path: root/lib/stdlib/doc/src/supervisor.xml

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

<erlref>
  <header>
    <copyright>
      <year>1996</year><year>2014</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>supervisor</title>
    <prepared></prepared>
    <docno></docno>
    <date></date>
    <rev></rev>
  </header>
  <module>supervisor</module>
  <modulesummary>Generic Supervisor Behaviour</modulesummary>
  <description>
    <p>A behaviour module for implementing a supervisor, a process which
      supervises other processes called child processes. A child
      process can either be another supervisor or a worker process.
      Worker processes are normally implemented using one of
      the <c>gen_event</c>, <c>gen_fsm</c>, or <c>gen_server</c>
      behaviours. A supervisor implemented using this module will have
      a standard set of interface functions and include functionality
      for tracing and error reporting. Supervisors are used to build a
      hierarchical process structure called a supervision tree, a
      nice way to structure a fault tolerant application. Refer to
      <em>OTP Design Principles</em> for more information.</p>
    <p>A supervisor expects the definition of which child processes to
      supervise to be specified in a callback module exporting a
      pre-defined set of functions.</p>
    <p>Unless otherwise stated, all functions in this module will fail
      if the specified supervisor does not exist or if bad arguments
      are given.</p>
  </description>

  <section>
    <title>Supervision Principles</title>
    <p>The supervisor is responsible for starting, stopping and
      monitoring its child processes. The basic idea of a supervisor is
      that it shall keep its child processes alive by restarting them
      when necessary.</p>
    <p>The children of a supervisor are defined as a list of
      <em>child specifications</em>. When the supervisor is started, the child
      processes are started in order from left to right according to
      this list. When the supervisor terminates, it first terminates
      its child processes in reversed start order, from right to left.</p>
    <marker id="sup_flags"/>
    <p>The properties of a supervisor are defined by the supervisor
      flags. This is the type definition for the supervisor flags:
    </p>
    <pre>sup_flags() = #{strategy => strategy(),         % optional
                intensity => non_neg_integer(), % optional
                period => pos_integer()}        % optional
      </pre>
    <p>A supervisor can have one of the following <em>restart
	strategies</em>, specified with the <c>strategy</c> key in the
      above map:
    </p>
    <list type="bulleted">
      <item>
        <p><c>one_for_one</c> - if one child process terminates and
          should be restarted, only that child process is
          affected. This is the default restart strategy.</p>
      </item>
      <item>
        <p><c>one_for_all</c> - if one child process terminates and
          should be restarted, all other child processes are terminated
          and then all child processes are restarted.</p>
      </item>
      <item>
        <p><c>rest_for_one</c> - if one child process terminates and
          should be restarted, the 'rest' of the child processes --
          i.e. the child processes after the terminated child process
          in the start order -- are terminated. Then the terminated
          child process and all child processes after it are restarted.</p>
      </item>
      <item>
        <p><c>simple_one_for_one</c> - a simplified <c>one_for_one</c>
          supervisor, where all child processes are dynamically added
          instances of the same process type, i.e. running the same
          code.</p>
        <p>The functions <c>delete_child/2</c>
          and <c>restart_child/2</c> are invalid for
          <c>simple_one_for_one</c> supervisors and will return
          <c>{error,simple_one_for_one}</c> if the specified supervisor
          uses this restart strategy.</p>
        <p>The function <c>terminate_child/2</c> can be used for
          children under <c>simple_one_for_one</c> supervisors by
          giving the child's <c>pid()</c> as the second argument. If
          instead the child specification identifier is used,
          <c>terminate_child/2</c> will return
          <c>{error,simple_one_for_one}</c>.</p>
        <p>Because a <c>simple_one_for_one</c> supervisor could have
          many children, it shuts them all down asynchronously. This
          means that the children will do their cleanup in parallel,
          and therefore the order in which they are stopped is not
          defined.</p>
      </item>
    </list>
    <p>To prevent a supervisor from getting into an infinite loop of
      child process terminations and restarts, a <em>maximum restart
      intensity</em> is defined using two integer values specified
      with the <c>intensity</c> and <c>period</c> keys in the above
      map. Assuming the values <c>MaxR</c> for <c>intensity</c>
      and <c>MaxT</c> for <c>period</c>, then if more than <c>MaxR</c>
      restarts occur within <c>MaxT</c> seconds, the supervisor will
      terminate all child processes and then itself. The default value
      for <c>intensity</c> is <c>1</c>, and the default value
      for <c>period</c> is <c>5</c>.
