erpc

erpc

erpc
Enhanced Remote Procedure Call
Module erpc was introduced in OTP 23.0.

This module provide services similar to Remote Procedure Calls. A remote procedure call is a method to call a function on a remote node and collect the answer. It is used for collecting information on a remote node, or for running a function with some specific side effects on the remote node.

This is an enhanced subset of the operations provided by the rpc module. Enhanced in the sense that it makes it possible to distinguish between returned value, raised exceptions, and other errors. erpc also has better performance and scalability than the original rpc implementation. However, current rpc module will utilize erpc in order to also provide these properties when possible.

In order for an erpc operation to succeed, the remote node also needs to support erpc. Typically only ordinary Erlang nodes as of OTP 23 have erpc support.

Note that it is up to the user to ensure that correct code to execute via erpc is available on the involved nodes.

Timeout relative to current time in milliseconds.

Infinite timeout. That is, the operation will never time out.

An absolute Erlang monotonic time timeout in milliseconds. That is, the operation will time out when erlang:monotonic_time(millisecond) returns a value larger than or equal to Timeout. Timeout is not allowed to identify a time further into the future than 4294967295 milliseconds. Identifying the timeout using an absolute timeout value is especially handy when you have a deadline for responses corresponding to a complete collection of requests (request_id_collection()) , since you do not have to recalculate the relative time until the deadline over and over again.

Types

Node = node()
Timeout = timeout_time()
Result = term()

The same as calling erpc:call(Node, erlang, apply, [Fun,[]], Timeout). May raise all the same exceptions as call/5 plus an {erpc, badarg} error exception if Fun is not a fun of zero arity.

The call erpc:call(Node,Fun) is the same as the call erpc:call(Node,Fun,infinity).

Types

Node = node()
Module = Function = atom()
Args = [term()]
Timeout = timeout_time()
Result = term()

Evaluates apply(Module, Function, Args) on node Node and returns the corresponding value Result. Timeout sets an upper time limit for the call operation to complete.

The call erpc:call(Node, Module, Function, Args) is equivalent to the call erpc:call(Node, Module, Function, Args, infinity)

The call() function only returns if the applied function successfully returned without raising any uncaught exceptions, the operation did not time out, and no failures occurred. In all other cases an exception is raised. The following exceptions, listed by exception class, can currently be raised by call():

The applied function called throw(Value) and did not catch this exception. The exception reason Value equals the argument passed to throw/1.

Exception reason:

The applied function called exit(ExitReason) and did not catch this exception. The exit reason ExitReason equals the argument passed to exit/1.

The process that applied the function received an exit signal and terminated due to this signal. The process terminated with exit reason ExitReason.

Exception reason:

A runtime error occurred which raised an error exception while applying the function, and the applied function did not catch the exception. The error reason ErrorReason indicates the type of error that occurred. StackTrace is formatted as when caught in a try/catch construct. The StackTrace is limited to the applied function and functions called by it.

The erpc operation failed. The following ERpcErrorReasons are the most common ones:

If any one of these are true:

  • Node is not an atom.

  • Module is not an atom.

  • Function is not an atom.

  • Args is not a list. Note that the list is not verified to be a proper list at the client side.

  • Timeout is invalid.

The connection to Node was lost or could not be established. The function may or may not be applied.

The erpc operation failed due to some system limit being reached. This typically due to failure to create a process on the remote node Node, but can be other things as well.

The erpc operation timed out. The function may or may not be applied.

The remote node Node does not support this erpc operation.

If the erpc operation fails, but it is unknown if the function is/will be applied (that is, a timeout or a connection loss), the caller will not receive any further information about the result if/when the applied function completes. If the applied function explicitly communicates with the calling process, such communication may, of course, reach the calling process.

Note

You cannot make any assumptions about the process that will perform the apply(). It may be the calling process itself, a server, or a freshly spawned process.

Types

Node = node()
Module = Function = atom()
Args = [term()]

Evaluates apply(Module, Function, Args) on node Node. No response is delivered to the calling process. cast() returns immediately after the cast request has been sent. Any failures beside bad arguments are silently ignored.

cast/4 fails with an {erpc, badarg} error exception if:

  • Node is not an atom.

  • Module is not an atom.

  • Function is not an atom.

  • Args is not a list. Note that the list is not verified to be a proper list at the client side.

Note

You cannot make any assumptions about the process that will perform the apply(). It may be a server, or a freshly spawned process.

