<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE erlref SYSTEM "erlref.dtd">
<erlref>
<header>
<copyright>
<year>1996</year><year>2013</year>
<holder>Ericsson AB. All Rights Reserved.</holder>
</copyright>
<legalnotice>
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
</legalnotice>
<title>net_kernel</title>
<prepared>Claes Wikstrom</prepared>
<docno>1</docno>
<date>1996-09-10</date>
<rev>A</rev>
</header>
<module>net_kernel</module>
<modulesummary>Erlang networking kernel.</modulesummary>
<description>
<p>The net kernel is a system process, registered as
<c>net_kernel</c>, which must be operational for distributed Erlang
to work. The purpose of this process is to implement parts of
the BIFs <c>spawn/4</c> and <c>spawn_link/4</c>, and to provide
monitoring of the network.</p>
<p>An Erlang node is started using command-line flag
<c>-name</c> or <c>-sname</c>:</p>
<pre>
$ <input>erl -sname foobar</input></pre>
<p>It is also possible to call <c>net_kernel:start([foobar])</c>
directly from the normal Erlang shell prompt:</p>
<pre>
1> <input>net_kernel:start([foobar, shortnames]).</input>
{ok,<0.64.0>}
(foobar@gringotts)2></pre>
<p>If the node is started with command-line flag <c>-sname</c>,
the node name is <c>foobar@Host</c>, where <c>Host</c> is
the short name of the host (not the fully qualified domain name).
If started with flag <c>-name</c>, the node name is <c>foobar@Host</c>,
where <c>Host</c> is the fully qualified domain name.
For more information, see
<seealso marker="erts:erl"><c>erl</c></seealso>.</p>
<p>Normally, connections are established automatically when
another node is referenced. This functionality can be disabled
by setting <c>Kernel</c> configuration parameter
<c>dist_auto_connect</c> to <c>false</c>, see
<seealso marker="kernel_app"><c>kernel(6)</c></seealso>. In this case,
connections must be established explicitly by calling
<seealso marker="#connect_node/1"><c>connect_node/1</c></seealso>.</p>
<p>Which nodes that are allowed to communicate with each other is handled
by the magic cookie system, see section
<seealso marker="doc/reference_manual:distributed">Distributed Erlang</seealso>
in the Erlang Reference Manual.</p>
</description>
<funcs>
<func>
<name name="allow" arity="1"/>
<fsummary>Permit access to a specified set of nodes</fsummary>
<desc>
<p>Permits access to the specified set of nodes.</p>
<p>Before the first call to <c>allow/1</c>, any node with the correct
cookie can be connected. When <c>allow/1</c> is called, a list
of allowed nodes is established. Any access attempts made from (or to)
nodes not in that list will be rejected.</p>
<p>Subsequent calls to <c>allow/1</c> will add the specified nodes
to the list of allowed nodes. It is not possible to remove nodes
from the list.</p>
<p>Returns <c>error</c> if any element in <c><anno>Nodes</anno></c> is not
an atom.</p>
</desc>
</func>
<func>
<name name="connect_node" arity="1"/>
<fsummary>Establish a connection to a node.</fsummary>
<desc>
<p>Establishes a connection to <c><anno>Node</anno></c>. Returns
<c>true</c> if successful, <c>false</c> if not, and <c>ignored</c>
if the local node is not alive.</p>
</desc>
</func>
<func>
<name name="get_net_ticktime" arity="0"/>
<fsummary>Get <c>net_ticktime</c>.</fsummary>
<desc>
<p>Gets <c>net_ticktime</c> (see
<seealso marker="kernel_app"><c>kernel(6)</c></seealso>).</p>
<p>Defined return values (<c><anno>Res</anno></c>):</p>
<taglist>
<tag><c><anno>NetTicktime</anno></c></tag>
<item><p><c>net_ticktime</c> is <c><anno>NetTicktime</anno></c>
seconds.</p></item>
<tag><c>{ongoing_change_to, <anno>NetTicktime</anno>}</c></tag>
<item><p><c>net_kernel</c> is currently changing
<c>net_ticktime</c> to <c><anno>NetTicktime</anno></c>
seconds.</p></item>
<tag><c>ignored</c></tag>
<item><p>The local node is not alive.</p></item>
</taglist>
</desc>
</func>
<func>
<name name="monitor_nodes" arity="1"/>
<name name="monitor_nodes" arity="2"/>
<fsummary>Subscribe to node status change messages.</fsummary>
<desc>
<p>The calling process subscribes or unsubscribes to node
status change messages. A <c>nodeup</c> message is delivered
to all subscribing processes when a new node is connected, and
a <c>nodedown</c> message is delivered when a node is
disconnected.</p>
<p>If <c><anno>Flag</anno></c> is <c>true</c>, a new subscription is
started. If <c><anno>Flag</anno></c> is <c>false</c>, all previous