    </p>
    <marker id="child_spec"/>
    <p>This is the type definition of a child specification:</p>
    <pre>child_spec() = #{id => child_id(),       % mandatory
                 start => mfargs(),      % mandatory
                 restart => restart(),   % optional
                 shutdown => shutdown(), % optional
                 type => worker(),       % optional
                 modules => modules()}   % optional</pre>
    <p>The old tuple format is kept for backwards compatibility,
      see <seealso marker="#type-child_spec">child_spec()</seealso>,
      but the map is preferred.
    </p>
    <list type="bulleted">
      <item>
        <p><c>id</c> is used to identify the child
          specification internally by the supervisor.</p>
	<p>The <c>id</c> key is mandatory.</p>
	<p>Note that this identifier on occations has been called
	  "name". As far as possible, the terms "identifier" or "id"
	  are now used but in order to keep backwards compatibility,
	  some occurences of "name" can still be found, for example
	  in error messages.</p>
      </item>
      <item>
        <p><c>start</c> defines the function call used to start the
          child process. It must be a module-function-arguments
          tuple <c>{M,F,A}</c> used as <c>apply(M,F,A)</c>.</p>
        <p>The start function <em>must create and link to</em> the child
          process, and must return <c>{ok,Child}</c> or
          <c>{ok,Child,Info}</c> where <c>Child</c> is the pid of
          the child process and <c>Info</c> an arbitrary term which is
          ignored by the supervisor.</p>
        <p>The start function can also return <c>ignore</c> if the child
          process for some reason cannot be started, in which case
          the child specification will be kept by the supervisor
	  (unless it is a temporary child) but the non-existing child
	  process will be ignored.</p>
        <p>If something goes wrong, the function may also return an
          error tuple <c>{error,Error}</c>.</p>
        <p>Note that the <c>start_link</c> functions of the different
          behaviour modules fulfill the above requirements.</p>
	<p>The <c>start</c> key is mandatory.</p>
      </item>
      <item>
        <p><c>restart</c> defines when a terminated child process
          shall be restarted. A <c>permanent</c> child process will
          always be restarted, a <c>temporary</c> child process will
          never be restarted (even when the supervisor's restart strategy
          is <c>rest_for_one</c> or <c>one_for_all</c> and a sibling's
          death causes the temporary process to be terminated) and a
          <c>transient</c> child process will be restarted only if
          it terminates abnormally, i.e. with another exit reason
          than <c>normal</c>, <c>shutdown</c> or <c>{shutdown,Term}</c>.</p>
	<p>The <c>restart</c> key is optional. If it is not given, the
	  default value <c>permanent</c> will be used.</p>
      </item>
      <item>
        <p><c>shutdown</c> defines how a child process shall be
          terminated. <c>brutal_kill</c> means the child process will
          be unconditionally terminated using <c>exit(Child,kill)</c>.
