28 | never be lower than this setting. |
| 29 | |
| 30 | IP Fragmentation: |
| 31 | |
| 32 | ipfrag_high_thresh - INTEGER |
| 33 | Maximum memory used to reassemble IP fragments. When |
| 34 | ipfrag_high_thresh bytes of memory is allocated for this purpose, |
| 35 | the fragment handler will toss packets until ipfrag_low_thresh |
| 36 | is reached. |
| 37 | |
| 38 | ipfrag_low_thresh - INTEGER |
| 39 | See ipfrag_high_thresh |
| 40 | |
| 41 | ipfrag_time - INTEGER |
| 42 | Time in seconds to keep an IP fragment in memory. |
| 43 | |
| 44 | ipfrag_secret_interval - INTEGER |
| 45 | Regeneration interval (in seconds) of the hash secret (or lifetime |
| 46 | for the hash secret) for IP fragments. |
| 47 | Default: 600 |
| 48 | |
Herbert Xu | 89cee8b | 2005-12-13 23:14:27 -0800 | [diff] [blame] | 49 | ipfrag_max_dist - INTEGER |
| 50 | ipfrag_max_dist is a non-negative integer value which defines the |
| 51 | maximum "disorder" which is allowed among fragments which share a |
| 52 | common IP source address. Note that reordering of packets is |
| 53 | not unusual, but if a large number of fragments arrive from a source |
| 54 | IP address while a particular fragment queue remains incomplete, it |
| 55 | probably indicates that one or more fragments belonging to that queue |
| 56 | have been lost. When ipfrag_max_dist is positive, an additional check |
| 57 | is done on fragments before they are added to a reassembly queue - if |
| 58 | ipfrag_max_dist (or more) fragments have arrived from a particular IP |
| 59 | address between additions to any IP fragment queue using that source |
| 60 | address, it's presumed that one or more fragments in the queue are |
| 61 | lost. The existing fragment queue will be dropped, and a new one |
| 62 | started. An ipfrag_max_dist value of zero disables this check. |
| 63 | |
| 64 | Using a very small value, e.g. 1 or 2, for ipfrag_max_dist can |
| 65 | result in unnecessarily dropping fragment queues when normal |
| 66 | reordering of packets occurs, which could lead to poor application |
| 67 | performance. Using a very large value, e.g. 50000, increases the |
| 68 | likelihood of incorrectly reassembling IP fragments that originate |
| 69 | from different IP datagrams, which could result in data corruption. |
| 70 | Default: 64 |
| 71 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 72 | INET peer storage: |
| 73 | |
| 74 | inet_peer_threshold - INTEGER |
| 75 | The approximate size of the storage. Starting from this threshold |
| 76 | entries will be thrown aggressively. This threshold also determines |
| 77 | entries' time-to-live and time intervals between garbage collection |
| 78 | passes. More entries, less time-to-live, less GC interval. |
| 79 | |
| 80 | inet_peer_minttl - INTEGER |
| 81 | Minimum time-to-live of entries. Should be enough to cover fragment |
| 82 | time-to-live on the reassembling side. This minimum time-to-live is |
| 83 | guaranteed if the pool size is less than inet_peer_threshold. |
| 84 | Measured in jiffies(1). |
| 85 | |
| 86 | inet_peer_maxttl - INTEGER |
| 87 | Maximum time-to-live of entries. Unused entries will expire after |
| 88 | this period of time if there is no memory pressure on the pool (i.e. |
| 89 | when the number of entries in the pool is very small). |
| 90 | Measured in jiffies(1). |
| 91 | |
| 92 | inet_peer_gc_mintime - INTEGER |
| 93 | Minimum interval between garbage collection passes. This interval is |
| 94 | in effect under high memory pressure on the pool. |
| 95 | Measured in jiffies(1). |
| 96 | |
| 97 | inet_peer_gc_maxtime - INTEGER |
| 98 | Minimum interval between garbage collection passes. This interval is |
| 99 | in effect under low (or absent) memory pressure on the pool. |
| 100 | Measured in jiffies(1). |
| 101 | |
| 102 | TCP variables: |
| 103 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 104 | somaxconn - INTEGER |
| 105 | Limit of socket listen() backlog, known in userspace as SOMAXCONN. |
| 106 | Defaults to 128. See also tcp_max_syn_backlog for additional tuning |
| 107 | for TCP sockets. |
| 108 | |
Stephen Hemminger | 9772efb | 2005-11-10 17:09:53 -0800 | [diff] [blame] | 109 | tcp_abc - INTEGER |
Stephen Hemminger | b3a8a40 | 2006-09-13 19:51:02 -0700 | [diff] [blame] | 110 | Controls Appropriate Byte Count (ABC) defined in RFC3465. |
| 111 | ABC is a way of increasing congestion window (cwnd) more slowly |
| 112 | in response to partial acknowledgments. |
| 113 | Possible values are: |
| 114 | 0 increase cwnd once per acknowledgment (no ABC) |
| 115 | 1 increase cwnd once per acknowledgment of full sized segment |
| 116 | 2 allow increase cwnd by two if acknowledgment is |
| 117 | of two segments to compensate for delayed acknowledgments. |
| 118 | Default: 0 (off) |
Stephen Hemminger | 9772efb | 2005-11-10 17:09:53 -0800 | [diff] [blame] | 119 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 120 | tcp_abort_on_overflow - BOOLEAN |
| 121 | If listening service is too slow to accept new connections, |
| 122 | reset them. Default state is FALSE. It means that if overflow |
| 123 | occurred due to a burst, connection will recover. Enable this |
| 124 | option _only_ if you are really sure that listening daemon |
| 125 | cannot be tuned to accept connections faster. Enabling this |
| 126 | option can harm clients of your server. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 127 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 128 | tcp_adv_win_scale - INTEGER |
| 129 | Count buffering overhead as bytes/2^tcp_adv_win_scale |
| 130 | (if tcp_adv_win_scale > 0) or bytes-bytes/2^(-tcp_adv_win_scale), |
| 131 | if it is <= 0. |
| 132 | Default: 2 |
| 133 | |
| 134 | tcp_allowed_congestion_control - STRING |
| 135 | Show/set the congestion control choices available to non-privileged |
| 136 | processes. The list is a subset of those listed in |
| 137 | tcp_available_congestion_control. |
| 138 | Default is "reno" and the default setting (tcp_congestion_control). |
| 139 | |
| 140 | tcp_app_win - INTEGER |
| 141 | Reserve max(window/2^tcp_app_win, mss) of window for application |
| 142 | buffer. Value 0 is special, it means that nothing is reserved. |
| 143 | Default: 31 |
| 144 | |
| 145 | tcp_available_congestion_control - STRING |
| 146 | Shows the available congestion control choices that are registered. |
| 147 | More congestion control algorithms may be available as modules, |
| 148 | but not loaded. |
| 149 | |
John Heffner | 71599cd | 2007-02-27 10:03:56 -0800 | [diff] [blame] | 150 | tcp_base_mss - INTEGER |
| 151 | The initial value of search_low to be used by Packetization Layer |
| 152 | Path MTU Discovery (MTU probing). If MTU probing is enabled, |
