1 Documentation for /proc/sys/net/*
2 (c) 1999 Terrehon Bowden <terrehon@pacbell.net>
3 Bodo Bauer <bb@ricochet.net>
4 (c) 2000 Jorge Nerin <comandante@zaralinux.com>
5 (c) 2009 Shen Feng <shen@cn.fujitsu.com>
7 For general info and legal blurb, please look in README.
9 ==============================================================
11 This file contains the documentation for the sysctl files in
14 The interface to the networking parts of the kernel is located in
15 /proc/sys/net. The following table shows all possible subdirectories. You may
16 see only some of them, depending on your kernel's configuration.
19 Table : Subdirectories in /proc/sys/net
20 ..............................................................................
21 Directory Content Directory Content
22 core General parameter appletalk Appletalk protocol
23 unix Unix domain sockets netrom NET/ROM
24 802 E802 protocol ax25 AX25
25 ethernet Ethernet protocol rose X.25 PLP layer
26 ipv4 IP version 4 x25 X.25 protocol
27 ipx IPX token-ring IBM token ring
28 bridge Bridging decnet DEC net
29 ipv6 IP version 6 tipc TIPC
30 ..............................................................................
32 1. /proc/sys/net/core - Network core options
33 -------------------------------------------------------
38 This enables Berkeley Packet Filter Just in Time compiler.
39 Currently supported on x86_64 architecture, bpf_jit provides a framework
40 to speed packet filtering, the one used by tcpdump/libpcap for example.
42 0 - disable the JIT (default value)
44 2 - enable the JIT and ask the compiler to emit traces on kernel log.
49 The maximum number of packets that kernel can handle on a NAPI interrupt,
50 it's a Per-CPU variable.
56 The default queuing discipline to use for network devices. This allows
57 overriding the default queue discipline of pfifo_fast with an
58 alternative. Since the default queuing discipline is created with the
59 no additional parameters so is best suited to queuing disciplines that
60 work well without configuration like stochastic fair queue (sfq),
61 CoDel (codel) or fair queue CoDel (fq_codel). Don't use queuing disciplines
62 like Hierarchical Token Bucket or Deficit Round Robin which require setting
63 up classes and bandwidths.
68 Low latency busy poll timeout for socket reads. (needs CONFIG_NET_RX_BUSY_POLL)
69 Approximate time in us to busy loop waiting for packets on the device queue.
70 This sets the default value of the SO_BUSY_POLL socket option.
71 Can be set or overridden per socket by setting socket option SO_BUSY_POLL,
72 which is the preferred method of enabling. If you need to enable the feature
73 globally via sysctl, a value of 50 is recommended.
74 Will increase power usage.
79 Low latency busy poll timeout for poll and select. (needs CONFIG_NET_RX_BUSY_POLL)
80 Approximate time in us to busy loop waiting for events.
81 Recommended value depends on the number of sockets you poll on.
82 For several sockets 50, for several hundreds 100.
83 For more than that you probably want to use epoll.
84 Note that only sockets with SO_BUSY_POLL set will be busy polled,
85 so you want to either selectively set SO_BUSY_POLL on those sockets or set
86 sysctl.net.busy_read globally.
87 Will increase power usage.
93 The default setting of the socket receive buffer in bytes.
98 The maximum receive socket buffer size in bytes.
102 Allow processes to receive tx timestamps looped together with the original
103 packet contents. If disabled, transmit timestamp requests from unprivileged
104 processes are dropped unless socket option SOF_TIMESTAMPING_OPT_TSONLY is set.
111 The default setting (in bytes) of the socket send buffer.
116 The maximum send socket buffer size in bytes.
118 message_burst and message_cost
119 ------------------------------
121 These parameters are used to limit the warning messages written to the kernel
122 log from the networking code. They enforce a rate limit to make a
123 denial-of-service attack impossible. A higher message_cost factor, results in
124 fewer messages that will be written. Message_burst controls when messages will
125 be dropped. The default settings limit warning messages to one every five
131 This sysctl is now unused.
133 This was used to control console messages from the networking stack that
134 occur because of problems on the network like duplicate address or bad
137 These messages are now emitted at KERN_DEBUG and can generally be enabled
138 and controlled by the dynamic_debug facility.
143 Maximum number of packets taken from all interfaces in one polling cycle (NAPI
144 poll). In one polling cycle interfaces which are registered to polling are
145 probed in a round-robin manner.
150 Maximum number of packets, queued on the INPUT side, when the interface
151 receives packets faster than kernel can process them.
156 RSS (Receive Side Scaling) enabled drivers use a 40 bytes host key that is
158 Some user space might need to gather its content even if drivers do not
159 provide ethtool -x support yet.