          An integer timeout value means that the supervisor will tell
          the child process to terminate by calling
          <c>exit(Child,shutdown)</c> and then wait for an exit signal
          with reason <c>shutdown</c> back from the child process. If
          no exit signal is received within the specified number of milliseconds,
          the child process is unconditionally terminated using
          <c>exit(Child,kill)</c>.</p>
        <p>If the child process is another supervisor, the shutdown time
          should be set to <c>infinity</c> to give the subtree ample
          time to shut down. It is also allowed to set it to <c>infinity</c>,
          if the child process is a worker.</p>
        <warning>
          <p>Be careful when setting the shutdown time to
          <c>infinity</c> when the child process is a worker. Because, in this
          situation, the termination of the supervision tree depends on the
          child process, it must be implemented in a safe way and its cleanup
          procedure must always return.</p>
        </warning>
        <p>Note that all child processes implemented using the standard
          OTP behaviour modules automatically adhere to the shutdown
          protocol.</p>
	<p>The <c>shutdown</c> key is optional. If it is not given,
	  the default value <c>5000</c> will be used if the child is
	  of type <c>worker</c>; and <c>infinity</c> will be used if
	  the child is of type <c>supervisor</c>.</p>
      </item>
      <item>
        <p><c>type</c> specifies if the child process is a supervisor or
          a worker.</p>
	<p>The <c>type</c> key is optional. If it is not given, the
	  default value <c>worker</c> will be used.</p>
      </item>
      <item>
        <p><c>modules</c> is used by the release handler during code
          replacement to determine which processes are using a certain
          module. As a rule of thumb, if the child process is a
          <c>supervisor</c>, <c>gen_server</c>, or <c>gen_fsm</c>,
	  this should be a list with one element <c>[Module]</c>,
	  where <c>Module</c> is the callback module. If the child
	  process is an event manager (<c>gen_event</c>) with a
	  dynamic set of callback modules, the value <c>dynamic</c>
	  shall be used. See <em>OTP Design Principles</em> for more
	  information about release handling.</p>
	<p>The <c>modules</c> key is optional. If it is not given, it
	  defaults to <c>[M]</c>, where <c>M</c> comes from the
	  child's start <c>{M,F,A}</c></p>
      </item>
      <item>
        <p>Internally, the supervisor also keeps track of the pid
          <c>Child</c> of the child process, or <c>undefined</c> if no
          pid exists.</p>
      </item>
    </list>
  </section>
  <datatypes>
    <datatype>
      <name name="child"/>
    </datatype>
    <datatype>
      <name name="child_id"/>
      <desc><p>Not a <c>pid()</c>.</p></desc>
    </datatype>
    <datatype>
      <name name="child_spec"/>
      <desc><p>The tuple format is kept for backwards compatibility
	 only. A map is preferred; see more details
	 <seealso marker="#child_spec">above</seealso>.</p></desc>
    </datatype>
    <datatype>
      <name name="mfargs"/>
      <desc>
	<p>The value <c>undefined</c> for <c><anno>A</anno></c> (the
	  argument list) is only to be used internally
	  in <c>supervisor</c>. If the restart type of the child
	  is <c>temporary</c>, then the process is never to be
	  restarted and therefore there is no need to store the real
	  argument list. The value <c>undefined</c> will then be
	  stored instead.</p>
      </desc>
    </datatype>
    <datatype>
      <name name="modules"/>
    </datatype>
    <datatype>
      <name name="restart"/>
    </datatype>
    <datatype>
      <name name="shutdown"/>
    </datatype>
    <datatype>
      <name name="strategy"/>
    </datatype>
    <datatype>
      <name name="sup_flags"/>
      <desc><p>The tuple format is kept for backwards compatibility
	 only. A map is preferred; see more details
	 <seealso marker="#sup_flags">above</seealso>.</p></desc>
    </datatype>
    <datatype>
      <name name="sup_ref"/>
    </datatype>
    <datatype>
      <name name="worker"/>
    </datatype>
  </datatypes>
  <funcs>
    <func>
      <name name="start_link" arity="2"/>
      <name name="start_link" arity="3"/>
      <fsummary>Create a supervisor process.</fsummary>
      <type name="startlink_ret"/>
      <type name="startlink_err"/>
      <type name="sup_name"/>
      <desc>
        <p>Creates a supervisor process as part of a supervision tree.