| 153 | this is the inital MSS used by the connection. |
| 154 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 155 | tcp_congestion_control - STRING |
| 156 | Set the congestion control algorithm to be used for new |
| 157 | connections. The algorithm "reno" is always available, but |
| 158 | additional choices may be available based on kernel configuration. |
| 159 | Default is set as part of kernel configuration. |
| 160 | |
| 161 | tcp_dsack - BOOLEAN |
| 162 | Allows TCP to send "duplicate" SACKs. |
| 163 | |
| 164 | tcp_ecn - BOOLEAN |
| 165 | Enable Explicit Congestion Notification in TCP. |
| 166 | |
| 167 | tcp_fack - BOOLEAN |
| 168 | Enable FACK congestion avoidance and fast retransmission. |
| 169 | The value is not used, if tcp_sack is not enabled. |
| 170 | |
| 171 | tcp_fin_timeout - INTEGER |
| 172 | Time to hold socket in state FIN-WAIT-2, if it was closed |
| 173 | by our side. Peer can be broken and never close its side, |
| 174 | or even died unexpectedly. Default value is 60sec. |
| 175 | Usual value used in 2.2 was 180 seconds, you may restore |
| 176 | it, but remember that if your machine is even underloaded WEB server, |
| 177 | you risk to overflow memory with kilotons of dead sockets, |
| 178 | FIN-WAIT-2 sockets are less dangerous than FIN-WAIT-1, |
| 179 | because they eat maximum 1.5K of memory, but they tend |
| 180 | to live longer. Cf. tcp_max_orphans. |
| 181 | |
Ilpo Järvinen | 8980806 | 2007-02-27 10:10:55 -0800 | [diff] [blame] | 182 | tcp_frto - INTEGER |
Ilpo Järvinen | cd99889 | 2007-09-20 11:35:26 -0700 | [diff] [blame] | 183 | Enables Forward RTO-Recovery (F-RTO) defined in RFC4138. |
| 184 | F-RTO is an enhanced recovery algorithm for TCP retransmission |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 185 | timeouts. It is particularly beneficial in wireless environments |
| 186 | where packet loss is typically due to random radio interference |
Ryousei Takano | 564262c1 | 2007-10-25 23:03:52 -0700 | [diff] [blame] | 187 | rather than intermediate router congestion. F-RTO is sender-side |
Ilpo Järvinen | cd99889 | 2007-09-20 11:35:26 -0700 | [diff] [blame] | 188 | only modification. Therefore it does not require any support from |
| 189 | the peer, but in a typical case, however, where wireless link is |
| 190 | the local access link and most of the data flows downlink, the |
Ryousei Takano | 564262c1 | 2007-10-25 23:03:52 -0700 | [diff] [blame] | 191 | faraway servers should have F-RTO enabled to take advantage of it. |
Ilpo Järvinen | cd99889 | 2007-09-20 11:35:26 -0700 | [diff] [blame] | 192 | If set to 1, basic version is enabled. 2 enables SACK enhanced |
| 193 | F-RTO if flow uses SACK. The basic version can be used also when |
Ryousei Takano | 564262c1 | 2007-10-25 23:03:52 -0700 | [diff] [blame] | 194 | SACK is in use though scenario(s) with it exists where F-RTO |
Ilpo Järvinen | cd99889 | 2007-09-20 11:35:26 -0700 | [diff] [blame] | 195 | interacts badly with the packet counting of the SACK enabled TCP |
| 196 | flow. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 197 | |
Ilpo Järvinen | 8980806 | 2007-02-27 10:10:55 -0800 | [diff] [blame] | 198 | tcp_frto_response - INTEGER |
| 199 | When F-RTO has detected that a TCP retransmission timeout was |
| 200 | spurious (i.e, the timeout would have been avoided had TCP set a |
| 201 | longer retransmission timeout), TCP has several options what to do |
| 202 | next. Possible values are: |
| 203 | 0 Rate halving based; a smooth and conservative response, |
| 204 | results in halved cwnd and ssthresh after one RTT |
| 205 | 1 Very conservative response; not recommended because even |
| 206 | though being valid, it interacts poorly with the rest of |
| 207 | Linux TCP, halves cwnd and ssthresh immediately |
| 208 | 2 Aggressive response; undoes congestion control measures |
| 209 | that are now known to be unnecessary (ignoring the |
| 210 | possibility of a lost retransmission that would require |
| 211 | TCP to be more cautious), cwnd and ssthresh are restored |
| 212 | to the values prior timeout |
| 213 | Default: 0 (rate halving based) |
| 214 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 215 | tcp_keepalive_time - INTEGER |
| 216 | How often TCP sends out keepalive messages when keepalive is enabled. |
| 217 | Default: 2hours. |
| 218 | |
| 219 | tcp_keepalive_probes - INTEGER |
| 220 | How many keepalive probes TCP sends out, until it decides that the |
| 221 | connection is broken. Default value: 9. |
| 222 | |
| 223 | tcp_keepalive_intvl - INTEGER |
| 224 | How frequently the probes are send out. Multiplied by |
| 225 | tcp_keepalive_probes it is time to kill not responding connection, |
| 226 | after probes started. Default value: 75sec i.e. connection |
| 227 | will be aborted after ~11 minutes of retries. |
| 228 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 229 | tcp_low_latency - BOOLEAN |
| 230 | If set, the TCP stack makes decisions that prefer lower |
| 231 | latency as opposed to higher throughput. By default, this |
| 232 | option is not set meaning that higher throughput is preferred. |
| 233 | An example of an application where this default should be |
| 234 | changed would be a Beowulf compute cluster. |
| 235 | Default: 0 |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 236 | |
| 237 | tcp_max_orphans - INTEGER |
| 238 | Maximal number of TCP sockets not attached to any user file handle, |
| 239 | held by system. If this number is exceeded orphaned connections are |
| 240 | reset immediately and warning is printed. This limit exists |
| 241 | only to prevent simple DoS attacks, you _must_ not rely on this |
| 242 | or lower the limit artificially, but rather increase it |
| 243 | (probably, after increasing installed memory), |
| 244 | if network conditions require more than default value, |
| 245 | and tune network services to linger and kill such states |
| 246 | more aggressively. Let me to remind again: each orphan eats |
| 247 | up to ~64K of unswappable memory. |
| 248 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 249 | tcp_max_syn_backlog - INTEGER |
| 250 | Maximal number of remembered connection requests, which are |
| 251 | still did not receive an acknowledgment from connecting client. |
| 252 | Default value is 1024 for systems with more than 128Mb of memory, |
| 253 | and 128 for low memory machines. If server suffers of overload, |
| 254 | try to increase this number. |
| 255 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 256 | tcp_max_tw_buckets - INTEGER |
| 257 | Maximal number of timewait sockets held by system simultaneously. |
| 258 | If this number is exceeded time-wait socket is immediately destroyed |