Types

Message = term()
RequestId = request_id()
Result = term()

Check if a message is a response to a call request previously made by the calling process using send_request/4. RequestId should be the value returned from the previously made send_request/4 call, and the corresponding response should not already have been received and handled to completion by check_response/2, receive_response/2, or wait_response/2. Message is the message to check.

If Message does not correspond to the response, the atom no_response is returned. If Message corresponds to the response, the call operation is completed and either the result is returned as {response, Result} where Result corresponds to the value returned from the applied function or an exception is raised. The exceptions that can be raised corresponds to the same exceptions as can be raised by call/4. That is, no {erpc, timeout} error exception can be raised. check_response() will fail with an {erpc, badarg} exception if/when an invalid RequestId is detected.

If the erpc operation fails, but it is unknown if the function is/will be applied (that is, a connection loss), the caller will not receive any further information about the result if/when the applied function completes. If the applied function explicitly communicates with the calling process, such communication may, of course, reach the calling process.

Types

Message = term()
RequestIdCollection = request_id_collection()
Delete = boolean()
Result = Label = term()
NewRequestIdCollection = request_id_collection()

Check if a message is a response to a call request corresponding to a request identifier saved in RequestIdCollection. All request identifiers of RequestIdCollection must correspond to requests that have been made using send_request/4 or send_request/6, and all requests must have been made by the process calling this function.

Label is the label associated with the request identifier of the request that the response corresponds to. A request identifier is associated with a label when adding a request identifier in a request identifier collection, or when sending the request using send_request/6.

Compared to check_response/2, the returned result associated with a specific request identifier or an exception associated with a specific request identifier will be wrapped in a 3-tuple. The first element of this tuple equals the value that would have been produced by check_response/2, the second element equals the Label associated with the specific request identifier, and the third element NewRequestIdCollection is a possibly modified request identifier collection. The error exception {erpc, badarg} is not associated with any specific request identifier, and will hence not be wrapped.

If RequestIdCollection is empty, the atom no_request will be returned. If Message does not correspond to any of the request identifiers in RequestIdCollection, the atom no_response is returned.

If Delete equals true, the association with Label will have been deleted from RequestIdCollection in the resulting NewRequestIdCollection. If Delete equals false, NewRequestIdCollection will equal RequestIdCollection. Note that deleting an association is not for free and that a collection containing already handled requests can still be used by subsequent calls to check_response/3, receive_response/3, and wait_response/3. However, without deleting handled associations, the above calls will not be able to detect when there are no more outstanding requests to handle, so you will have to keep track of this some other way than relying on a no_request return. Note that if you pass a collection only containing associations of already handled or abandoned requests to check_response/3, it will always return no_response.

Note that a response might have been consumed uppon an {erpc, badarg} exception and if so, will be lost for ever.

Types

Nodes = [atom()]
Timeout = timeout_time()
Result = term()

The same as calling erpc:multicall(Nodes, erlang, apply, [Fun,[]], Timeout). May raise all the same exceptions as multicall/5 plus an {erpc, badarg} error exception if Fun is not a fun of zero arity.

The call erpc:multicall(Nodes,Fun) is the same as the call erpc:multicall(Nodes,Fun, infinity).

Types

Nodes = [atom()]
Module = Function = atom()
Args = [term()]
Timeout = timeout_time()
Result =
    [{ok, ReturnValue :: term()} | caught_call_exception()]
caught_call_exception() =
    {throw, Throw :: term()} |
    {exit, {exception, Reason :: term()}} |
    {error,
     {exception, Reason :: term(), StackTrace :: [stack_item()]}} |
    {exit, {signal, Reason :: term()}} |
    {error, {erpc, Reason :: term()}}

stack_item() =
    {Module :: atom(),
     Function :: atom(),
     Arity :: arity() | (Args :: [term()]),
     Location ::
         [{file, Filename :: string()} |
          {line, Line :: integer() >= 1}]}

Performs multiple call operations in parallel on multiple nodes. That is, evaluates apply(Module, Function, Args) on the nodes Nodes in parallel. Timeout sets an upper time limit for all call operations to complete. The result is returned as a list where the result from each node is placed at the same position as the node name is placed in Nodes. Each item in the resulting list is formatted as either:

The call operation for this specific node returned Result.

The call operation for this specific node raised an exception of class Class with exception reason ExceptionReason. These correspond to the exceptions that call/5 can raise.

multicall/5 fails with an {erpc, badarg} error exception if:

  • Nodes is not a proper list of atoms. Note that some requests may already have been sent when the failure occurs. That is, the function may or may not be applied on some nodes.