          The function will, among other things, ensure that
          the supervisor is linked to the calling process (its
          supervisor).</p>
        <p>The created supervisor process calls <c><anno>Module</anno>:init/1</c> to
          find out about restart strategy, maximum restart intensity
          and child processes. To ensure a synchronized start-up
          procedure, <c>start_link/2,3</c> does not return until
          <c><anno>Module</anno>:init/1</c> has returned and all child processes
          have been started.</p>
        <p>If <c><anno>SupName</anno>={local,Name}</c>, the supervisor is registered
          locally as <c>Name</c> using <c>register/2</c>. If
          <c><anno>SupName</anno>={global,Name}</c> the supervisor is registered
          globally as <c>Name</c> using <c>global:register_name/2</c>. If
          <c><anno>SupName</anno>={via,<anno>Module</anno>,<anno>Name</anno>}</c> the supervisor
	  is registered as <c>Name</c> using the registry represented by
	  <c>Module</c>. The <c>Module</c> callback must export the functions
	  <c>register_name/2</c>, <c>unregister_name/1</c> and <c>send/2</c>,
	  which shall behave like the corresponding functions in <c>global</c>.
	  Thus, <c>{via,global,<anno>Name</anno>}</c> is a valid reference.</p>
          <p>If no name is provided, the supervisor is not registered.</p>
        <p><c><anno>Module</anno></c> is the name of the callback module.</p>
        <p><c><anno>Args</anno></c> is an arbitrary term which is passed as
          the argument to <c><anno>Module</anno>:init/1</c>.</p>
        <p>If the supervisor and its child processes are successfully
          created (i.e. if all child process start functions return
          <c>{ok,Child}</c>, <c>{ok,Child,Info}</c>, or <c>ignore</c>),
          the function returns <c>{ok,Pid}</c>, where <c>Pid</c> is
          the pid of the supervisor. If there already exists a process
          with the specified <c><anno>SupName</anno></c>, the function returns
          <c>{error,{already_started,Pid}}</c>, where <c>Pid</c> is
          the pid of that process.</p>
        <p>If <c><anno>Module</anno>:init/1</c> returns <c>ignore</c>, this function
          returns <c>ignore</c> as well, and the supervisor terminates
          with reason <c>normal</c>.
          If <c><anno>Module</anno>:init/1</c> fails or returns an incorrect value,
          this function returns <c>{error,Term}</c> where <c>Term</c>
          is a term with information about the error, and the supervisor
          terminates with reason <c>Term</c>.</p>
        <p>If any child process start function fails or returns an error
          tuple or an erroneous value, the supervisor will first terminate
          all already started child processes with reason <c>shutdown</c>
          and then terminate itself and return
          <c>{error, {shutdown, Reason}}</c>.</p>
      </desc>
    </func>
    <func>
      <name name="start_child" arity="2"/>
      <fsummary>Dynamically add a child process to a supervisor.</fsummary>
      <type name="startchild_ret"/>
      <type name="startchild_err"/>
      <desc>
        <p>Dynamically adds a child specification to the supervisor
          <c><anno>SupRef</anno></c> which starts the corresponding child process.</p>
        <p><marker id="SupRef"/><c><anno>SupRef</anno></c> can be:</p>
        <list type="bulleted">
          <item>the pid,</item>
          <item><c>Name</c>, if the supervisor is locally registered,</item>
          <item><c>{Name,Node}</c>, if the supervisor is locally
           registered at another node, or</item>
          <item><c>{global,Name}</c>, if the supervisor is globally
           registered.</item>
          <item><c>{via,Module,Name}</c>, if the supervisor is registered
	  through an alternative process registry.</item>
        </list>
        <p><c><anno>ChildSpec</anno></c> must be a valid child specification
          (unless the supervisor is a <c>simple_one_for_one</c>
          supervisor; see below). The child process will be started by
          using the start function as defined in the child
          specification.</p>
        <p>In the case of a <c>simple_one_for_one</c> supervisor,
          the child specification defined in <c>Module:init/1</c> will
          be used, and <c><anno>ChildSpec</anno></c> shall instead be an arbitrary