| 259 | and warning is printed. This limit exists only to prevent |
| 260 | simple DoS attacks, you _must_ not lower the limit artificially, |
| 261 | but rather increase it (probably, after increasing installed memory), |
| 262 | if network conditions require more than default value. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 263 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 264 | tcp_mem - vector of 3 INTEGERs: min, pressure, max |
| 265 | min: below this number of pages TCP is not bothered about its |
| 266 | memory appetite. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 267 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 268 | pressure: when amount of memory allocated by TCP exceeds this number |
| 269 | of pages, TCP moderates its memory consumption and enters memory |
| 270 | pressure mode, which is exited when memory consumption falls |
| 271 | under "min". |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 272 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 273 | max: number of pages allowed for queueing by all TCP sockets. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 274 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 275 | Defaults are calculated at boot time from amount of available |
| 276 | memory. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 277 | |
John Heffner | 71599cd | 2007-02-27 10:03:56 -0800 | [diff] [blame] | 278 | tcp_moderate_rcvbuf - BOOLEAN |
| 279 | If set, TCP performs receive buffer autotuning, attempting to |
| 280 | automatically size the buffer (no greater than tcp_rmem[2]) to |
| 281 | match the size required by the path for full throughput. Enabled by |
| 282 | default. |
| 283 | |
| 284 | tcp_mtu_probing - INTEGER |
| 285 | Controls TCP Packetization-Layer Path MTU Discovery. Takes three |
| 286 | values: |
| 287 | 0 - Disabled |
| 288 | 1 - Disabled by default, enabled when an ICMP black hole detected |
| 289 | 2 - Always enabled, use initial MSS of tcp_base_mss. |
| 290 | |
| 291 | tcp_no_metrics_save - BOOLEAN |
| 292 | By default, TCP saves various connection metrics in the route cache |
| 293 | when the connection closes, so that connections established in the |
| 294 | near future can use these to set initial conditions. Usually, this |
| 295 | increases overall performance, but may sometimes cause performance |
Simon Arlott | 0f035b8 | 2007-10-20 01:30:25 +0200 | [diff] [blame] | 296 | degradation. If set, TCP will not cache metrics on closing |
John Heffner | 71599cd | 2007-02-27 10:03:56 -0800 | [diff] [blame] | 297 | connections. |
| 298 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 299 | tcp_orphan_retries - INTEGER |
| 300 | How may times to retry before killing TCP connection, closed |
| 301 | by our side. Default value 7 corresponds to ~50sec-16min |
| 302 | depending on RTO. If you machine is loaded WEB server, |
| 303 | you should think about lowering this value, such sockets |
| 304 | may consume significant resources. Cf. tcp_max_orphans. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 305 | |
| 306 | tcp_reordering - INTEGER |
| 307 | Maximal reordering of packets in a TCP stream. |
| 308 | Default: 3 |
| 309 | |
| 310 | tcp_retrans_collapse - BOOLEAN |
| 311 | Bug-to-bug compatibility with some broken printers. |
| 312 | On retransmit try to send bigger packets to work around bugs in |
| 313 | certain TCP stacks. |
| 314 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 315 | tcp_retries1 - INTEGER |
| 316 | How many times to retry before deciding that something is wrong |
| 317 | and it is necessary to report this suspicion to network layer. |
| 318 | Minimal RFC value is 3, it is default, which corresponds |
| 319 | to ~3sec-8min depending on RTO. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 320 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 321 | tcp_retries2 - INTEGER |
| 322 | How may times to retry before killing alive TCP connection. |
| 323 | RFC1122 says that the limit should be longer than 100 sec. |
| 324 | It is too small number. Default value 15 corresponds to ~13-30min |
| 325 | depending on RTO. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 326 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 327 | tcp_rfc1337 - BOOLEAN |
| 328 | If set, the TCP stack behaves conforming to RFC1337. If unset, |
| 329 | we are not conforming to RFC, but prevent TCP TIME_WAIT |
| 330 | assassination. |
| 331 | Default: 0 |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 332 | |
| 333 | tcp_rmem - vector of 3 INTEGERs: min, default, max |
| 334 | min: Minimal size of receive buffer used by TCP sockets. |
| 335 | It is guaranteed to each TCP socket, even under moderate memory |
| 336 | pressure. |
| 337 | Default: 8K |
| 338 | |
| 339 | default: default size of receive buffer used by TCP sockets. |
| 340 | This value overrides net.core.rmem_default used by other protocols. |
| 341 | Default: 87380 bytes. This value results in window of 65535 with |
| 342 | default setting of tcp_adv_win_scale and tcp_app_win:0 and a bit |
| 343 | less for default tcp_app_win. See below about these variables. |
| 344 | |
| 345 | max: maximal size of receive buffer allowed for automatically |
| 346 | selected receiver buffers for TCP socket. This value does not override |
| 347 | net.core.rmem_max, "static" selection via SO_RCVBUF does not use this. |
| 348 | Default: 87380*2 bytes. |
| 349 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 350 | tcp_sack - BOOLEAN |
| 351 | Enable select acknowledgments (SACKS). |
Rick Jones | 15d99e0 | 2006-03-20 22:40:29 -0800 | [diff] [blame] | 352 | |
David S. Miller | 35089bb | 2006-06-13 22:33:04 -0700 | [diff] [blame] | 353 | tcp_slow_start_after_idle - BOOLEAN |
| 354 | If set, provide RFC2861 behavior and time out the congestion |
| 355 | window after an idle period. An idle period is defined at |
| 356 | the current RTO. If unset, the congestion window will not |
| 357 | be timed out after an idle period. |
| 358 | Default: 1 |
| 359 | |
Stephen Hemminger | ef56e62 | 2006-11-09 16:37:26 -0800 | [diff] [blame] | 360 | tcp_stdurg - BOOLEAN |
| 361 | Use the Host requirements interpretation of the TCP urg pointer field. |
| 362 | Most hosts use the older BSD interpretation, so if you turn this on |
| 363 | Linux might not communicate correctly with them. |
| 364 | Default: FALSE |
| 365 | |
| 366 | tcp_synack_retries - INTEGER |
| 367 | Number of times SYNACKs for a passive TCP connection attempt will |
| 368 | be retransmitted. Should not be higher than 255. Default value |
| 369 | is 5, which corresponds to ~180seconds. |
| 370 | |
| 371 | tcp_syncookies - BOOLEAN |