  • Module is not an atom.

  • Function is not an atom.

  • Args is not a list. Note that the list is not verified to be a proper list at the client side.

The call erpc:multicall(Nodes, Module, Function, Args) is equivalent to the call erpc:multicall(Nodes, Module, Function, Args, infinity). These calls are also equivalent to calling my_multicall(Nodes, Module, Function, Args) below if one disregard performance and failure behavior. multicall() can utilize a selective receive optimization which removes the need to scan the message queue from the beginning in order to find a matching message. The send_request()/receive_response() combination can, however, not utilize this optimization.

my_multicall(Nodes, Module, Function, Args) ->
  ReqIds = lists:map(fun (Node) ->
                       erpc:send_request(Node, Module, Function, Args)
                     end,
                     Nodes),
  lists:map(fun (ReqId) ->
              try
                {ok, erpc:receive_response(ReqId, infinity)}
              catch
                Class:Reason ->
                  {Class, Reason}
              end
            end,
            ReqIds).

If an erpc operation fails, but it is unknown if the function is/will be applied (that is, a timeout, connection loss, or an improper Nodes list), the caller will not receive any further information about the result if/when the applied function completes. If the applied function communicates with the calling process, such communication may, of course, reach the calling process.

Note

You cannot make any assumptions about the process that will perform the apply(). It may be the calling process itself, a server, or a freshly spawned process.

Types

Nodes = [node()]
Module = Function = atom()
Args = [term()]

Evaluates apply(Module, Function, Args) on the nodes Nodes. No response is delivered to the calling process. multicast() returns immediately after the cast requests have been sent. Any failures beside bad arguments are silently ignored.

multicast/4 fails with an {erpc, badarg} error exception if:

  • Nodes is not a proper list of atoms. Note that some requests may already have been sent when the failure occurs. That is, the function may or may not be applied on some nodes.

  • Module is not an atom.

  • Function is not an atom.

  • Args is not a list. Note that the list is not verified to be a proper list at the client side.

Note

You cannot make any assumptions about the process that will perform the apply(). It may be a server, or a freshly spawned process.

Types

RequestId = request_id()
Timeout = timeout_time()
Result = term()

Receive a response to a call request previously made by the calling process using send_request/4. RequestId should be the value returned from the previously made send_request/4 call, and the corresponding response should not already have been received and handled to completion by receive_response(), check_response/4, or wait_response/4.

Timeout sets an upper time limit on how long to wait for a response. If the operation times out, the request identified by RequestId will be abandoned, then an {erpc, timeout} error exception will be raised. That is, no response corresponding to the request will ever be received after a timeout. If a response is received, the call operation is completed and either the result is returned or an exception is raised. The exceptions that can be raised corresponds to the same exceptions as can be raised by call/5. receive_response/2 will fail with an {erpc, badarg} exception if/when an invalid RequestId is detected or if an invalid Timeout is passed.

A call to the function my_call(Node, Module, Function, Args, Timeout) below is equivalent to the call erpc:call(Node, Module, Function, Args, Timeout) if one disregards performance. call() can utilize a selective receive optimization which removes the need to scan the message queue from the beginning in order to find a matching message. The send_request()/receive_response() combination can, however, not utilize this optimization.

my_call(Node, Module, Function, Args, Timeout) ->
  RequestId = erpc:send_request(Node, Module, Function, Args),
  erpc:receive_response(RequestId, Timeout).

If the erpc operation fails, but it is unknown if the function is/will be applied (that is, a timeout, or a connection loss), the caller will not receive any further information about the result if/when the applied function completes. If the applied function explicitly communicates with the calling process, such communication may, of course, reach the calling process.

Types

RequestIdCollection = request_id_collection()
Timeout = timeout_time()
Delete = boolean()
Result = Label = term()
NewRequestIdCollection = request_id_collection()

Receive a response to a call request corresponding to a request identifier saved in RequestIdCollection. All request identifiers of RequestIdCollection must correspond to requests that have been made using send_request/4 or send_request/6, and all requests must have been made by the process calling this function.

Label is the label associated with the request identifier of the request that the response corresponds to. A request identifier is associated with a label when adding a request identifier in a request identifier collection, or when sending the request using send_request/6.