          list of terms <c><anno>List</anno></c>. The child process will then be
          started by appending <c><anno>List</anno></c> to the existing start
          function arguments, i.e. by calling
          <c>apply(M, F, A++<anno>List</anno>)</c> where <c>{M,F,A}</c> is the start
          function defined in the child specification.</p>
        <p>If there already exists a child specification with
          the specified identifier, <c><anno>ChildSpec</anno></c> is discarded, and
          the function returns <c>{error,already_present}</c> or
          <c>{error,{already_started,<anno>Child</anno>}}</c>, depending on if
          the corresponding child process is running or not.</p>
        <p>If the child process start function returns <c>{ok,<anno>Child</anno>}</c>
          or <c>{ok,<anno>Child</anno>,<anno>Info</anno>}</c>, the child specification and pid are
          added to the supervisor and the function returns the same
          value.</p>
        <p>If the child process start function returns <c>ignore</c>,
          the child specification is added to the supervisor (unless the
          supervisor is a <c>simple_one_for_one</c> supervisor, see below),
          the pid is set to <c>undefined</c> and the function returns
          <c>{ok,undefined}</c>.
        </p>
        <p>In the case of a <c>simple_one_for_one</c> supervisor, when a child
          process start function returns <c>ignore</c> the functions returns
          <c>{ok,undefined}</c> and no child is added to the supervisor.
        </p>
        <p>If the child process start function returns an error tuple or
          an erroneous value, or if it fails, the child specification is
          discarded, and the function returns <c>{error,Error}</c> where
          <c>Error</c> is a term containing information about the error
          and child specification.</p>
      </desc>
    </func>
    <func>
      <name name="terminate_child" arity="2"/>
      <fsummary>Terminate a child process belonging to a supervisor.</fsummary>
      <desc>
        <p>Tells the supervisor <c><anno>SupRef</anno></c> to terminate the given
          child.</p>

        <p>If the supervisor is not <c>simple_one_for_one</c>,
          <c><anno>Id</anno></c> must be the child specification
          identifier. The process, if there is one, is terminated and,
          unless it is a temporary child, the child specification is
          kept by the supervisor. The child process may later be
          restarted by the supervisor. The child process can also be
          restarted explicitly by calling
          <c>restart_child/2</c>. Use <c>delete_child/2</c> to remove
          the child specification.</p>

	<p>If the child is temporary, the child specification is deleted as
	  soon as the process terminates. This means
	  that <c>delete_child/2</c> has no meaning,
	  and <c>restart_child/2</c> can not be used for these
	  children.</p>

        <p>If the supervisor is <c>simple_one_for_one</c>, <c><anno>Id</anno></c>
          must be the child process' <c>pid()</c>. If the specified
          process is alive, but is not a child of the given
          supervisor, the function will return
          <c>{error,not_found}</c>. If the child specification
          identifier is given instead of a <c>pid()</c>, the
          function will return <c>{error,simple_one_for_one}</c>.</p>
        <p>If successful, the function returns <c>ok</c>. If there is
          no child specification with the specified <c><anno>Id</anno></c>, the
          function returns <c>{error,not_found}</c>.</p>
        <p>See <seealso marker="#SupRef"><c>start_child/2</c></seealso>
          for a description of <c><anno>SupRef</anno></c>.</p>
      </desc>
    </func>
    <func>
      <name name="delete_child" arity="2"/>
      <fsummary>Delete a child specification from a supervisor.</fsummary>
      <desc>
        <p>Tells the supervisor <c><anno>SupRef</anno></c> to delete the child
          specification identified by <c><anno>Id</anno></c>. The corresponding child
          process must not be running. Use <c>terminate_child/2</c> to
          terminate it.</p>
        <p>See <seealso marker="#SupRef"><c>start_child/2</c></seealso>
          for a description of <c><anno>SupRef</anno></c>.</p>
        <p>If successful, the function returns <c>ok</c>. If the child