| 372 | Only valid when the kernel was compiled with CONFIG_SYNCOOKIES |
| 373 | Send out syncookies when the syn backlog queue of a socket |
| 374 | overflows. This is to prevent against the common 'syn flood attack' |
| 375 | Default: FALSE |
| 376 | |
| 377 | Note, that syncookies is fallback facility. |
| 378 | It MUST NOT be used to help highly loaded servers to stand |
| 379 | against legal connection rate. If you see synflood warnings |
| 380 | in your logs, but investigation shows that they occur |
| 381 | because of overload with legal connections, you should tune |
| 382 | another parameters until this warning disappear. |
| 383 | See: tcp_max_syn_backlog, tcp_synack_retries, tcp_abort_on_overflow. |
| 384 | |
| 385 | syncookies seriously violate TCP protocol, do not allow |
| 386 | to use TCP extensions, can result in serious degradation |
| 387 | of some services (f.e. SMTP relaying), visible not by you, |
| 388 | but your clients and relays, contacting you. While you see |
| 389 | synflood warnings in logs not being really flooded, your server |
| 390 | is seriously misconfigured. |
| 391 | |
| 392 | tcp_syn_retries - INTEGER |
| 393 | Number of times initial SYNs for an active TCP connection attempt |
| 394 | will be retransmitted. Should not be higher than 255. Default value |
| 395 | is 5, which corresponds to ~180seconds. |
| 396 | |
| 397 | tcp_timestamps - BOOLEAN |
| 398 | Enable timestamps as defined in RFC1323. |
| 399 | |
| 400 | tcp_tso_win_divisor - INTEGER |
| 401 | This allows control over what percentage of the congestion window |
| 402 | can be consumed by a single TSO frame. |
| 403 | The setting of this parameter is a choice between burstiness and |
| 404 | building larger TSO frames. |
| 405 | Default: 3 |
| 406 | |
| 407 | tcp_tw_recycle - BOOLEAN |
| 408 | Enable fast recycling TIME-WAIT sockets. Default value is 0. |
| 409 | It should not be changed without advice/request of technical |
| 410 | experts. |
| 411 | |
| 412 | tcp_tw_reuse - BOOLEAN |
| 413 | Allow to reuse TIME-WAIT sockets for new connections when it is |
| 414 | safe from protocol viewpoint. Default value is 0. |
| 415 | It should not be changed without advice/request of technical |
| 416 | experts. |
| 417 | |
| 418 | tcp_window_scaling - BOOLEAN |
| 419 | Enable window scaling as defined in RFC1323. |
| 420 | |
| 421 | tcp_wmem - vector of 3 INTEGERs: min, default, max |
| 422 | min: Amount of memory reserved for send buffers for TCP socket. |
| 423 | Each TCP socket has rights to use it due to fact of its birth. |
| 424 | Default: 4K |
| 425 | |
| 426 | default: Amount of memory allowed for send buffers for TCP socket |
| 427 | by default. This value overrides net.core.wmem_default used |
| 428 | by other protocols, it is usually lower than net.core.wmem_default. |
| 429 | Default: 16K |
| 430 | |
| 431 | max: Maximal amount of memory allowed for automatically selected |
| 432 | send buffers for TCP socket. This value does not override |
| 433 | net.core.wmem_max, "static" selection via SO_SNDBUF does not use this. |
| 434 | Default: 128K |
| 435 | |
| 436 | tcp_workaround_signed_windows - BOOLEAN |
| 437 | If set, assume no receipt of a window scaling option means the |
| 438 | remote TCP is broken and treats the window as a signed quantity. |
| 439 | If unset, assume the remote TCP is not broken even if we do |
| 440 | not receive a window scaling option from them. |
| 441 | Default: 0 |
| 442 | |
Chris Leech | 72d0b7a | 2007-03-08 09:57:35 -0800 | [diff] [blame] | 443 | tcp_dma_copybreak - INTEGER |
| 444 | Lower limit, in bytes, of the size of socket reads that will be |
| 445 | offloaded to a DMA copy engine, if one is present in the system |
| 446 | and CONFIG_NET_DMA is enabled. |
| 447 | Default: 4096 |
| 448 | |
Hideo Aoki | 95766ff | 2007-12-31 00:29:24 -0800 | [diff] [blame] | 449 | UDP variables: |
| 450 | |
| 451 | udp_mem - vector of 3 INTEGERs: min, pressure, max |
| 452 | Number of pages allowed for queueing by all UDP sockets. |
| 453 | |
| 454 | min: Below this number of pages UDP is not bothered about its |
| 455 | memory appetite. When amount of memory allocated by UDP exceeds |
| 456 | this number, UDP starts to moderate memory usage. |
| 457 | |
| 458 | pressure: This value was introduced to follow format of tcp_mem. |
| 459 | |
| 460 | max: Number of pages allowed for queueing by all UDP sockets. |
| 461 | |
| 462 | Default is calculated at boot time from amount of available memory. |
| 463 | |
| 464 | udp_rmem_min - INTEGER |
| 465 | Minimal size of receive buffer used by UDP sockets in moderation. |
| 466 | Each UDP socket is able to use the size for receiving data, even if |
| 467 | total pages of UDP sockets exceed udp_mem pressure. The unit is byte. |
| 468 | Default: 4096 |
| 469 | |
| 470 | udp_wmem_min - INTEGER |
| 471 | Minimal size of send buffer used by UDP sockets in moderation. |
| 472 | Each UDP socket is able to use the size for sending data, even if |
| 473 | total pages of UDP sockets exceed udp_mem pressure. The unit is byte. |
| 474 | Default: 4096 |
| 475 | |
Paul Moore | 8802f61 | 2006-08-03 16:45:49 -0700 | [diff] [blame] | 476 | CIPSOv4 Variables: |
| 477 | |
| 478 | cipso_cache_enable - BOOLEAN |
| 479 | If set, enable additions to and lookups from the CIPSO label mapping |
| 480 | cache. If unset, additions are ignored and lookups always result in a |
| 481 | miss. However, regardless of the setting the cache is still |
| 482 | invalidated when required when means you can safely toggle this on and |
| 483 | off and the cache will always be "safe". |
| 484 | Default: 1 |
| 485 | |
| 486 | cipso_cache_bucket_size - INTEGER |
| 487 | The CIPSO label cache consists of a fixed size hash table with each |
| 488 | hash bucket containing a number of cache entries. This variable limits |
| 489 | the number of entries in each hash bucket; the larger the value the |
| 490 | more CIPSO label mappings that can be cached. When the number of |
| 491 | entries in a given hash bucket reaches this limit adding new entries |
| 492 | causes the oldest entry in the bucket to be removed to make room. |
| 493 | Default: 10 |
| 494 | |
| 495 | cipso_rbm_optfmt - BOOLEAN |
| 496 | Enable the "Optimized Tag 1 Format" as defined in section 3.4.2.6 of |
| 497 | the CIPSO draft specification (see Documentation/netlabel for details). |
| 498 | This means that when set the CIPSO tag will be padded with empty |
| 499 | categories in order to make the packet data 32-bit aligned. |
| 500 | Default: 0 |
| 501 | |
| 502 | cipso_rbm_structvalid - BOOLEAN |
| 503 | If set, do a very strict check of the CIPSO option when |
| 504 | ip_options_compile() is called. If unset, relax the checks done during |
