View Source erl_call
Call/start a distributed Erlang node.
Description
erl_call makes it possible to start and/or communicate with a distributed
Erlang node. It is built upon the Erl_Interface library as an example
application. Its purpose is to use a Unix shell script to interact with a
distributed Erlang node. It performs all communication with the Erlang rex
server, using the standard Erlang RPC facility. It does not require any special
software to be run at the Erlang target node.
The main use is to either start a distributed Erlang node or to make an ordinary
function call. However, it is also possible to pipe an Erlang module to
erl_call and have it compiled, or to pipe a sequence of Erlang expressions to
be evaluated (similar to the Erlang shell).
Options, which cause stdin to be read, can be used with advantage, as scripts
from within (Unix) shell scripts. Another nice use of erl_call could be from
(HTTP) CGI-bin scripts.
erl_call <options>
Starts/calls Erlang.
Each option flag is described below with its name, type, and meaning.
-a [Mod [Fun [Args]]]]- (Optional.) Applies the specified function and returns the result.Modmust be specified. However,startand[]are assumed for unspecifiedFunandArgs, respectively.Argsis to be in the same format as forerlang:apply/3inERTSexcept only a subset of all terms are allowed. The allowed term types are:list(andstringrepresentation of list, that is "example"),tuple,atomandnumber.Notice that this flag takes exactly one argument, so quoting can be necessary to group
Mod,Fun, andArgsin a manner dependent on the behavior of your command shell.-address [Hostname:]Port- (One of-n,-name,-snameor-addressis required.)Hostnameis the hostname of the machine that is running the peer node thaterl_callshall communicate with. The default hostname is the hostname of the local machine.Portis the port number of the node thaterl_callshall communicate with. The-addressflag cannot be combined with any of the flags-n,-name,-snameor-s.The
-addressflag is typically useful when one wants to call a node that is running on machine without an accessible epmd instance.-c Cookie- (Optional.) Use this option to specify a certain cookie. If no cookie is specified, the~/.erlang.cookiefile is read and its content is used as cookie. The Erlang node we want to communicate with must have the same cookie.-d- (Optional.) Debug mode. This causes all I/O to be output to the~/.erl_call.out.Nodenamefile, whereNodenameis the node name of the Erlang node in question.-e- (Optional.) Reads a sequence of Erlang expressions, separated by comma (,) and ended with a full stop (.), fromstdinuntil EOF (Control-D). Evaluates the expressions and returns the result from the last expression. Returns{ok,Result}on success.-fetch_stdout- (Optional.) Executes the code, specified with the-aor-eoption, in a new process that has a group leader that forwards all stdout (standard output) data so that it is printed to stdout of theerl_callprocess. This means that stdout data that are written during the execution of the called code, by the code and by descendant processes, will be forwarded (given that the group leader has not been changed by a call toerlang:group_leader/2).The printed data is UTF-8 encoded.
This option is only relevant together with the option
-aor-e.See the documentation of the I/O protocol, for more information about the group leader concept.
Note
This option only works when
erl_callis interacting with a node with a version greater or equal to OTP-24.-h HiddenName- (Optional.) Specifies the name of the hidden node thaterl_callrepresents.-m- (Optional.) Reads an Erlang module fromstdinand compiles it.-n Node- (One of-n,-name,-snameor-addressis required.) Has the same meaning as-nameand can still be used for backward compatibility reasons.-name Node- (One of-n,-name,-snameor-addressis required.)Nodeis the name of the peer node to be started or communicated with. It is assumed thatNodeis started witherl -name, which means that fully qualified long node names are used. If option-sis specified, an Erlang node will (if necessary) be started witherl -name.-no_result_term- (Optional.) Do not print the result term. This option is only relevant together with the options-aand-e.-q- (Optional.) Halts the Erlang node specified with switch-n. This switch overrides switch-s.-r- (Optional.) Generates a random name of the hidden node thaterl_callrepresents.-R- (Optional.) Request a dynamic random name, of the hidden node thaterl_callrepresents, from the peer node. Supported since OTP 23. Prefer-Rover-rwhen doing repeated requests toward the same peer node.-s- (Optional.) Starts a distributed Erlang node if necessary. This means that in a sequence of calls, where '-s' and '-n Node' are constant, only the first call starts the Erlang node. This makes the rest of the communication very fast. This flag is currently only available on Unix-like platforms (Linux, Mac OS X, Solaris, and so on).-sname Node- (One of-n,-name,-snameor-addressis required.)Nodeis the name of the peer node to be started or communicated with. It is assumed thatNodeis started witherl -sname, which means that short node names are used. If option-sis specified, an Erlang node is started (if necessary) witherl -sname.-timeout Seconds- (Optional.) Aborts theerl_callprocess after the timeout expires. Note that this does not abort commands that have already been started with-a,-e, or similar.-v- (Optional.) Prints a lot ofverboseinformation. This is only useful for the developer and maintainer oferl_call.-x ErlScript- (Optional.) Specifies another name of the Erlang startup script to be used. If not specified, the standarderlstartup script is used.
