Erlang logo
User's Guide
Reference Manual
Release Notes

Reference Manual
Version 2.8

Expand All
Contract All

Table of Contents





Functions to Starting and Controlling Slave Nodes


This module provides functions for starting Erlang slave nodes. All slave nodes which are started by a master will terminate automatically when the master terminates. All TTY output produced at the slave will be sent back to the master node. File I/O is done via the master.

Slave nodes on other hosts than the current one are started with the program rsh. The user must be allowed to rsh to the remote hosts without being prompted for a password. This can be arranged in a number of ways (refer to the rsh documentation for details). A slave node started on the same host as the master inherits certain environment values from the master, such as the current directory and the environment variables. For what can be assumed about the environment when a slave is started on another host, read the documentation for the rsh program.

An alternative to the rsh program can be specified on the command line to erl as follows: -rsh Program.

The slave node should use the same file system at the master. At least, Erlang/OTP should be installed in the same place on both computers and the same version of Erlang should be used.

Currently, a node running on Windows NT can only start slave nodes on the host on which it is running.

The master node must be alive.


start(Host) -> {ok, Node} | {error, Reason}
start(Host, Name) -> {ok, Node} | {error, Reason}
start(Host, Name, Args) -> {ok, Node} | {error, Reason}


Name = atom() | string()
Args = string()
Node = node()
Reason = timeout | no_rsh | {already_running, Node}

Starts a slave node on the host Host. Host names need not necessarily be specified as fully qualified names; short names can also be used. This is the same condition that applies to names of distributed Erlang nodes.

The name of the started node will be Name@Host. If no name is provided, the name will be the same as the node which executes the call (with the exception of the host name part of the node name).

The slave node resets its user process so that all terminal I/O which is produced at the slave is automatically relayed to the master. Also, the file process will be relayed to the master.

The Args argument is used to set erl command line arguments. If provided, it is passed to the new node and can be used for a variety of purposes. See erl(1)

As an example, suppose that we want to start a slave node at host H with the node name Name@H, and we also want the slave node to have the following properties:

  • directory Dir should be added to the code path;

  • the Mnesia directory should be set to M;

  • the unix DISPLAY environment variable should be set to the display of the master node.

The following code is executed to achieve this:

E = " -env DISPLAY " ++ net_adm:localhost() ++ ":0 ",
Arg = "-mnesia_dir " ++ M ++ " -pa " ++ Dir ++ E,
slave:start(H, Name, Arg).

If successful, the function returns {ok, Node}, where Node is the name of the new node. Otherwise it returns {error, Reason}, where Reason can be one of:


The master node failed to get in contact with the slave node. This can happen in a number of circumstances:

  • Erlang/OTP is not installed on the remote host
  • the file system on the other host has a different structure to the the master
  • the Erlang nodes have different cookies.

There is no rsh program on the computer.

{already_running, Node}

A node with the name Name@Host already exists.

start_link(Host) -> {ok, Node} | {error, Reason}
start_link(Host, Name) -> {ok, Node} | {error, Reason}
start_link(Host, Name, Args) -> {ok, Node} | {error, Reason}


Name = atom() | string()
Args = string()
Node = node()
Reason = timeout | no_rsh | {already_running, Node}

Starts a slave node in the same way as start/1,2,3, except that the slave node is linked to the currently executing process. If that process terminates, the slave node also terminates.

See start/1,2,3 for a description of arguments and return values.

stop(Node) -> ok


Node = node()

Stops (kills) a node.

pseudo([Master | ServerList]) -> ok


Master = node()
ServerList = [atom()]

Calls pseudo(Master, ServerList). If we want to start a node from the command line and set up a number of pseudo servers, an Erlang runtime system can be started as follows:

% erl -name abc -s slave pseudo klacke@super x --

pseudo(Master, ServerList) -> ok


Master = node()
ServerList = [atom()]

Starts a number of pseudo servers. A pseudo server is a server with a registered name which does absolutely nothing but pass on all message to the real server which executes at a master node. A pseudo server is an intermediary which only has the same registered name as the real server.

For example, if we have started a slave node N and want to execute pxw graphics code on this node, we can start the server pxw_server as a pseudo server at the slave node. The following code illustrates:

rpc:call(N, slave, pseudo, [node(), [pxw_server]]).

relay(Pid) -> no_return()


Pid = pid()

Runs a pseudo server. This function never returns any value and the process which executes the function will receive messages. All messages received will simply be passed on to Pid.