I have some apps in my cluster, I need to start some of them sometimes on different hosts.
The story is that the Erlang cluster is already running, so even though I have my .app resource file per application stating which applications should be started before mine, this only works to create a startup script, not to start an app in a already running node.
At the moment I have a custom routine that uses application:get_key(Application,applications) to extract the dependencies and start them separately before starting the given application.
I was wondering if there isn't a better way of doing this.
Since Erlang R16B02, there is also application:ensure_all_started.
Frankly, the standard tools for doing this in Erlang are unnecessarily annoying right now. I tend to put the following boiler-plate in my application callback module:
-module(myapp_app).
-export([start/0]).
start() -> a_start(myapp, permanent).
a_start(App, Type) ->
start_ok(App, Type, application:start(App, Type)).
start_ok(_App, _Type, ok) -> ok;
start_ok(_App, _Type, {error, {already_started, _App}}) -> ok;
start_ok(App, Type, {error, {not_started, Dep}}) ->
ok = a_start(Dep, Type),
a_start(App, Type);
start_ok(App, _Type, {error, Reason}) ->
erlang:error({app_start_failed, App, Reason}).
You can then add -s myapp_app to your erlang command line and this will start the app and all its dependencies recursively. Why this function isn't in the application module I don't know :)
There is a working example of this custom erlang app startup code in my Erlang Factory 2012 SFBay example app.
When starting the app outside of the startup script you do need to start the dependencies first. You could build the smarts to do this into the app itself so that when the app starts it will start any required dependencies before it needs them.
One place I've seen this done is in Mochiweb apps. The default app templates include code for loading dependencies on startup:
-module(some_app).
-export([start/0, stop/0]).
ensure_started(App) ->
case application:start(App) of
ok ->
ok;
{error, {already_started, App}} ->
ok
end.
%% #spec start() -> ok
%% #doc Start the some_app server.
start() ->
some_app_deps:ensure(),
ensure_started(crypto),
application:start(some_app).
%% #spec stop() -> ok
%% #doc Stop the some_app server.
stop() ->
application:stop(some_app).
If you write your app under "OTP Design Principles", you will have to make yourappname.app file, which will contains `applications' section. This section defines which other applications you want to be started before yours. Here is stated:
applications
All applications which must be started before this
application is started. systools uses this list to generate correct
boot scripts. Defaults to [], but note that all applications have
dependencies to at least kernel and stdlib.
So if you use releases, this dependency resolution will be solved by systools.
Related
I have an Erlang application that is a REST service. I have a relx file and when I launch:
rebar3 release
I get an executable
./_build/default/rel/myapp/bin/myapp
When I launch that, the service starts and I can hit my service on localhost:80.
Now I'm trying to write a test suite to test some API calls. In my common test init_per_suite(Config) function, I want to launch my app, something like:
-module(apitest_SUITE).
-include_lib("common_test/include/ct.hrl").
-export([all/0]).
-export([test1/1, init_per_suite/1, end_per_suite/1]).
all() -> [test1].
init_per_suite(Config) ->
%LAUNCH MY APP HERE!!!
%LAUNCHING ../../_build/default/rel/myapp/bin/myapp SEEMS WRONG TO ME
[].
end_per_suite(Config) ->
%KILL MY APP HERE!!!
ok.
test1(Config) ->
httpc:get("localhost:80"). %e.g.
What is the proper way to launch my release from this common_test suite and do this testing?
BTW I am launching my tests using
rebar3 ct
OK, I figured out a way to do this (not sure if it is the best way):
init_per_suite(Config) ->
{ok, _} = application:ensure_all_started(myapp),
[].
%tests that hit localhost:port..
end_per_suite(Config) ->
ok = application:stop(myapp).
I had the same problem. You method can start the application but can't load the sys.config file. If the application depends on some configuration in this file, it will be started fail.
I'm building a system that needs to use a previously built OTP application (lets call it X). If I want to build a new OTP application / module, how can I use the application that already exists from a module, for instance?
I assumed I could call start, since it follows the application behaviour, and so I built a minimalistic application Y that has the following code:
y.erl:
-module(y).
-behaviour(application).
start(_StartType, _StartArgs) ->
io:format("going to call x_app~n"),
{ok, _} = x_app:start([]),
io:format("called x_app~n"),
y:start_link().
stop(_State) ->
ok = x_app:stop([]),
ok.
Rebar compiles this code successfully and generates no warnings.
rel/y/bin/y start outputs nothing at all (I hoped to get the output of at least one io:format)
rel/y/bin/y stop outputs Node is not running!
You need to list application x as a dependent application in your application's .app resource file, or since you're using rebar, in your .app.src file:
{application, your_app,
[{description,"your application"},
{vsn, "0.1"},
{modules,[]},
{registered, []},
{mod,{your_app,[]}},
{env, []},
{applications,[kernel, stdlib, x]}]}.