161 myhost:~# cat /proc/sys/net/core/netdev_rss_key
162 84:50:f4:00:a8:15:d1:a7:e9:7f:1d:60:35:c7:47:25:42:97:74:ca:56:bb:b6:a1:d8: ... (52 bytes total)
164 File contains nul bytes if no driver ever called netdev_rss_key_fill() function.
166 /proc/sys/net/core/netdev_rss_key contains 52 bytes of key,
167 but most drivers only use 40 bytes of it.
169 myhost:~# ethtool -x eth0
170 RX flow hash indirection table for eth0 with 8 RX ring(s):
173 84:50:f4:00:a8:15:d1:a7:e9:7f:1d:60:35:c7:47:25:42:97:74:ca:56:bb:b6:a1:d8:43:e3:c9:0c:fd:17:55:c2:3a:4d:69:ed:f1:42:89
175 netdev_tstamp_prequeue
176 ----------------------
178 If set to 0, RX packet timestamps can be sampled after RPS processing, when
179 the target CPU processes packets. It might give some delay on timestamps, but
180 permit to distribute the load on several cpus.
182 If set to 1 (default), timestamps are sampled as soon as possible, before
188 Maximum ancillary buffer size allowed per socket. Ancillary data is a sequence
189 of struct cmsghdr structures with appended data.
191 2. /proc/sys/net/unix - Parameters for Unix domain sockets
192 -------------------------------------------------------
194 There is only one file in this directory.
195 unix_dgram_qlen limits the max number of datagrams queued in Unix domain
196 socket's buffer. It will not take effect unless PF_UNIX flag is specified.
199 3. /proc/sys/net/ipv4 - IPV4 settings
200 -------------------------------------------------------
201 Please see: Documentation/networking/ip-sysctl.txt and ipvs-sysctl.txt for
202 descriptions of these entries.
206 -------------------------------------------------------
208 The /proc/sys/net/appletalk directory holds the Appletalk configuration data
209 when Appletalk is loaded. The configurable parameters are:
214 The amount of time we keep an ARP entry before expiring it. Used to age out
220 The amount of time we will spend trying to resolve an Appletalk address.
222 aarp-retransmit-limit
223 ---------------------
225 The number of times we will retransmit a query before giving up.
230 Controls the rate at which expires are checked.
232 The directory /proc/net/appletalk holds the list of active Appletalk sockets
235 The fields indicate the DDP type, the local address (in network:node format)
236 the remote address, the size of the transmit pending queue, the size of the
237 received queue (bytes waiting for applications to read) the state and the uid
240 /proc/net/atalk_iface lists all the interfaces configured for appletalk.It
241 shows the name of the interface, its Appletalk address, the network range on
242 that address (or network number for phase 1 networks), and the status of the
245 /proc/net/atalk_route lists each known network route. It lists the target
246 (network) that the route leads to, the router (may be directly connected), the
247 route flags, and the device the route is using.
251 -------------------------------------------------------
253 The IPX protocol has no tunable values in proc/sys/net.
255 The IPX protocol does, however, provide proc/net/ipx. This lists each IPX
256 socket giving the local and remote addresses in Novell format (that is
257 network:node:port). In accordance with the strange Novell tradition,
258 everything but the port is in hex. Not_Connected is displayed for sockets that
259 are not tied to a specific remote address. The Tx and Rx queue sizes indicate
260 the number of bytes pending for transmission and reception. The state
261 indicates the state the socket is in and the uid is the owning uid of the
264 The /proc/net/ipx_interface file lists all IPX interfaces. For each interface
265 it gives the network number, the node number, and indicates if the network is
266 the primary network. It also indicates which device it is bound to (or
267 Internal for internal networks) and the Frame Type if appropriate. Linux
268 supports 802.3, 802.2, 802.2 SNAP and DIX (Blue Book) ethernet framing for
271 The /proc/net/ipx_route table holds a list of IPX routes. For each route it
272 gives the destination network, the router node (or Directly) and the network
273 address of the router (or Connected) for internal networks.
276 -------------------------------------------------------
281 The TIPC protocol now has a tunable for the receive memory, similar to the
282 tcp_rmem - i.e. a vector of 3 INTEGERs: (min, default, max)
284 # cat /proc/sys/net/tipc/tipc_rmem
285 4252725 34021800 68043600
288 The max value is set to CONN_OVERLOAD_LIMIT, and the default and min values
289 are scaled (shifted) versions of that same value. Note that the min value
290 is not at this point in time used in any meaningful way, but the triplet is
291 preserved in order to be consistent with things like tcp_rmem.
296 TIPC name table updates are distributed asynchronously in a cluster, without
297 any form of transaction handling. This means that different race scenarios are
298 possible. One such is that a name withdrawal sent out by one node and received
299 by another node may arrive after a second, overlapping name publication already
300 has been accepted from a third node, although the conflicting updates
301 originally may have been issued in the correct sequential order.
302 If named_timeout is nonzero, failed topology updates will be placed on a defer
303 queue until another event arrives that clears the error, or until the timeout
304 expires. Value is in milliseconds.