          specification identified by <c><anno>Id</anno></c> exists but
          the corresponding child process is running or about to be restarted,
	  the function returns <c>{error,running}</c> or
	  <c>{error,restarting}</c>, respectively. If the child specification
          identified by <c><anno>Id</anno></c> does not exist, the function
	  returns <c>{error,not_found}</c>.</p>
      </desc>
    </func>
    <func>
      <name name="restart_child" arity="2"/>
      <fsummary>Restart a terminated child process belonging to a supervisor.</fsummary>
      <desc>
        <p>Tells the supervisor <c><anno>SupRef</anno></c> to restart
          a child process corresponding to the child specification
          identified by <c><anno>Id</anno></c>. The child
          specification must exist, and the corresponding child process
          must not be running.</p>
	<p>Note that for temporary children, the child specification
	  is automatically deleted when the child terminates; thus
	  it is not possible to restart such children.</p>
        <p>See <seealso marker="#SupRef"><c>start_child/2</c></seealso>
          for a description of <c>SupRef</c>.</p>
        <p>If the child specification identified
          by <c><anno>Id</anno></c> does not exist, the function
          returns <c>{error,not_found}</c>. If the child specification
          exists but the corresponding process is already running, the
          function returns
          <c>{error,running}</c>.</p>
        <p>If the child process start function
          returns <c>{ok,<anno>Child</anno>}</c>
          or <c>{ok,<anno>Child</anno>,<anno>Info</anno>}</c>, the pid
          is added to the supervisor and the function returns the same
          value.</p>
        <p>If the child process start function returns <c>ignore</c>,
          the pid remains set to <c>undefined</c>, and the function
          returns <c>{ok,undefined}</c>.</p>
        <p>If the child process start function returns an error tuple
          or an erroneous value, or if it fails, the function returns
          <c>{error,<anno>Error</anno>}</c>
          where <c><anno>Error</anno></c> is a term containing
          information about the error.</p>
      </desc>
    </func>
    <func>
      <name name="which_children" arity="1"/>
      <fsummary>Return information about all children specifications and
        child processes belonging to a supervisor.</fsummary>
      <desc>
        <p>Returns a newly created list with information about all child
          specifications and child processes belonging to
          the supervisor <c><anno>SupRef</anno></c>.</p>
        <p>Note that calling this function when supervising a large
          number of children under low memory conditions can cause an
          out of memory exception.</p>
        <p>See <seealso marker="#SupRef"><c>start_child/2</c></seealso> for a description of
          <c>SupRef</c>.</p>
        <p>The information given for each child specification/process
          is:</p>
        <list type="bulleted">
          <item>
            <p><c><anno>Id</anno></c> - as defined in the child specification or
              <c>undefined</c> in the case of a
              <c>simple_one_for_one</c> supervisor.</p>
          </item>
          <item>
            <p><c><anno>Child</anno></c> - the pid of the corresponding child
              process, the atom <c>restarting</c> if the process is about to be
	      restarted, or <c>undefined</c> if there is no such process.</p>
          </item>
          <item>
            <p><c><anno>Type</anno></c> - as defined in the child specification.</p>
          </item>
          <item>
            <p><c><anno>Modules</anno></c> - as defined in the child specification.</p>
          </item>
        </list>
      </desc>
    </func>
    <func>
      <name name="count_children" arity="1"/>
      <fsummary>Return counts for the number of child specifications,
        active children, supervisors, and workers.</fsummary>
      <desc>
        <p>Returns a property list (see <c>proplists</c>) containing the
          counts for each of the following elements of the supervisor's
          child specifications and managed processes:</p>
        <list type="bulleted">
          <item>
            <p><c>specs</c> - the total count of children, dead or alive.</p>
          </item>
          <item>