Compared to receive_response/2, the returned result associated with a specific request identifier or an exception associated with a specific request identifier will be wrapped in a 3-tuple. The first element of this tuple equals the value that would have been produced by receive_response/2, the second element equals the Label associated with the specific request identifier, and the third element NewRequestIdCollection is a possibly modified request identifier collection. The error exceptions {erpc, badarg} and {erpc, timeout} are not associated with any specific request identifiers, and will hence not be wrapped.

If RequestIdCollection is empty, the atom no_request will be returned.

If the operation times out, all requests identified by RequestIdCollection will be abandoned, then an {erpc, timeout} error exception will be raised. That is, no responses corresponding to any of the request identifiers in RequestIdCollection will ever be received after a timeout. The difference between receive_response/3 and wait_response/3 is that receive_response/3 abandons the requests at timeout so that any potential future responses are ignored, while wait_response/3 does not.

If Delete equals true, the association with Label will have been deleted from RequestIdCollection in the resulting NewRequestIdCollection. If Delete equals false, NewRequestIdCollection will equal RequestIdCollection. Note that deleting an association is not for free and that a collection containing already handled requests can still be used by subsequent calls to receive_response/3, check_response/3, and wait_response/3. However, without deleting handled associations, the above calls will not be able to detect when there are no more outstanding requests to handle, so you will have to keep track of this some other way than relying on a no_request return. Note that if you pass a collection only containing associations of already handled or abandoned requests to receive_response/3, it will always block until a timeout determined by Timeout is triggered.

Note that a response might have been consumed uppon an {erpc, badarg} exception and if so, will be lost for ever.

Saves RequestId and associates a Label with the request identifier by adding this information to RequestIdCollection and returning the resulting request identifier collection.

Returns a new empty request identifier collection. A request identifier collection can be utilized in order the handle multiple outstanding requests.

Request identifiers of requests made by send_request/4 can be saved in a request identifier collection using reqids_add/3. Such a collection of request identifiers can later be used in order to get one response corresponding to a request in the collection by passing the collection as argument to check_response/3, receive_response/3, and wait_response/3.

reqids_size/1 can be used to determine the amount of request identifiers in a request identifier collection.

Returns the amount of request identifiers saved in RequestIdCollection.

Returns a list of {RequestId, Label} tuples which corresponds to all request identifiers with their associated labels present in the RequestIdCollection collection.

Types

Node = node()
RequestId = request_id()

The same as calling erpc:send_request(Node, erlang, apply, [Fun, []]).

Fails with an {erpc, badarg} error exception if:

  • Node is not an atom.

  • Fun is not a fun of zero arity.

Note

You cannot make any assumptions about the process that will perform the apply(). It may be a server, or a freshly spawned process.

Types

Node = node()
Module = Function = atom()
Args = [term()]
RequestId = request_id()

Send an asynchronous call request to the node Node. send_request/4 returns a request identifier that later is to be passed to either receive_response/2, wait_response/2, or, check_response/2 in order to get the response of the call request. Besides passing the request identifier directly to these functions, it can also be added in a request identifier collection using reqids_add/3. Such a collection of request identifiers can later be used in order to get one response corresponding to a request in the collection by passing the collection as argument to receive_response/3, wait_response/3, or, check_response/3. If you are about to save the request identifier in a request identifier collection, you may want to consider using send_request/6 instead.

A call to the function my_call(Node, Module, Function, Args, Timeout) below is equivalent to the call erpc:call(Node, Module, Function, Args, Timeout) if one disregards performance. call() can utilize a selective receive optimization which removes the need to scan the message queue from the beginning in order to find a matching message. The send_request()/receive_response() combination can, however, not utilize this optimization.

my_call(Node, Module, Function, Args, Timeout) ->
  RequestId = erpc:send_request(Node, Module, Function, Args),
  erpc:receive_response(RequestId, Timeout).

Fails with an {erpc, badarg} error exception if:

  • Node is not an atom.

  • Module is not an atom.

  • Function is not an atom.

  • Args is not a list. Note that the list is not verified to be a proper list at the client side.

Note

You cannot make any assumptions about the process that will perform the apply(). It may be a server, or a freshly spawned process.

Types

Node = node()
Label = term()
RequestIdCollection = NewRequestIdCollection = request_id_collection()

The same as calling erpc:send_request(Node, erlang, apply, [Fun,[]]), Label, RequestIdCollection).

Fails with an {erpc, badarg} error exception if:

  • Node is not an atom.

  • Fun is not a fun of zero arity.

  • RequestIdCollection is detected not to be request identifier collection.

Note

You cannot make any assumptions about the process that will perform the apply(). It may be a server, or a freshly spawned process.

