Erlang logo
User's Guide
Reference Manual
Release Notes
PDF
Top

STDLIB
Reference Manual
Version 2.8


Expand All
Contract All

Table of Contents

supervisor

MODULE

supervisor

MODULE SUMMARY

Generic Supervisor Behaviour

DESCRIPTION

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 gen_event, gen_fsm, or gen_server 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 OTP Design Principles for more information.

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.

Unless otherwise stated, all functions in this module will fail if the specified supervisor does not exist or if bad arguments are given.

Supervision Principles

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.

The children of a supervisor are defined as a list of child specifications. 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.

The properties of a supervisor are defined by the supervisor flags. This is the type definition for the supervisor flags:

sup_flags() = #{strategy => strategy(),         % optional
                intensity => non_neg_integer(), % optional
                period => pos_integer()}        % optional
      

A supervisor can have one of the following restart strategies, specified with the strategy key in the above map:

  • one_for_one - if one child process terminates and should be restarted, only that child process is affected. This is the default restart strategy.

  • one_for_all - if one child process terminates and should be restarted, all other child processes are terminated and then all child processes are restarted.

  • rest_for_one - 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.

  • simple_one_for_one - a simplified one_for_one supervisor, where all child processes are dynamically added instances of the same process type, i.e. running the same code.

    The functions delete_child/2 and restart_child/2 are invalid for simple_one_for_one supervisors and will return {error,simple_one_for_one} if the specified supervisor uses this restart strategy.

    The function terminate_child/2 can be used for children under simple_one_for_one supervisors by giving the child's pid() as the second argument. If instead the child specification identifier is used, terminate_child/2 will return {error,simple_one_for_one}.

    Because a simple_one_for_one 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.

To prevent a supervisor from getting into an infinite loop of child process terminations and restarts, a maximum restart intensity is defined using two integer values specified with the intensity and period keys in the above map. Assuming the values MaxR for intensity and MaxT for period, then if more than MaxR restarts occur within MaxT seconds, the supervisor will terminate all child processes and then itself. The default value for intensity is 1, and the default value for period is 5.

This is the type definition of a child specification:

child_spec() = #{id => child_id(),       % mandatory
                 start => mfargs(),      % mandatory
                 restart => restart(),   % optional
                 shutdown => shutdown(), % optional
                 type => worker(),       % optional
                 modules => modules()}   % optional

The old tuple format is kept for backwards compatibility, see child_spec(), but the map is preferred.

  • id is used to identify the child specification internally by the supervisor.

    The id key is mandatory.

    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.

  • start defines the function call used to start the child process. It must be a module-function-arguments tuple {M,F,A} used as apply(M,F,A).

    The start function must create and link to the child process, and must return {ok,Child} or {ok,Child,Info} where Child is the pid of the child process and Info an arbitrary term which is ignored by the supervisor.

    The start function can also return ignore 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.

    If something goes wrong, the function may also return an error tuple {error,Error}.

    Note that the start_link functions of the different behaviour modules fulfill the above requirements.

    The start key is mandatory.

  • restart defines when a terminated child process shall be restarted. A permanent child process will always be restarted, a temporary child process will never be restarted (even when the supervisor's restart strategy is rest_for_one or one_for_all and a sibling's death causes the temporary process to be terminated) and a transient child process will be restarted only if it terminates abnormally, i.e. with another exit reason than normal, shutdown or {shutdown,Term}.

    The restart key is optional. If it is not given, the default value permanent will be used.

  • shutdown defines how a child process shall be terminated. brutal_kill means the child process will be unconditionally terminated using exit(Child,kill). An integer timeout value means that the supervisor will tell the child process to terminate by calling exit(Child,shutdown) and then wait for an exit signal with reason shutdown back from the child process. If no exit signal is received within the specified number of milliseconds, the child process is unconditionally terminated using exit(Child,kill).

    If the child process is another supervisor, the shutdown time should be set to infinity to give the subtree ample time to shut down. It is also allowed to set it to infinity, if the child process is a worker.

