path: root/lib/kernel/doc/src/net_kernel.xml

<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE erlref SYSTEM "erlref.dtd">

<erlref>
  <header>
    <copyright>
      <year>1996</year><year>2018</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,&lt;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 Kernel configuration parameter
      <c>dist_auto_connect</c> to <c>never</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>
    <warning>
      <p>
        Starting a distributed node without also specifying
        <seealso marker="erts:erl#proto_dist"><c>-proto_dist inet_tls</c></seealso>
        will expose the node to attacks that may give the attacker
        complete access to the node and in extension the cluster.
        When using un-secure distributed nodes, make sure that the
        network is configured to keep potential attackers out.
        See the <seealso marker="ssl:ssl_distribution">
        Using SSL for Erlang Distribution</seealso> User's Guide
        for details on how to setup a secure distributed node.
      </p>
    </warning>
  </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 a connection was established or was already
	  established or if <c><anno>Node</anno></c> is the local node
	  itself. Returns <c>false</c> if the connection attempt failed, 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="getopts" arity="2"/>
      <fsummary>Get distribution socket options.</fsummary>
      <desc>
        <p>Get one or more options for the distribution socket
	connected to <c><anno>Node</anno></c>.</p>
        <p>If <c><anno>Node</anno></c> is a connected node
	the return value is the same as from
	<seealso marker="inet#getopts/2"><c>inet:getopts(Sock, Options)</c></seealso>
	where <c>Sock</c> is the distribution socket for <c><anno>Node</anno></c>.</p>
	<p>Returns <c>ignored</c> if the local node is not alive or
	<c>{error, noconnection}</c> if <c><anno>Node</anno></c> is not connected.</p>
      </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 Kernel version 2.11.4, and ERTS 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 Kernel
          versions before 2.11.4.</p>
        <p>As from Kernel 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 Kernel version 2.13, and ERTS 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 Kernel
          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, depending on which
	      distribution module or process that is used be any term,
	      but for the standard TCP distribution module it is
	      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 name="setopts" arity="2"/>
      <fsummary>Set distribution socket options.</fsummary>
      <desc>
        <p>Set one or more options for distribution sockets.
	Argument <c><anno>Node</anno></c> can be either one node name
	or the atom <c>new</c> to affect the distribution sockets of all
	future connected nodes.</p>
        <p>The return value is the same as from
	<seealso marker="inet#setopts/2"><c>inet:setopts/2</c></seealso>
	or <c>{error, noconnection}</c> if <c><anno>Node</anno></c> is not
	a connected node or <c>new</c>.</p>
	<p>If <c><anno>Node</anno></c> is <c>new</c> the <c><anno>Options</anno></c>
	will then also be added to kernel configration parameters
	<seealso marker="kernel:kernel_app#inet_dist_listen_options">inet_dist_listen_options</seealso>
	and
	<seealso marker="kernel:kernel_app#inet_dist_connect_options">inet_dist_connect_options</seealso>.</p>
	<p>Returns <c>ignored</c> if the local node is not alive.</p>
      </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>