Note in the very last line that x is listed as an application dependency. This results in the Erlang application controller ensuring that x is started before it starts your application. And if you're starting your application interactively in an Erlang shell via application:ensure_all_started/1,2 this declaration will ensure that x is started first before your app starts.
Currently my Erlang application is started within an escript (TCP server) and all works fine since it uses the default port I provided. Now I want to pass the port via the escript to the application but I have no idea how. (The app runs a supervisor)
script.escript
!/usr/bin/env escript
%% -*- erlang -*-
-export([main/1]).
main([UDPort, TCPort]) ->
U = list_to_integer(UDPort),
T = list_to_integer(TCPort),
app:start(), %% Want to pass T into the startup.
receive
_ -> ok
end;
...
app.erl
-module(app).
-behaviour(application).
-export([start/0, start/2, stop/0, stop/1]).
-define(PORT, 4300).
start () -> application:start(?MODULE). %% This is called by the escript.
stop () -> application:stop(?MODULE).
start (_StartType, _StartArgs) -> supervisor:start(?PORT).
stop (_State) -> ok.
I'm honestly not sure if this is possible with using application but I thought it best to just ask.
The common way is to start things from whatever shell just calling
erl -run foo
But you can also do
erl -appname key value
to set an environment value and then
application:get_env(appname, key)
to get the value you are looking for.
That said...
I like to have service applications be things that don't have to shut down to be (re)configured. I usually include some message protocol like {config, Aspect, Setting} or similar that can alter the basic state of a service on the fly. Because I often do this I usually just wind up having whatever script starts up the application also send a configuration message to it.
So with this in mind, consider this rough conceptual example:
!/usr/bin/env escript
%% -*- erlang -*-
-export([main/1]).
main([UDPort, TCPort]) ->
U = list_to_integer(UDPort),
T = list_to_integer(TCPort),
ok = case whereis(app) of
undefined -> app:start();
_Pid -> ok
end,
ok = set_ports(U, T).
%% Just an illustration.
%% Making this a synchronous gen_server/gen_fsm call is way better.
set_ports(U, T) ->
app ! {config, listen, {tcp, T}},
app ! {config, listen, {udp, U}},
ok.
Now not only is the startup script a startup script, it is also a config script. The point isn't to have a startup script, it is to have a service running on the ports you designated. This isn't a conceptual fit for all tools, of course, but it should give you some ideas. There is also the practice of putting a config file somewhere the application knows to look and just reading terms from it, among other techniques (like including ports in the application specification, etc.).
Edit
I just realized you are doing this in an escript which will spawn a new node every time you call it. To make the technique above work properly you would need to make the escript name a node for the service to run on, and locate it if it already exists.
I want to pass some arguments to supervisor:init/1 function and it is desirable, that the application's interface was so:
redis_pool:start() % start all instances
redis_pool:start(Names) % start only given instances
Here is the application:
-module(redis_pool).
-behaviour(application).
...
start() -> % start without params
application:ensure_started(?APP_NAME, transient).
start(Names) -> % start with some params
% I want to pass Names to supervisor init function
% in order to do that I have to bypass application:ensure_started
% which is not GOOD :(
application:load(?APP_NAME),
case start(normal, [Names]) of
{ok, _Pid} -> ok;
{error, {already_started, _Pid}} -> ok
end.
start(_StartType, StartArgs) ->
redis_pool_sup:start_link(StartArgs).
Here is the supervisor:
init([]) ->
{ok, Config} = get_config(),
Names = proplists:get_keys(Config),
init([Names]);
init([Names]) ->
{ok, Config} = get_config(),
PoolSpecs = lists:map(fun(Name) ->
PoolName = pool_utils:name_for(Name),
{[Host, Port, Db], PoolSize} = proplists:get_value(Name, Config),
PoolArgs = [{name, {local, PoolName}},
{worker_module, eredis},
{size, PoolSize},
{max_overflow, 0}],
poolboy:child_spec(PoolName, PoolArgs, [Host, Port, Db])
end, Names),
{ok, {{one_for_one, 10000, 1}, PoolSpecs}}.
As you can see, current implementation is ugly and may be buggy. The question is how I can pass some arguments and start application and supervisor (with params who were given to start/1) ?
One option is to start application and run redis pools in two separate phases.
redis_pool:start(),
redis_pool:run([] | Names).
But what if I want to run supervisor children (redis pool) when my app starts?
Thank you.
The application callback Module:start/2 is not an API to call in order to start the application. It is called when the application is started by application:start/1,2. This means that overloading it to provide differing parameters is probably the wrong thing to do.
In particular, application:start will be called directly if someone adds your application as a dependency of theirs (in the foo.app file). At this point, they have no control over the parameters, since they come from your .app file, in the {mod, {Mod, Args}} term.