Examples
To start an Erlang node and call erlang:time/0:
erl_call -s -a 'erlang time' -n madonna
{18,27,34}To terminate an Erlang node by calling erlang:halt/0:
erl_call -s -a 'erlang halt' -n madonnaTo apply with many arguments:
erl_call -s -a 'lists seq [1,10]' -n madonnaTo evaluate some expressions (the input ends with EOF (Control-D)):
erl_call -s -e -n madonna
statistics(runtime),
X=1,
Y=2,
{_,T}=statistics(runtime),
{X+Y,T}.
^D
{ok,{3,0}}To compile a module and run it (again, the input ends with EOF (Control-D)):
(In the example, the output has been formatted afterwards.)
erl_call -s -m -a procnames -n madonna
-module(procnames).
-compile(export_all).
start() ->
P = processes(),
F = fun(X) -> {X,process_info(X,registered_name)} end,
lists:map(F,[],P).
^D
[{<madonna@chivas.du.etx.ericsson.se,0,0>,
{registered_name,init}},
{<madonna@chivas.du.etx.ericsson.se,2,0>,
{registered_name,erl_prim_loader}},
{<madonna@chivas.du.etx.ericsson.se,4,0>,
{registered_name,error_logger}},
{<madonna@chivas.du.etx.ericsson.se,5,0>,
{registered_name,application_controller}},
{<madonna@chivas.du.etx.ericsson.se,6,0>,
{registered_name,kernel}},
{<madonna@chivas.du.etx.ericsson.se,7,0>,
[]},
{<madonna@chivas.du.etx.ericsson.se,8,0>,
{registered_name,kernel_sup}},
{<madonna@chivas.du.etx.ericsson.se,9,0>,
{registered_name,net_sup}},
{<madonna@chivas.du.etx.ericsson.se,10,0>,
{registered_name,net_kernel}},
{<madonna@chivas.du.etx.ericsson.se,11,0>,
[]},
{<madonna@chivas.du.etx.ericsson.se,12,0>,
{registered_name,global_name_server}},
{<madonna@chivas.du.etx.ericsson.se,13,0>,
{registered_name,auth}},
{<madonna@chivas.du.etx.ericsson.se,14,0>,
{registered_name,rex}},
{<madonna@chivas.du.etx.ericsson.se,15,0>,
[]},
{<madonna@chivas.du.etx.ericsson.se,16,0>,
{registered_name,file_server}},
{<madonna@chivas.du.etx.ericsson.se,17,0>,
{registered_name,code_server}},
{<madonna@chivas.du.etx.ericsson.se,20,0>,
{registered_name,user}},
{<madonna@chivas.du.etx.ericsson.se,38,0>,
[]}]To forward standard output without printing the result term (again, the input ends with EOF (Control-D)):
erl_call -s -e -sname madonna -fetch_stdout -no_result_term
io:format("Number of schedulers: ~p~n", [erlang:system_info(schedulers)]),
io:format("Number of logical cores: ~p~n", [erlang:system_info(logical_processors_available)]).
^D
Number of schedulers: 8
Number of logical cores: 8