subscriptions started with the same <c><anno>Options</anno></c>
are stopped. Two
option lists are considered the same if they contain the same
set of options.</p>
<p>As from <c>Kernel</c> version 2.11.4, and <c>ERTS</c> version
5.5.4, the following is guaranteed:</p>
<list type="bulleted">
<item><p><c>nodeup</c> messages are delivered before delivery
of any message from the remote node passed through the
newly established connection.</p></item>
<item><p><c>nodedown</c> messages are not delivered until all
messages from the remote node that have been passed
through the connection have been delivered.</p></item>
</list>
<p>Notice that this is <em>not</em> guaranteed for <c>Kernel</c>
versions before 2.11.4.</p>
<p>As from <c>Kernel</c> version 2.11.4, subscriptions can also be
made before the <c>net_kernel</c> server is started, that is,
<c>net_kernel:monitor_nodes/[1,2]</c> does not return
<c>ignored</c>.</p>
<p>As from <c>Kernel</c> version 2.13, and <c>ERTS</c> version
5.7, the following is guaranteed:</p>
<list type="bulleted">
<item><p><c>nodeup</c> messages are delivered after the
corresponding node appears in results from
<c>erlang:nodes/X</c>.</p></item>
<item><p><c>nodedown</c> messages are delivered after the
corresponding node has disappeared in results from
<c>erlang:nodes/X</c>.</p></item>
</list>
<p>Notice that this is <em>not</em> guaranteed for <c>Kernel</c>
versions before 2.13.</p>
<p>The format of the node status change messages depends on
<c><anno>Options</anno></c>. If <c><anno>Options</anno></c> is
<c>[]</c>, which is the default, the format is as follows:</p>
<code type="none">
{nodeup, Node} | {nodedown, Node}
Node = node()</code>
<p>If <c><anno>Options</anno></c> is not <c>[]</c>, the format is
as follows:</p>
<code type="none">
{nodeup, Node, InfoList} | {nodedown, Node, InfoList}
Node = node()
InfoList = [{Tag, Val}]</code>
<p><c>InfoList</c> is a list of tuples. Its contents depends on
<c><anno>Options</anno></c>, see below.</p>
<p>Also, when <c>OptionList == []</c>, only visible nodes, that
is, nodes that appear in the result of
<seealso marker="erts:erlang#nodes/0"><c>erlang:nodes/0</c></seealso>,
are monitored.</p>
<p><c><anno>Option</anno></c> can be any of the following:</p>
<taglist>
<tag><c>{node_type, NodeType}</c></tag>
<item>
<p>Valid values for <c>NodeType</c>:</p>
<taglist>
<tag><c>visible</c></tag>
<item><p>Subscribe to node status change messages for visible
nodes only. The tuple <c>{node_type, visible}</c> is
included in <c>InfoList</c>.</p></item>
<tag><c>hidden</c></tag>
<item><p>Subscribe to node status change messages for hidden
nodes only. The tuple <c>{node_type, hidden}</c> is
included in <c>InfoList</c>.</p></item>
<tag><c>all</c></tag>
<item><p>Subscribe to node status change messages for both
visible and hidden nodes. The tuple
<c>{node_type, visible | hidden}</c> is included in
<c>InfoList</c>.</p></item>
</taglist>
</item>
<tag><c>nodedown_reason</c></tag>
<item>
<p>The tuple <c>{nodedown_reason, Reason}</c> is included in
<c>InfoList</c> in <c>nodedown</c> messages.</p>
<p><c>Reason</c> can be any of the following:</p>
<taglist>
<tag><c>connection_setup_failed</c></tag>
<item><p>The connection setup failed (after <c>nodeup</c>
messages were sent).</p></item>
<tag><c>no_network</c></tag>
<item><p>No network is available.</p></item>
<tag><c>net_kernel_terminated</c></tag>
<item><p>The <c>net_kernel</c> process terminated.</p></item>
<tag><c>shutdown</c></tag>
<item><p>Unspecified connection shutdown.</p></item>
<tag><c>connection_closed</c></tag>
<item><p>The connection was closed.</p></item>
<tag><c>disconnect</c></tag>
<item><p>The connection was disconnected (forced from the
current node).</p></item>
<tag><c>net_tick_timeout</c></tag>
<item><p>Net tick time-out.</p></item>
<tag><c>send_net_tick_failed</c></tag>
<item><p>Failed to send net tick over the connection.</p></item>
<tag><c>get_status_failed</c></tag>
<item><p>Status information retrieval from the <c>Port</c>
holding the connection failed.</p></item>
</taglist>
</item>
</taglist>
</desc>
</func>
<func>
<name name="set_net_ticktime" arity="1"/>
<name name="set_net_ticktime" arity="2"/>
<fsummary>Set <c>net_ticktime</c>.</fsummary>
<desc>
<p>Sets <c>net_ticktime</c> (see
<seealso marker="kernel_app"><c>kernel(6)</c></seealso>) to
<c><anno>NetTicktime</anno></c> seconds.