    Warning

    Be careful when setting the shutdown time to infinity 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.

    Note that all child processes implemented using the standard OTP behaviour modules automatically adhere to the shutdown protocol.

    The shutdown key is optional. If it is not given, the default value 5000 will be used if the child is of type worker; and infinity will be used if the child is of type supervisor.

  • type specifies if the child process is a supervisor or a worker.

    The type key is optional. If it is not given, the default value worker will be used.

  • modules 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 supervisor, gen_server, or gen_fsm, this should be a list with one element [Module], where Module is the callback module. If the child process is an event manager (gen_event) with a dynamic set of callback modules, the value dynamic shall be used. See OTP Design Principles for more information about release handling.

    The modules key is optional. If it is not given, it defaults to [M], where M comes from the child's start {M,F,A}

  • Internally, the supervisor also keeps track of the pid Child of the child process, or undefined if no pid exists.

DATA TYPES

child() = undefined | pid()

child_id() = term()

Not a pid().

child_spec() =
    #{id => child_id(),
      start => mfargs(),
      restart => restart(),
      shutdown => shutdown(),
      type => worker(),
      modules => modules()} |
    {Id :: child_id(),
     StartFunc :: mfargs(),
     Restart :: restart(),
     Shutdown :: shutdown(),
     Type :: worker(),
     Modules :: modules()}

The tuple format is kept for backwards compatibility only. A map is preferred; see more details above.

mfargs() =
    {M :: module(), F :: atom(), A :: [term()] | undefined}

The value undefined for A (the argument list) is only to be used internally in supervisor. If the restart type of the child is temporary, then the process is never to be restarted and therefore there is no need to store the real argument list. The value undefined will then be stored instead.

modules() = [module()] | dynamic

restart() = permanent | transient | temporary

shutdown() = brutal_kill | timeout()

strategy() =
    one_for_all | one_for_one | rest_for_one | simple_one_for_one

sup_flags() =
    #{strategy => strategy(),
      intensity => integer() >= 0,
      period => integer() >= 1} |
    {RestartStrategy :: strategy(),
     Intensity :: integer() >= 0,
     Period :: integer() >= 1}

The tuple format is kept for backwards compatibility only. A map is preferred; see more details above.

sup_ref() =
    (Name :: atom()) |
    {Name :: atom(), Node :: node()} |
    {global, Name :: atom()} |
    {via, Module :: module(), Name :: any()} |
    pid()

worker() = worker | supervisor

EXPORTS

start_link(Module, Args) -> startlink_ret()
start_link(SupName, Module, Args) -> startlink_ret()

Types:

SupName = sup_name()
Module = module()
Args = term()
startlink_ret() =
    {ok, pid()} | ignore | {error, startlink_err()}

startlink_err() =
    {already_started, pid()} | {shutdown, term()} | term()

sup_name() =
    {local, Name :: atom()} |
    {global, Name :: atom()} |
    {via, Module :: module(), Name :: any()}

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).

The created supervisor process calls Module:init/1 to find out about restart strategy, maximum restart intensity and child processes. To ensure a synchronized start-up procedure, start_link/2,3 does not return until Module:init/1 has returned and all child processes have been started.

If SupName={local,Name}, the supervisor is registered locally as Name using register/2. If SupName={global,Name} the supervisor is registered globally as Name using global:register_name/2. If SupName={via,Module,Name} the supervisor is registered as Name using the registry represented by Module. The Module callback must export the functions register_name/2, unregister_name/1 and send/2, which shall behave like the corresponding functions in global. Thus, {via,global,Name} is a valid reference.

If no name is provided, the supervisor is not registered.

Module is the name of the callback module.

Args is an arbitrary term which is passed as the argument to Module:init/1.

If the supervisor and its child processes are successfully created (i.e. if all child process start functions return {ok,Child}, {ok,Child,Info}, or ignore), the function returns {ok,Pid}, where Pid is the pid of the supervisor. If there already exists a process with the specified SupName, the function returns {error,{already_started,Pid}}, where Pid is the pid of that process.