| 505 | ip_options_compile(). Either way is "safe" as errors are caught else |
| 506 | where in the CIPSO processing code but setting this to 0 (False) should |
| 507 | result in less work (i.e. it should be faster) but could cause problems |
| 508 | with other implementations that require strict checking. |
| 509 | Default: 0 |
| 510 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 511 | IP Variables: |
| 512 | |
| 513 | ip_local_port_range - 2 INTEGERS |
| 514 | Defines the local port range that is used by TCP and UDP to |
| 515 | choose the local port. The first number is the first, the |
| 516 | second the last local port number. Default value depends on |
| 517 | amount of memory available on the system: |
| 518 | > 128Mb 32768-61000 |
| 519 | < 128Mb 1024-4999 or even less. |
| 520 | This number defines number of active connections, which this |
| 521 | system can issue simultaneously to systems not supporting |
| 522 | TCP extensions (timestamps). With tcp_tw_recycle enabled |
| 523 | (i.e. by default) range 1024-4999 is enough to issue up to |
| 524 | 2000 connections per second to systems supporting timestamps. |
| 525 | |
| 526 | ip_nonlocal_bind - BOOLEAN |
| 527 | If set, allows processes to bind() to non-local IP addresses, |
| 528 | which can be quite useful - but may break some applications. |
| 529 | Default: 0 |
| 530 | |
| 531 | ip_dynaddr - BOOLEAN |
| 532 | If set non-zero, enables support for dynamic addresses. |
| 533 | If set to a non-zero value larger than 1, a kernel log |
| 534 | message will be printed when dynamic address rewriting |
| 535 | occurs. |
| 536 | Default: 0 |
| 537 | |
| 538 | icmp_echo_ignore_all - BOOLEAN |
David S. Miller | 7ce31246 | 2005-10-03 16:07:30 -0700 | [diff] [blame] | 539 | If set non-zero, then the kernel will ignore all ICMP ECHO |
| 540 | requests sent to it. |
| 541 | Default: 0 |
| 542 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 543 | icmp_echo_ignore_broadcasts - BOOLEAN |
David S. Miller | 7ce31246 | 2005-10-03 16:07:30 -0700 | [diff] [blame] | 544 | If set non-zero, then the kernel will ignore all ICMP ECHO and |
| 545 | TIMESTAMP requests sent to it via broadcast/multicast. |
| 546 | Default: 1 |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 547 | |
| 548 | icmp_ratelimit - INTEGER |
| 549 | Limit the maximal rates for sending ICMP packets whose type matches |
| 550 | icmp_ratemask (see below) to specific targets. |
Stephen Hemminger | 6dbf4bc | 2008-07-01 19:29:07 -0700 | [diff] [blame] | 551 | 0 to disable any limiting, |
| 552 | otherwise the minimal space between responses in milliseconds. |
| 553 | Default: 1000 |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 554 | |
| 555 | icmp_ratemask - INTEGER |
| 556 | Mask made of ICMP types for which rates are being limited. |
| 557 | Significant bits: IHGFEDCBA9876543210 |
| 558 | Default mask: 0000001100000011000 (6168) |
| 559 | |
| 560 | Bit definitions (see include/linux/icmp.h): |
| 561 | 0 Echo Reply |
| 562 | 3 Destination Unreachable * |
| 563 | 4 Source Quench * |
| 564 | 5 Redirect |
| 565 | 8 Echo Request |
| 566 | B Time Exceeded * |
| 567 | C Parameter Problem * |
| 568 | D Timestamp Request |
| 569 | E Timestamp Reply |
| 570 | F Info Request |
| 571 | G Info Reply |
| 572 | H Address Mask Request |
| 573 | I Address Mask Reply |
| 574 | |
| 575 | * These are rate limited by default (see default mask above) |
| 576 | |
| 577 | icmp_ignore_bogus_error_responses - BOOLEAN |
| 578 | Some routers violate RFC1122 by sending bogus responses to broadcast |
| 579 | frames. Such violations are normally logged via a kernel warning. |
| 580 | If this is set to TRUE, the kernel will not give such warnings, which |
| 581 | will avoid log file clutter. |
| 582 | Default: FALSE |
| 583 | |
Horms | 95f7daf | 2006-02-02 17:02:25 -0800 | [diff] [blame] | 584 | icmp_errors_use_inbound_ifaddr - BOOLEAN |
| 585 | |
| 586 | If zero, icmp error messages are sent with the primary address of |
| 587 | the exiting interface. |
| 588 | |
| 589 | If non-zero, the message will be sent with the primary address of |
| 590 | the interface that received the packet that caused the icmp error. |
| 591 | This is the behaviour network many administrators will expect from |
| 592 | a router. And it can make debugging complicated network layouts |
| 593 | much easier. |
| 594 | |
| 595 | Note that if no primary address exists for the interface selected, |
| 596 | then the primary address of the first non-loopback interface that |
Matt LaPlante | d6bc8ac | 2006-10-03 22:54:15 +0200 | [diff] [blame] | 597 | has one will be used regardless of this setting. |
Horms | 95f7daf | 2006-02-02 17:02:25 -0800 | [diff] [blame] | 598 | |
| 599 | Default: 0 |
| 600 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 601 | igmp_max_memberships - INTEGER |
| 602 | Change the maximum number of multicast groups we can subscribe to. |
| 603 | Default: 20 |
| 604 | |
| 605 | conf/interface/* changes special settings per interface (where "interface" is |
| 606 | the name of your network interface) |
| 607 | conf/all/* is special, changes the settings for all interfaces |
| 608 | |
| 609 | |
| 610 | log_martians - BOOLEAN |
| 611 | Log packets with impossible addresses to kernel log. |
| 612 | log_martians for the interface will be enabled if at least one of |
| 613 | conf/{all,interface}/log_martians is set to TRUE, |
| 614 | it will be disabled otherwise |
| 615 | |
| 616 | accept_redirects - BOOLEAN |
| 617 | Accept ICMP redirect messages. |
| 618 | accept_redirects for the interface will be enabled if: |
| 619 | - both conf/{all,interface}/accept_redirects are TRUE in the case forwarding |
| 620 | for the interface is enabled |
| 621 | or |
| 622 | - at least one of conf/{all,interface}/accept_redirects is TRUE in the case |
| 623 | forwarding for the interface is disabled |
| 624 | accept_redirects for the interface will be disabled otherwise |
| 625 | default TRUE (host) |
| 626 | FALSE (router) |
| 627 | |
| 628 | forwarding - BOOLEAN |
| 629 | Enable IP forwarding on this interface. |
| 630 | |
| 631 | mc_forwarding - BOOLEAN |
| 632 | Do multicast routing. The kernel needs to be compiled with CONFIG_MROUTE |
| 633 | and a multicast routing daemon is required. |
| 634 | conf/all/mc_forwarding must also be set to TRUE to enable multicast routing |
| 635 | for the interface |
| 636 | |
| 637 | medium_id - INTEGER |
| 638 | Integer value used to differentiate the devices by the medium they |
| 639 | are attached to. Two devices can have different id values when |
| 640 | the broadcast packets are received only on one of them. |