<c><anno>TransitionPeriod</anno></c> defaults to <c>60</c>.</p>
<p>Some definitions:</p>
<taglist>
<tag>Minimum transition traffic interval (<c>MTTI</c>)</tag>
<item><p><c>minimum(<anno>NetTicktime</anno>,
PreviousNetTicktime)*1000 div 4</c> milliseconds.</p></item>
<tag>Transition period</tag>
<item><p>The time of the least number of consecutive <c>MTTI</c>s
to cover <c><anno>TransitionPeriod</anno></c> seconds following
the call to <c>set_net_ticktime/2</c> (that is,
((<c><anno>TransitionPeriod</anno>*1000 - 1) div MTTI + 1)*MTTI</c>
milliseconds).</p></item>
</taglist>
<p>If
<c><![CDATA[NetTicktime < PreviousNetTicktime]]></c>,
the <c>net_ticktime</c> change is done at the end of
the transition period; otherwise at the beginning. During
the transition period, <c>net_kernel</c> ensures that
there is outgoing traffic on all connections at least
every <c>MTTI</c> millisecond.</p>
<note>
<p>The <c>net_ticktime</c> changes must be initiated on all
nodes in the network (with the same <c><anno>NetTicktime</anno></c>)
before the end of any transition period on any node;
otherwise connections can erroneously be disconnected.</p>
</note>
<p>Returns one of the following:</p>
<taglist>
<tag><c>unchanged</c></tag>
<item>
<p><c>net_ticktime</c> already has the value of
<c><anno>NetTicktime</anno></c> and is left unchanged.</p>
</item>
<tag><c>change_initiated</c></tag>
<item>
<p><c>net_kernel</c> initiated the change of
<c>net_ticktime</c> to <c><anno>NetTicktime</anno></c>
seconds.</p>
</item>
<tag><c>{ongoing_change_to, <anno>NewNetTicktime</anno>}</c></tag>
<item>
<p>The request is <em>ignored</em> because
<c>net_kernel</c> is busy changing <c>net_ticktime</c> to
<c><anno>NewNetTicktime</anno></c> seconds.</p>
</item>
</taglist>
</desc>
</func>
<func>
<name>start([Name]) -> {ok, pid()} | {error, Reason}</name>
<name>start([Name, NameType]) -> {ok, pid()} | {error, Reason}</name>
<name>start([Name, NameType, Ticktime]) -> {ok, pid()} | {error, Reason}</name>
<fsummary>Turn an Erlang runtime system into a distributed node.</fsummary>
<type>
<v>Name = atom()</v>
<v>NameType = shortnames | longnames</v>
<v>Reason = {already_started, pid()} | term()</v>
</type>
<desc>
<p>Turns a non-distributed node into a distributed node by
starting <c>net_kernel</c> and other necessary processes.</p>
<p>Notice that the argument is a list with exactly one, two, or
three arguments. <c>NameType</c> defaults to <c>longnames</c>
and <c>Ticktime</c> to <c>15000</c>.</p>
</desc>
</func>
<func>
<name name="stop" arity="0"/>
<fsummary>Turn a node into a non-distributed Erlang runtime system.</fsummary>
<desc>
<p>Turns a distributed node into a non-distributed node. For
other nodes in the network, this is the same as the node
going down. Only possible when the net kernel was started using
<seealso marker="#start/1"><c>start/1</c></seealso>,
otherwise <c>{error, not_allowed}</c> is returned. Returns
<c>{error, not_found}</c> if the local node is not alive.</p>
</desc>
</func>
</funcs>
</erlref>