Some possible solutions:
Application Configuration File
Require that the parameters be in the application configuration file; you can retrieve them with application:get_env/2,3.
Don't start a supervisor
This means one of two things: becoming a library application (removing the {mod, Mod} term from your .app file) -- you don't need an application behaviour; or starting a dummy supervisor that does nothing.
Then, when someone wants to use your library, they can call an API to create the pool supervisor, and graft it into their supervision tree. This is what poolboy does with poolboy:child_spec.
Or, your application-level supervisor can be a normal supervisor, with no children by default, and you can provide an API to start children of that, via supervisor:start_child. This is (more or less) what cowboy does.
You can pass arguments in the AppDescr argument to application:load/1 (though its a mighty big tuple already...) as {mod, {Module, StartArgs}} according to the docs ("according to the docs" as in, I don't recall doing it this way myself, ever: http://www.erlang.org/doc/apps/kernel/application.html#load-1).
application:load({application, some_app, {mod, {Module, [Stuff]}}})
Without knowing anything about the internals of the application you're starting, its hard to say which way is best, but a common way to do this is to start up the application and then send it a message containing the data you want it to know.
You could make receipt of the message form tell the application to go through a configuration assertion procedure, so that the same message you send on startup is also the same sort of thing you would send it to reconfigure it on the fly. I find this more useful than one-shotting arguments on startup.
In any case, it is usually better to think in terms of starting something, then asking it to do something for you, than to try telling it everything in init parameters. This can be as simple as having it start up and wait for some message that will tell the listener to then spin up the supervisor the way you're trying to here -- isolated one step from the application inclusion issues RL mentioned in his answer.
I have an Erlang application based on Cowboy and I would like to test it.
Previously I used wooga's library etest_http for this kind of tasks, but I would like to start using common tests since I noticed that this is the way used in cowboy. I have tried to setup a very basic test but I am not able to run it properly.
Can anybody provide me a sample for testing the basic example echo_get and tell me what is the correct way to run the test from the console using the Makefile contained in the example?
The example make used only for build echo_get app. So to test echo_get app you can add test suite and call make && rebar -v 1 skip_deps=true ct (rebar should be in PATH) from shell.
Also you need etest and etest_http in your Erlang PATH or add it with rebar.config in your application. You can use httpc or curl with os:cmd/1 instead ehttp_test :)
test/my_test_SUITE.erl (full example)
-module(my_test_SUITE).
-compile(export_all).
-include_lib("common_test/include/ct.hrl").
% etest macros
-include_lib ("etest/include/etest.hrl").
% etest_http macros
-include_lib ("etest_http/include/etest_http.hrl").
suite() ->
[{timetrap,{seconds,30}}].
init_per_suite(Config) ->
%% start your app or release here
%% for example start echo_get release
os:cmd("./_rel/bin/echo_get_example start"),
Config.
end_per_suite(_Config) ->
%% stop your app or release here
%% for example stop echo_get release
os:cmd("./_rel/bin/echo_get_example stop")
ok.
init_per_testcase(_TestCase, Config) ->
Config.
end_per_testcase(_TestCase, _Config) ->
ok.
all() ->
[my_test_case].
my_test_case(_Config) ->
Response = ?perform_get("http://localhost:8080/?echo=saymyname"),
?assert_status(200, Response),
?assert_body("saymyname", Response).
ok.
The following starts the hello_world application and its dependencies, but uses the most recently compiled versions rather than the one in ./_rel; that may or may not be what you want, but it does avoid the timer:sleep(1000)
-module(hello_world_SUITE).
-include_lib("common_test/include/ct.hrl").
-export([all/0, init_per_suite/1, end_per_suite/1]).
-export([http_get_hello_world/1]).
all() ->
[http_get_hello_world].
init_per_suite(Config) ->
{ok, App_Start_List} = start([hello_world]),
inets:start(),
[{app_start_list, App_Start_List}|Config].
end_per_suite(Config) ->
inets:stop(),
stop(?config(app_start_list, Config)),
Config.
http_get_hello_world(_Config) ->
{ok, {{_Version, 200, _ReasonPhrase}, _Headers, Body}} =
httpc:request(get, {"http://localhost:8080", []}, [], []),
Body = "Hello World!\n".
start(Apps) ->
{ok, do_start(_To_start = Apps, _Started = [])}.
do_start([], Started) ->
Started;
do_start([App|Apps], Started) ->
case application:start(App) of
ok ->
do_start(Apps, [App|Started]);
{error, {not_started, Dep}} ->
do_start([Dep|[App|Apps]], Started)
end.
stop(Apps) ->
_ = [ application:stop(App) || App <- Apps ],
ok.
Use https://github.com/extend/gun
to run the provided example (considering 'gun' in in 'user' folder):
ct_run -suite twitter_SUITE.erl -logdir ./results -pa /home/user/gun/deps/*/ebin /home/user/gun/ebin