| 641 | The default value 0 means that the device is the only interface |
| 642 | to its medium, value of -1 means that medium is not known. |
| 643 | |
| 644 | Currently, it is used to change the proxy_arp behavior: |
| 645 | the proxy_arp feature is enabled for packets forwarded between |
| 646 | two devices attached to different media. |
| 647 | |
| 648 | proxy_arp - BOOLEAN |
| 649 | Do proxy arp. |
| 650 | proxy_arp for the interface will be enabled if at least one of |
| 651 | conf/{all,interface}/proxy_arp is set to TRUE, |
| 652 | it will be disabled otherwise |
| 653 | |
| 654 | shared_media - BOOLEAN |
| 655 | Send(router) or accept(host) RFC1620 shared media redirects. |
| 656 | Overrides ip_secure_redirects. |
| 657 | shared_media for the interface will be enabled if at least one of |
| 658 | conf/{all,interface}/shared_media is set to TRUE, |
| 659 | it will be disabled otherwise |
| 660 | default TRUE |
| 661 | |
| 662 | secure_redirects - BOOLEAN |
| 663 | Accept ICMP redirect messages only for gateways, |
| 664 | listed in default gateway list. |
| 665 | secure_redirects for the interface will be enabled if at least one of |
| 666 | conf/{all,interface}/secure_redirects is set to TRUE, |
| 667 | it will be disabled otherwise |
| 668 | default TRUE |
| 669 | |
| 670 | send_redirects - BOOLEAN |
| 671 | Send redirects, if router. |
| 672 | send_redirects for the interface will be enabled if at least one of |
| 673 | conf/{all,interface}/send_redirects is set to TRUE, |
| 674 | it will be disabled otherwise |
| 675 | Default: TRUE |
| 676 | |
| 677 | bootp_relay - BOOLEAN |
| 678 | Accept packets with source address 0.b.c.d destined |
| 679 | not to this host as local ones. It is supposed, that |
| 680 | BOOTP relay daemon will catch and forward such packets. |
| 681 | conf/all/bootp_relay must also be set to TRUE to enable BOOTP relay |
| 682 | for the interface |
| 683 | default FALSE |
| 684 | Not Implemented Yet. |
| 685 | |
| 686 | accept_source_route - BOOLEAN |
| 687 | Accept packets with SRR option. |
| 688 | conf/all/accept_source_route must also be set to TRUE to accept packets |
| 689 | with SRR option on the interface |
| 690 | default TRUE (router) |
| 691 | FALSE (host) |
| 692 | |
| 693 | rp_filter - BOOLEAN |
| 694 | 1 - do source validation by reversed path, as specified in RFC1812 |
| 695 | Recommended option for single homed hosts and stub network |
| 696 | routers. Could cause troubles for complicated (not loop free) |
| 697 | networks running a slow unreliable protocol (sort of RIP), |
| 698 | or using static routes. |
| 699 | |
| 700 | 0 - No source validation. |
| 701 | |
| 702 | conf/all/rp_filter must also be set to TRUE to do source validation |
| 703 | on the interface |
| 704 | |
| 705 | Default value is 0. Note that some distributions enable it |
| 706 | in startup scripts. |
| 707 | |
| 708 | arp_filter - BOOLEAN |
| 709 | 1 - Allows you to have multiple network interfaces on the same |
| 710 | subnet, and have the ARPs for each interface be answered |
| 711 | based on whether or not the kernel would route a packet from |
| 712 | the ARP'd IP out that interface (therefore you must use source |
| 713 | based routing for this to work). In other words it allows control |
| 714 | of which cards (usually 1) will respond to an arp request. |
| 715 | |
| 716 | 0 - (default) The kernel can respond to arp requests with addresses |
| 717 | from other interfaces. This may seem wrong but it usually makes |
| 718 | sense, because it increases the chance of successful communication. |
| 719 | IP addresses are owned by the complete host on Linux, not by |
| 720 | particular interfaces. Only for more complex setups like load- |
| 721 | balancing, does this behaviour cause problems. |
| 722 | |
| 723 | arp_filter for the interface will be enabled if at least one of |
| 724 | conf/{all,interface}/arp_filter is set to TRUE, |
| 725 | it will be disabled otherwise |
| 726 | |
| 727 | arp_announce - INTEGER |
| 728 | Define different restriction levels for announcing the local |
| 729 | source IP address from IP packets in ARP requests sent on |
| 730 | interface: |
| 731 | 0 - (default) Use any local address, configured on any interface |
| 732 | 1 - Try to avoid local addresses that are not in the target's |
| 733 | subnet for this interface. This mode is useful when target |
| 734 | hosts reachable via this interface require the source IP |
| 735 | address in ARP requests to be part of their logical network |
| 736 | configured on the receiving interface. When we generate the |
| 737 | request we will check all our subnets that include the |
| 738 | target IP and will preserve the source address if it is from |
| 739 | such subnet. If there is no such subnet we select source |
| 740 | address according to the rules for level 2. |
| 741 | 2 - Always use the best local address for this target. |
| 742 | In this mode we ignore the source address in the IP packet |
| 743 | and try to select local address that we prefer for talks with |
| 744 | the target host. Such local address is selected by looking |
| 745 | for primary IP addresses on all our subnets on the outgoing |
| 746 | interface that include the target IP address. If no suitable |
| 747 | local address is found we select the first local address |
| 748 | we have on the outgoing interface or on all other interfaces, |
| 749 | with the hope we will receive reply for our request and |
| 750 | even sometimes no matter the source IP address we announce. |
| 751 | |
| 752 | The max value from conf/{all,interface}/arp_announce is used. |
| 753 | |
| 754 | Increasing the restriction level gives more chance for |
| 755 | receiving answer from the resolved target while decreasing |
| 756 | the level announces more valid sender's information. |
| 757 | |
| 758 | arp_ignore - INTEGER |
| 759 | Define different modes for sending replies in response to |
| 760 | received ARP requests that resolve local target IP addresses: |
| 761 | 0 - (default): reply for any local target IP address, configured |
| 762 | on any interface |
| 763 | 1 - reply only if the target IP address is local address |
| 764 | configured on the incoming interface |
| 765 | 2 - reply only if the target IP address is local address |
| 766 | configured on the incoming interface and both with the |
| 767 | sender's IP address are part from same subnet on this interface |
| 768 | 3 - do not reply for local addresses configured with scope host, |
| 769 | only resolutions for global and link addresses are replied |
| 770 | 4-7 - reserved |
| 771 | 8 - do not reply for all local addresses |
| 772 | |
| 773 | The max value from conf/{all,interface}/arp_ignore is used |