If Module:init/1 returns ignore, this function returns ignore as well, and the supervisor terminates with reason normal. If Module:init/1 fails or returns an incorrect value, this function returns {error,Term} where Term is a term with information about the error, and the supervisor terminates with reason Term.

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 shutdown and then terminate itself and return {error, {shutdown, Reason}}.

start_child(SupRef, ChildSpec) -> startchild_ret()

Types:

SupRef = sup_ref()
ChildSpec = child_spec() | (List :: [term()])
startchild_ret() =
    {ok, Child :: child()} |
    {ok, Child :: child(), Info :: term()} |
    {error, startchild_err()}

startchild_err() =
    already_present | {already_started, Child :: child()} | term()

Dynamically adds a child specification to the supervisor SupRef which starts the corresponding child process.

SupRef can be:

  • the pid,
  • Name, if the supervisor is locally registered,
  • {Name,Node}, if the supervisor is locally registered at another node, or
  • {global,Name}, if the supervisor is globally registered.
  • {via,Module,Name}, if the supervisor is registered through an alternative process registry.

ChildSpec must be a valid child specification (unless the supervisor is a simple_one_for_one supervisor; see below). The child process will be started by using the start function as defined in the child specification.

In the case of a simple_one_for_one supervisor, the child specification defined in Module:init/1 will be used, and ChildSpec shall instead be an arbitrary list of terms List. The child process will then be started by appending List to the existing start function arguments, i.e. by calling apply(M, F, A++List) where {M,F,A} is the start function defined in the child specification.

If there already exists a child specification with the specified identifier, ChildSpec is discarded, and the function returns {error,already_present} or {error,{already_started,Child}}, depending on if the corresponding child process is running or not.

If the child process start function returns {ok,Child} or {ok,Child,Info}, the child specification and pid are added to the supervisor and the function returns the same value.

If the child process start function returns ignore, the child specification is added to the supervisor (unless the supervisor is a simple_one_for_one supervisor, see below), the pid is set to undefined and the function returns {ok,undefined}.

In the case of a simple_one_for_one supervisor, when a child process start function returns ignore the functions returns {ok,undefined} and no child is added to the supervisor.

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 {error,Error} where Error is a term containing information about the error and child specification.

terminate_child(SupRef, Id) -> Result

Types:

SupRef = sup_ref()
Id = pid() | child_id()
Result = ok | {error, Error}
Error = not_found | simple_one_for_one

Tells the supervisor SupRef to terminate the given child.

If the supervisor is not simple_one_for_one, Id 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 restart_child/2. Use delete_child/2 to remove the child specification.

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

If the supervisor is simple_one_for_one, Id must be the child process' pid(). If the specified process is alive, but is not a child of the given supervisor, the function will return {error,not_found}. If the child specification identifier is given instead of a pid(), the function will return {error,simple_one_for_one}.

If successful, the function returns ok. If there is no child specification with the specified Id, the function returns {error,not_found}.

See start_child/2 for a description of SupRef.

delete_child(SupRef, Id) -> Result

Types:

SupRef = sup_ref()
Result = ok | {error, Error}
Error = running | restarting | not_found | simple_one_for_one

Tells the supervisor SupRef to delete the child specification identified by Id. The corresponding child process must not be running. Use terminate_child/2 to terminate it.

See start_child/2 for a description of SupRef.

If successful, the function returns ok. If the child specification identified by Id exists but the corresponding child process is running or about to be restarted, the function returns {error,running} or {error,restarting}, respectively. If the child specification identified by Id does not exist, the function returns {error,not_found}.

restart_child(SupRef, Id) -> Result

Types:

SupRef = sup_ref()
Result =
    {ok, Child :: child()} |
    {ok, Child :: child(), Info :: term()} |
    {error, Error}
Error =
    running | restarting | not_found | simple_one_for_one | term()

Tells the supervisor SupRef to restart a child process corresponding to the child specification identified by Id. The child specification must exist, and the corresponding child process must not be running.