            <p><c>active</c> - the count of all actively running child processes
              managed by this supervisor. In the case of <c>simple_one_for_one</c>
              supervisors, no check is carried out to ensure that each child process
              is still alive, though the result provided here is likely to be very
              accurate unless the supervisor is heavily overloaded.</p>
          </item>
          <item>
            <p><c>supervisors</c> - the count of all children marked as
              child_type = supervisor in the spec list, whether or not the
              child process is still alive.</p>
          </item>
          <item>
            <p><c>workers</c> - the count of all children marked as
              child_type = worker in the spec list, whether or not the child
              process is still alive.</p>
          </item>
        </list>
        <p>See <seealso marker="#SupRef"><c>start_child/2</c></seealso>
          for a description of <c><anno>SupRef</anno></c>.</p>
      </desc>
    </func>
    <func>
      <name name="check_childspecs" arity="1"/>
      <fsummary>Check if children specifications are syntactically correct.</fsummary>
      <desc>
        <p>This function takes a list of child specification as argument
          and returns <c>ok</c> if all of them are syntactically
          correct, or <c>{error,<anno>Error</anno>}</c> otherwise.</p>
      </desc>
    </func>
    <func>
      <name name="get_childspec" arity="2"/>
      <fsummary>Return the child specification map for the given
	child.</fsummary>
      <desc>
        <p>Returns the child specification map for the child identified
          by <c>Id</c> under supervisor <c>SupRef</c>. The returned
          map contains all keys, both mandatory and optional.</p>
        <p>See <seealso marker="#SupRef"><c>start_child/2</c></seealso>
          for a description of <c><anno>SupRef</anno></c>.</p>
      </desc>
    </func>
  </funcs>

  <section>
    <title>CALLBACK FUNCTIONS</title>
    <p>The following functions must be exported from a
      <c>supervisor</c> callback module.</p>
  </section>
  <funcs>
    <func>
      <name>Module:init(Args) -> Result</name>
      <fsummary>Return a supervisor specification.</fsummary>
      <type>
        <v>Args = term()</v>
        <v>Result = {ok,{SupFlags,[ChildSpec]}} | ignore</v>
        <v>&nbsp;SupFlags = <seealso marker="#type-sup_flags">sup_flags()</seealso></v>
        <v>&nbsp;ChildSpec = <seealso marker="#type-child_spec">child_spec()</seealso></v>
      </type>
      <desc>
        <p>Whenever a supervisor is started using
          <c>supervisor:start_link/2,3</c>, this function is called by
          the new process to find out about restart strategy, maximum
          restart intensity, and child specifications.</p>
        <p><c>Args</c> is the <c>Args</c> argument provided to the start
          function.</p>
        <p><c>SupFlags</c> is the supervisor flags defining the
          restart strategy and max restart intensity for the
          supervisor. <c>[ChildSpec]</c> is a list of valid child
          specifications defining which child processes the supervisor
          shall start and monitor. See the discussion about
          Supervision Principles above.</p>
        <p>Note that when the restart strategy is
          <c>simple_one_for_one</c>, the list of child specifications
          must be a list with one child specification only.
          (The child specification identifier is ignored.) No child process is then started
          during the initialization phase, but all children are assumed
          to be started dynamically using
          <c>supervisor:start_child/2</c>.</p>
        <p>The function may also return <c>ignore</c>.</p>
	<p>Note that this function might also be called as a part of a
	  code upgrade procedure. For this reason, the function should
	  not have any side effects. See
	  <seealso marker="doc/design_principles:appup_cookbook#sup">Design
	  Principles</seealso> for more information about code upgrade
	  of supervisors.</p>
      </desc>
    </func>
  </funcs>

  <section>
    <title>SEE ALSO</title>
    <p><seealso marker="gen_event">gen_event(3)</seealso>, 
      <seealso marker="gen_fsm">gen_fsm(3)</seealso>, 
      <seealso marker="gen_server">gen_server(3)</seealso>, 
      <seealso marker="sys">sys(3)</seealso></p>
  </section>
</erlref>