Types

Node = node()
Module = Function = atom()
Args = [term()]
Label = term()
RequestIdCollection = NewRequestIdCollection = request_id_collection()

Send an asynchronous call request to the node Node. The Label will be associated with the request identifier of the operation and added to the returned request identifier collection NewRequestIdCollection. The collection can later be used in order to get one response corresponding to a request in the collection by passing the collection as argument to receive_response/3, wait_response/3, or, check_response/3.

The same as calling erpc:reqids_add(erpc:send_request(Node, Module, Function, Args), Label, RequestIdCollection), but calling send_request/6 is slightly more efficient.

Fails with an {erpc, badarg} error exception if:

  • Node is not an atom.

  • Module is not an atom.

  • Function is not an atom.

  • Args is not a list. Note that the list is not verified to be a proper list at the client side.

  • RequestIdCollection is detected not to be request identifier collection.

Note

You cannot make any assumptions about the process that will perform the apply(). It may be a server, or a freshly spawned process.

Types

RequestId = request_id()
Result = term()

The same as calling erpc:wait_response(RequestId, 0). That is, poll for a response message to a call request previously made by the calling process.

Types

RequestId = request_id()
WaitTime = timeout_time()
Result = term()

Wait or poll for a response message to a call request previously made by the calling process using send_request/4. RequestId should be the value returned from the previously made send_request() call, and the corresponding response should not already have been received and handled to completion by check_response/2, receive_response/2, or wait_response().

WaitTime sets an upper time limit on how long to wait for a response. If no response is received before the WaitTime timeout has triggered, the atom no_response is returned. It is valid to continue waiting for a response as many times as needed up until a response has been received and completed by check_response(), receive_response(), or wait_response(). If a response is received, the call operation is completed and either the result is returned as {response, Result} where Result corresponds to the value returned from the applied function or an exception is raised. The exceptions that can be raised corresponds to the same exceptions as can be raised by call/4. That is, no {erpc, timeout} error exception can be raised. wait_response/2 will fail with an {erpc, badarg} exception if/when an invalid RequestId is detected or if an invalid WaitTime is passed.

If the erpc operation fails, but it is unknown if the function is/will be applied (that is, a too large wait time value, or a connection loss), the caller will not receive any further information about the result if/when the applied function completes. If the applied function explicitly communicates with the calling process, such communication may, of course, reach the calling process.

Types

RequestIdCollection = request_id_collection()
WaitTime = timeout_time()
Delete = boolean()
Label = term()
NewRequestIdCollection = request_id_collection()
Result = term()

Wait or poll for a response to a call request corresponding to a request identifier saved in RequestIdCollection. All request identifiers of RequestIdCollection must correspond to requests that have been made using send_request/4 or send_request/6, and all requests must have been made by the process calling this function.

Label is the label associated with the request identifier of the request that the response corresponds to. A request identifier is associated with a label when adding a request identifier in a request identifier collection, or when sending the request using send_request/6.

Compared to wait_response/2, the returned result associated with a specific request identifier or an exception associated with a specific request identifier will be wrapped in a 3-tuple. The first element of this tuple equals the value that would have been produced by wait_response/2, the second element equals the Label associated with the specific request identifier, and the third element NewRequestIdCollection is a possibly modified request identifier collection. The error exception {erpc, badarg} is not associated with any specific request identifier, and will hence not be wrapped.

If RequestIdCollection is empty, no_request will be returned. If no response is received before the WaitTime timeout has triggered, the atom no_response is returned. It is valid to continue waiting for a response as many times as needed up until a response has been received and completed by check_response(), receive_response(), or wait_response(). The difference between receive_response/3 and wait_response/3 is that receive_response/3 abandons requests at timeout so that any potential future responses are ignored, while wait_response/3 does not.

If Delete equals true, the association with Label will have been deleted from RequestIdCollection in the resulting NewRequestIdCollection. If Delete equals false, NewRequestIdCollection will equal RequestIdCollection. Note that deleting an association is not for free and that a collection containing already handled requests can still be used by subsequent calls to wait_response/3, check_response/3, and receive_response/3. However, without deleting handled associations, the above calls will not be able to detect when there are no more outstanding requests to handle, so you will have to keep track of this some other way than relying on a no_request return. Note that if you pass a collection only containing associations of already handled or abandoned requests to wait_response/3, it will always block until a timeout determined by WaitTime is triggered and then return no_response.

Note that a response might have been consumed uppon an {erpc, badarg} exception and if so, will be lost for ever.