Functions for starting and controlling slave nodes.
This module provides functions for starting Erlang slave nodes. All slave nodes that are started by a master terminate automatically when the master terminates. All terminal output produced at the slave is sent back to the master node. File I/O is done through the master.
Slave nodes on other hosts than the current one are started with
the ssh
program. The user must be allowed to ssh
to
the remote hosts without being prompted for a password. This can
be arranged in a number of ways (for details, see the ssh
documentation). 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, see the documentation for the ssh
program.
An alternative to the ssh
program can be specified on
the command line to
erl(1)
as follows:
-rsh Program
Note that the command specified with the -rsh
flag is
treated as a file name which may contain spaces. It is thus not
possible to include any command line options. The remote node will
be launched as "$RSH" "$REMOTE_HOSTNAME" erl -detached -noinput
...
, so the
erl
command must be found in the path on the remote host.
The slave node is to use the same file system at the master. At least, Erlang/OTP is to be installed in the same place on both computers and the same version of Erlang is to be used.
A node running on Windows can only start slave nodes on the host on which it is running.
The master node must be alive.
Functions
Master = node()
ServerList = [atom()]
Calls pseudo(Master, ServerList)
. If you 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 that does nothing but pass on all message to the real server that executes at a master node. A pseudo server is an intermediary that only has the same registered name as the real server.
For example, if you have started a slave node N
and
want to execute pxw
graphics code on this node, you can
start server pxw_server
as a pseudo server at
the slave node. This is illustrated as follows:
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 that executes the function receives
messages. All messages received are simply passed on to
.
start(Host) -> {ok, Node} | {error, Reason}
Host = inet:hostname()
Node = node()
Reason = timeout | no_rsh | {already_running, Node}
start(Host, Name) -> {ok, Node} | {error, Reason}
Host = inet:hostname()
Name = atom() | string()
Node = node()
Reason = timeout | no_rsh | {already_running, Node}
start(Host, Name, Args) -> {ok, Node} | {error, Reason}
Host = inet:hostname()
Name = atom() | string()
Args = string()
Node = node()
Reason = timeout | no_rsh | {already_running, Node}
Starts a slave node on 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 becomes
. If no
name is provided, the name becomes the same as the node that
executes the call (except the host name part of the node name).
The slave node resets its user
process so that all
terminal I/O that is produced at the slave is automatically
relayed to the master. Also, the file process is relayed
to the master.
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 you want to start a slave node at
host H
with node name Name@H
and
want the slave node to have the following properties:
- Directory
Dir
is to be added to the code path. - The Mnesia directory is to be set to
M
. - The Unix
DISPLAY
environment variable is to 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).
The function returns {ok,
, where
is the name of the new node, otherwise
{error,
, where
can be one of:
timeout
The master node failed to get in contact with the slave node. This can occur 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.
no_rsh
No remote shell program was found on the computer. Note
that ssh
is used by default, but this can be overridden
with the -rsh
flag.
{already_running, Node }
A node with name
already exists.
start_link(Host) -> {ok, Node} | {error, Reason}
Host = inet:hostname()
Node = node()
Reason = timeout | no_rsh | {already_running, Node}
start_link(Host, Name) -> {ok, Node} | {error, Reason}
Host = inet:hostname()
Name = atom() | string()
Node = node()
Reason = timeout | no_rsh | {already_running, Node}
start_link(Host, Name, Args) -> {ok, Node} | {error, Reason}
Host = inet:hostname()
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.
For a description of arguments and return values, see
start/1,2,3
.
stop(Node) -> ok
Node = node()
Stops (kills) a node.