| 774 | when ARP request is received on the {interface} |
| 775 | |
Neil Horman | c1b1bce | 2006-03-20 22:40:03 -0800 | [diff] [blame] | 776 | arp_accept - BOOLEAN |
| 777 | Define behavior when gratuitous arp replies are received: |
| 778 | 0 - drop gratuitous arp frames |
| 779 | 1 - accept gratuitous arp frames |
| 780 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 781 | app_solicit - INTEGER |
| 782 | The maximum number of probes to send to the user space ARP daemon |
| 783 | via netlink before dropping back to multicast probes (see |
| 784 | mcast_solicit). Defaults to 0. |
| 785 | |
| 786 | disable_policy - BOOLEAN |
| 787 | Disable IPSEC policy (SPD) for this interface |
| 788 | |
| 789 | disable_xfrm - BOOLEAN |
| 790 | Disable IPSEC encryption on this interface, whatever the policy |
| 791 | |
| 792 | |
| 793 | |
| 794 | tag - INTEGER |
| 795 | Allows you to write a number, which can be used as required. |
| 796 | Default value is 0. |
| 797 | |
| 798 | (1) Jiffie: internal timeunit for the kernel. On the i386 1/100s, on the |
| 799 | Alpha 1/1024s. See the HZ define in /usr/include/asm/param.h for the exact |
| 800 | value on your system. |
| 801 | |
| 802 | Alexey Kuznetsov. |
| 803 | kuznet@ms2.inr.ac.ru |
| 804 | |
| 805 | Updated by: |
| 806 | Andi Kleen |
| 807 | ak@muc.de |
| 808 | Nicolas Delon |
| 809 | delon.nicolas@wanadoo.fr |
| 810 | |
| 811 | |
| 812 | |
| 813 | |
| 814 | /proc/sys/net/ipv6/* Variables: |
| 815 | |
| 816 | IPv6 has no global variables such as tcp_*. tcp_* settings under ipv4/ also |
| 817 | apply to IPv6 [XXX?]. |
| 818 | |
| 819 | bindv6only - BOOLEAN |
| 820 | Default value for IPV6_V6ONLY socket option, |
| 821 | which restricts use of the IPv6 socket to IPv6 communication |
| 822 | only. |
| 823 | TRUE: disable IPv4-mapped address feature |
| 824 | FALSE: enable IPv4-mapped address feature |
| 825 | |
| 826 | Default: FALSE (as specified in RFC2553bis) |
| 827 | |
| 828 | IPv6 Fragmentation: |
| 829 | |
| 830 | ip6frag_high_thresh - INTEGER |
| 831 | Maximum memory used to reassemble IPv6 fragments. When |
| 832 | ip6frag_high_thresh bytes of memory is allocated for this purpose, |
| 833 | the fragment handler will toss packets until ip6frag_low_thresh |
| 834 | is reached. |
| 835 | |
| 836 | ip6frag_low_thresh - INTEGER |
| 837 | See ip6frag_high_thresh |
| 838 | |
| 839 | ip6frag_time - INTEGER |
| 840 | Time in seconds to keep an IPv6 fragment in memory. |
| 841 | |
| 842 | ip6frag_secret_interval - INTEGER |
| 843 | Regeneration interval (in seconds) of the hash secret (or lifetime |
| 844 | for the hash secret) for IPv6 fragments. |
| 845 | Default: 600 |
| 846 | |
| 847 | conf/default/*: |
| 848 | Change the interface-specific default settings. |
| 849 | |
| 850 | |
| 851 | conf/all/*: |
| 852 | Change all the interface-specific settings. |
| 853 | |
| 854 | [XXX: Other special features than forwarding?] |
| 855 | |
| 856 | conf/all/forwarding - BOOLEAN |
| 857 | Enable global IPv6 forwarding between all interfaces. |
| 858 | |
| 859 | IPv4 and IPv6 work differently here; e.g. netfilter must be used |
| 860 | to control which interfaces may forward packets and which not. |
| 861 | |
| 862 | This also sets all interfaces' Host/Router setting |
| 863 | 'forwarding' to the specified value. See below for details. |
| 864 | |
| 865 | This referred to as global forwarding. |
| 866 | |
YOSHIFUJI Hideaki | fbea49e | 2006-09-22 14:43:49 -0700 | [diff] [blame] | 867 | proxy_ndp - BOOLEAN |
| 868 | Do proxy ndp. |
| 869 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 870 | conf/interface/*: |
| 871 | Change special settings per interface. |
| 872 | |
| 873 | The functional behaviour for certain settings is different |
| 874 | depending on whether local forwarding is enabled or not. |
| 875 | |
| 876 | accept_ra - BOOLEAN |
| 877 | Accept Router Advertisements; autoconfigure using them. |
| 878 | |
| 879 | Functional default: enabled if local forwarding is disabled. |
| 880 | disabled if local forwarding is enabled. |
| 881 | |
YOSHIFUJI Hideaki | 65f5c7c | 2006-03-20 16:55:08 -0800 | [diff] [blame] | 882 | accept_ra_defrtr - BOOLEAN |
| 883 | Learn default router in Router Advertisement. |
| 884 | |
| 885 | Functional default: enabled if accept_ra is enabled. |
| 886 | disabled if accept_ra is disabled. |
| 887 | |
YOSHIFUJI Hideaki | c4fd30e | 2006-03-20 16:55:26 -0800 | [diff] [blame] | 888 | accept_ra_pinfo - BOOLEAN |
Matt LaPlante | 2fe0ae7 | 2006-10-03 22:50:39 +0200 | [diff] [blame] | 889 | Learn Prefix Information in Router Advertisement. |
YOSHIFUJI Hideaki | c4fd30e | 2006-03-20 16:55:26 -0800 | [diff] [blame] | 890 | |
| 891 | Functional default: enabled if accept_ra is enabled. |
| 892 | disabled if accept_ra is disabled. |
| 893 | |
YOSHIFUJI Hideaki | 09c884d | 2006-03-20 17:07:03 -0800 | [diff] [blame] | 894 | accept_ra_rt_info_max_plen - INTEGER |
| 895 | Maximum prefix length of Route Information in RA. |
| 896 | |
| 897 | Route Information w/ prefix larger than or equal to this |
| 898 | variable shall be ignored. |
| 899 | |
| 900 | Functional default: 0 if accept_ra_rtr_pref is enabled. |
| 901 | -1 if accept_ra_rtr_pref is disabled. |
| 902 | |
YOSHIFUJI Hideaki | 930d6ff | 2006-03-20 17:05:30 -0800 | [diff] [blame] | 903 | accept_ra_rtr_pref - BOOLEAN |
| 904 | Accept Router Preference in RA. |
| 905 | |
| 906 | Functional default: enabled if accept_ra is enabled. |
| 907 | disabled if accept_ra is disabled. |
| 908 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 909 | accept_redirects - BOOLEAN |
| 910 | Accept Redirects. |
| 911 | |
| 912 | Functional default: enabled if local forwarding is disabled. |
| 913 | disabled if local forwarding is enabled. |
| 914 | |
YOSHIFUJI Hideaki | 0bcbc92 | 2007-04-24 14:58:30 -0700 | [diff] [blame] | 915 | accept_source_route - INTEGER |
| 916 | Accept source routing (routing extension header). |
| 917 | |
YOSHIFUJI Hideaki | bb4dbf9 | 2007-07-10 22:55:49 -0700 | [diff] [blame] | 918 | >= 0: Accept only routing header type 2. |
YOSHIFUJI Hideaki | 0bcbc92 | 2007-04-24 14:58:30 -0700 | [diff] [blame] | 919 | < 0: Do not accept routing header. |
| 920 | |
| 921 | Default: 0 |
| 922 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 923 | autoconf - BOOLEAN |
| 924 | Autoconfigure addresses using Prefix Information in Router |
| 925 | Advertisements. |
| 926 | |
YOSHIFUJI Hideaki | c4fd30e | 2006-03-20 16:55:26 -0800 | [diff] [blame] | 927 | Functional default: enabled if accept_ra_pinfo is enabled. |
| 928 | disabled if accept_ra_pinfo is disabled. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 929 | |
| 930 | dad_transmits - INTEGER |