Note that for temporary children, the child specification is automatically deleted when the child terminates; thus it is not possible to restart such children.

See start_child/2 for a description of SupRef.

If the child specification identified by Id does not exist, the function returns {error,not_found}. If the child specification exists but the corresponding process is already running, the function returns {error,running}.

If the child process start function returns {ok,Child} or {ok,Child,Info}, the pid is added to the supervisor and the function returns the same value.

If the child process start function returns ignore, the pid remains set to undefined, and the function returns {ok,undefined}.

If the child process start function returns an error tuple or an erroneous value, or if it fails, the function returns {error,Error} where Error is a term containing information about the error.

which_children(SupRef) -> [{Id, Child, Type, Modules}]

Types:

SupRef = sup_ref()
Id = child_id() | undefined
Child = child() | restarting
Type = worker()
Modules = modules()

Returns a newly created list with information about all child specifications and child processes belonging to the supervisor SupRef.

Note that calling this function when supervising a large number of children under low memory conditions can cause an out of memory exception.

See start_child/2 for a description of SupRef.

The information given for each child specification/process is:

  • Id - as defined in the child specification or undefined in the case of a simple_one_for_one supervisor.

  • Child - the pid of the corresponding child process, the atom restarting if the process is about to be restarted, or undefined if there is no such process.

  • Type - as defined in the child specification.

  • Modules - as defined in the child specification.

count_children(SupRef) -> PropListOfCounts

Types:

SupRef = sup_ref()
PropListOfCounts = [Count]
Count =
    {specs, ChildSpecCount :: integer() >= 0} |
    {active, ActiveProcessCount :: integer() >= 0} |
    {supervisors, ChildSupervisorCount :: integer() >= 0} |
    {workers, ChildWorkerCount :: integer() >= 0}

Returns a property list (see proplists) containing the counts for each of the following elements of the supervisor's child specifications and managed processes:

  • specs - the total count of children, dead or alive.

  • active - the count of all actively running child processes managed by this supervisor. In the case of simple_one_for_one 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.

  • supervisors - the count of all children marked as child_type = supervisor in the spec list, whether or not the child process is still alive.

  • workers - the count of all children marked as child_type = worker in the spec list, whether or not the child process is still alive.

See start_child/2 for a description of SupRef.

check_childspecs(ChildSpecs) -> Result

Types:

ChildSpecs = [child_spec()]
Result = ok | {error, Error :: term()}

This function takes a list of child specification as argument and returns ok if all of them are syntactically correct, or {error,Error} otherwise.

get_childspec(SupRef, Id) -> Result

Types:

SupRef = sup_ref()
Id = pid() | child_id()
Result = {ok, child_spec()} | {error, Error}
Error = not_found

Returns the child specification map for the child identified by Id under supervisor SupRef. The returned map contains all keys, both mandatory and optional.

See start_child/2 for a description of SupRef.

CALLBACK FUNCTIONS

The following functions must be exported from a supervisor callback module.

EXPORTS

Module:init(Args) -> Result

Types:

Args = term()
Result = {ok,{SupFlags,[ChildSpec]}} | ignore
 SupFlags = sup_flags()
 ChildSpec = child_spec()

Whenever a supervisor is started using supervisor:start_link/2,3, this function is called by the new process to find out about restart strategy, maximum restart intensity, and child specifications.

Args is the Args argument provided to the start function.

SupFlags is the supervisor flags defining the restart strategy and max restart intensity for the supervisor. [ChildSpec] is a list of valid child specifications defining which child processes the supervisor shall start and monitor. See the discussion about Supervision Principles above.

Note that when the restart strategy is simple_one_for_one, 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 supervisor:start_child/2.

The function may also return ignore.

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 Design Principles for more information about code upgrade of supervisors.

SEE ALSO