Build Status / link to build job

taskeduler

Highly Distributed Application for executing tasks in a concurrent and parallel manner

Overview

This distributed service allows to execute scheduled tasks using a task controller list, provided in the config. Using this module as a dependency of any project enables the main application to define workers and do whatever it needs to accomplish a set of tasks, this is suitable for services that need to sync or process periodically any amount of data, the best example would be obtaining a list of customers and process them using an interval of time. All modules for tasks must implement ´-behavior(task).´, the controller module is used to obtain the list of tasks to execute. If used as a main application, see example functions test_controller.erl for testing, these are the mandatory implementations by the consumer application.

If your main application requires additional endpoints, there is no need to define a new service, all you have to do is to specify the module that will contain the functions exposed as endpoints in the config as {rest_handler, your_module}, all functions exposed with arity of 2 will be public and searchable by the service in that module, so function_name(UrlParam1, Body), will be used if it is requested, if no request Body is received (usually GET), in the request and no url param, these will come as undefined atoms, the endpoint in this case will be your_api/v1/function_name/param1 or your_api/v1/function_name, if the function needs to be async can be called as your_api/v1/async/function_name/param1 or your_api/v1/async/function_name.

Please see project definition documentation

Ownership

taskeduler is owned by the elpaisa. Feel free to email me."

Dependencies

This service depends on Cowboy for defining the http API

How to consume taskeduler

Endpoint	Parameters	Description
start		Starts all workers
start/1	"worker-name", {test_daily, test_regular}	Starts a specific worker
stop		Stops all workers
stop/1	"worker-name", {test_daily, test_regular}	Stops a specific worker
status		Responds the status of all workers
status/1	"worker-name", {test_daily, test_regular}	Responds status for a specific worker
re-sync		Starts all workers and runs an loop for the given interval of days defined in sys.config
re-sync/1	"worker-name", {test_daily, test_regular}	Runs re-sync for a specific worker

How to contribute

Always submit issues for discussion before creating any pull requests.

How to report defects

Open github issues

Running locally

For running locally please use.

git clone [email protected]:elpaisa/taskeduler.git.git
cd taskeduler
make shell

taskeduler requires configuration parameters. Copy the sys.config.tmpl as config/sys.config and edit it accordingly:

cp sys.config.tmpl config/sys.config
vim sys.config

%%-*- mode: erlang -*-
[
  {taskeduler, [
    {max_workers, 12},
    {max_retries, 12},
    {controller_module, test_controller},
    {rest_handler, test_controller},
    {service_name, "/taskeduler/v1"},
    {nodes, ['nonode@nohost']},
    {workers, [
        {test,
            [ {re_sync_days, 7}, {max_fetch, 2000}],
            [ {test_regular, 60, minutes}, {test_daily, 1, days} ] %% Worker definitions
        }
    ]}
  ]},

  {sasl, [
    {sasl_error_logger, false}
  ]}
].
% vim:ft=erlang

Once you have filled out the above changes to the sys.config, you can start taskeduler:

make shell

Or

./rebar3 shell

Testing

To run the tests, execute:

make test

For unit tests and for common (integration) tests, if they are different. Note that the make test target should generally execute whatever testing can be run by continuous integration: thus, if the common tests require some specialized setup/environment that CI is not configured for, this target should not include ct.

Unit Tests

To run the unit tests alone, execute:

make eunit

OR

./rebar3 eunit

If there is something unique about the eunit configuration, or the output of the tests or code coverage are sent somewhere different than conventional, describe here how to view those reports.

Common tests

To run the common tests alone, execute:

make ct

The common tests for taskeduler are integration tests that require all dependencies above to be available. The node started for the common test uses the erl.config file in the project's root directory.

HTTP API documentation

In addition to viewing the API documentation, you can also build it locally for testing updates to documentation prior to deploying it. To build documentation locally, you must install apidoc (see instructions [here][install-apidoc]). Once you have it installed, you can build the documentation:

make doc

To view the documentation after it is built, open the file doc/api/index.html:

% Open in Mac OS X with default browser
open docs/index.html

Doing a release

This command creates a tarball in _build/default/rel/taskeduler/, the full application is self contained in there.

make release

From Jenkins

To create an rpm for installation, execute the below command, that will do test, create a self contained release and build an rpm ready for deployment, later on ./_build/default/rel/taskeduler/bin/taskeduler start will create a service, add this to chkconfig in the server in order to do it a real centos service.

make rpm