| 931 | The amount of Duplicate Address Detection probes to send. |
| 932 | Default: 1 |
| 933 | |
| 934 | forwarding - BOOLEAN |
| 935 | Configure interface-specific Host/Router behaviour. |
| 936 | |
| 937 | Note: It is recommended to have the same setting on all |
| 938 | interfaces; mixed router/host scenarios are rather uncommon. |
| 939 | |
| 940 | FALSE: |
| 941 | |
| 942 | By default, Host behaviour is assumed. This means: |
| 943 | |
| 944 | 1. IsRouter flag is not set in Neighbour Advertisements. |
| 945 | 2. Router Solicitations are being sent when necessary. |
| 946 | 3. If accept_ra is TRUE (default), accept Router |
| 947 | Advertisements (and do autoconfiguration). |
| 948 | 4. If accept_redirects is TRUE (default), accept Redirects. |
| 949 | |
| 950 | TRUE: |
| 951 | |
| 952 | If local forwarding is enabled, Router behaviour is assumed. |
| 953 | This means exactly the reverse from the above: |
| 954 | |
| 955 | 1. IsRouter flag is set in Neighbour Advertisements. |
| 956 | 2. Router Solicitations are not sent. |
| 957 | 3. Router Advertisements are ignored. |
| 958 | 4. Redirects are ignored. |
| 959 | |
| 960 | Default: FALSE if global forwarding is disabled (default), |
| 961 | otherwise TRUE. |
| 962 | |
| 963 | hop_limit - INTEGER |
| 964 | Default Hop Limit to set. |
| 965 | Default: 64 |
| 966 | |
| 967 | mtu - INTEGER |
| 968 | Default Maximum Transfer Unit |
| 969 | Default: 1280 (IPv6 required minimum) |
| 970 | |
YOSHIFUJI Hideaki | 52e1635 | 2006-03-20 17:05:47 -0800 | [diff] [blame] | 971 | router_probe_interval - INTEGER |
| 972 | Minimum interval (in seconds) between Router Probing described |
| 973 | in RFC4191. |
| 974 | |
| 975 | Default: 60 |
| 976 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 977 | router_solicitation_delay - INTEGER |
| 978 | Number of seconds to wait after interface is brought up |
| 979 | before sending Router Solicitations. |
| 980 | Default: 1 |
| 981 | |
| 982 | router_solicitation_interval - INTEGER |
| 983 | Number of seconds to wait between Router Solicitations. |
| 984 | Default: 4 |
| 985 | |
| 986 | router_solicitations - INTEGER |
| 987 | Number of Router Solicitations to send until assuming no |
| 988 | routers are present. |
| 989 | Default: 3 |
| 990 | |
| 991 | use_tempaddr - INTEGER |
| 992 | Preference for Privacy Extensions (RFC3041). |
| 993 | <= 0 : disable Privacy Extensions |
| 994 | == 1 : enable Privacy Extensions, but prefer public |
| 995 | addresses over temporary addresses. |
| 996 | > 1 : enable Privacy Extensions and prefer temporary |
| 997 | addresses over public addresses. |
| 998 | Default: 0 (for most devices) |
| 999 | -1 (for point-to-point devices and loopback devices) |
| 1000 | |
| 1001 | temp_valid_lft - INTEGER |
| 1002 | valid lifetime (in seconds) for temporary addresses. |
| 1003 | Default: 604800 (7 days) |
| 1004 | |
| 1005 | temp_prefered_lft - INTEGER |
| 1006 | Preferred lifetime (in seconds) for temporary addresses. |
| 1007 | Default: 86400 (1 day) |
| 1008 | |
| 1009 | max_desync_factor - INTEGER |
| 1010 | Maximum value for DESYNC_FACTOR, which is a random value |
| 1011 | that ensures that clients don't synchronize with each |
| 1012 | other and generate new addresses at exactly the same time. |
| 1013 | value is in seconds. |
| 1014 | Default: 600 |
| 1015 | |
| 1016 | regen_max_retry - INTEGER |
| 1017 | Number of attempts before give up attempting to generate |
| 1018 | valid temporary addresses. |
| 1019 | Default: 5 |
| 1020 | |
| 1021 | max_addresses - INTEGER |
| 1022 | Number of maximum addresses per interface. 0 disables limitation. |
| 1023 | It is recommended not set too large value (or 0) because it would |
| 1024 | be too easy way to crash kernel to allow to create too much of |
| 1025 | autoconfigured addresses. |
| 1026 | Default: 16 |
| 1027 | |
YOSHIFUJI Hideaki | 778d80b | 2008-06-28 14:17:11 +0900 | [diff] [blame] | 1028 | disable_ipv6 - BOOLEAN |
| 1029 | Disable IPv6 operation. |
| 1030 | Default: FALSE (enable IPv6 operation) |
| 1031 | |
YOSHIFUJI Hideaki | 1b34be7 | 2008-06-28 14:18:38 +0900 | [diff] [blame] | 1032 | accept_dad - INTEGER |
| 1033 | Whether to accept DAD (Duplicate Address Detection). |
| 1034 | 0: Disable DAD |
| 1035 | 1: Enable DAD (default) |
| 1036 | 2: Enable DAD, and disable IPv6 operation if MAC-based duplicate |
| 1037 | link-local address has been found. |
| 1038 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1039 | icmp/*: |
| 1040 | ratelimit - INTEGER |
| 1041 | Limit the maximal rates for sending ICMPv6 packets. |
Stephen Hemminger | 6dbf4bc | 2008-07-01 19:29:07 -0700 | [diff] [blame] | 1042 | 0 to disable any limiting, |
| 1043 | otherwise the minimal space between responses in milliseconds. |
| 1044 | Default: 1000 |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1045 | |
| 1046 | |
| 1047 | IPv6 Update by: |
| 1048 | Pekka Savola <pekkas@netcore.fi> |
| 1049 | YOSHIFUJI Hideaki / USAGI Project <yoshfuji@linux-ipv6.org> |
| 1050 | |
| 1051 | |
| 1052 | /proc/sys/net/bridge/* Variables: |
| 1053 | |
| 1054 | bridge-nf-call-arptables - BOOLEAN |
| 1055 | 1 : pass bridged ARP traffic to arptables' FORWARD chain. |
| 1056 | 0 : disable this. |
| 1057 | Default: 1 |
| 1058 | |
| 1059 | bridge-nf-call-iptables - BOOLEAN |
| 1060 | 1 : pass bridged IPv4 traffic to iptables' chains. |
| 1061 | 0 : disable this. |
| 1062 | Default: 1 |
| 1063 | |
| 1064 | bridge-nf-call-ip6tables - BOOLEAN |
| 1065 | 1 : pass bridged IPv6 traffic to ip6tables' chains. |
| 1066 | 0 : disable this. |
| 1067 | Default: 1 |
| 1068 | |
| 1069 | bridge-nf-filter-vlan-tagged - BOOLEAN |
Michael Milner | 516299d | 2007-04-12 22:14:23 -0700 | [diff] [blame] | 1070 | 1 : pass bridged vlan-tagged ARP/IP/IPv6 traffic to {arp,ip,ip6}tables. |
| 1071 | 0 : disable this. |
| 1072 | Default: 1 |
| 1073 | |
| 1074 | bridge-nf-filter-pppoe-tagged - BOOLEAN |
| 1075 | 1 : pass bridged pppoe-tagged IP/IPv6 traffic to {ip,ip6}tables. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1076 | 0 : disable this. |
| 1077 | Default: 1 |
| 1078 | |
| 1079 | |
| 1080 | UNDOCUMENTED: |
| 1081 | |
| 1082 | dev_weight FIXME |
| 1083 | discovery_slots FIXME |
| 1084 | discovery_timeout FIXME |
| 1085 | fast_poll_increase FIXME |
| 1086 | ip6_queue_maxlen FIXME |
| 1087 | lap_keepalive_time FIXME |
| 1088 | lo_cong FIXME |
| 1089 | max_baud_rate FIXME |
| 1090 | max_dgram_qlen FIXME |
| 1091 | max_noreply_time FIXME |
| 1092 | max_tx_data_size FIXME |
| 1093 | max_tx_window FIXME |
| 1094 | min_tx_turn_time FIXME |
| 1095 | mod_cong FIXME |
| 1096 | no_cong FIXME |
| 1097 | no_cong_thresh FIXME |
| 1098 | slot_timeout FIXME |
| 1099 | warn_noreply_time FIXME |
| 1100 | |