Download the PHP package instasent/resque-bundle without Composer

Intro to InstasentResqueBundle

This is a fork from https://github.com/michelsalib/InstasentResqueBundle in order to keep its maintenance.

The Instasent resque bundle provides integration of php-resque to Symfony. It is inspired from resque, a Redis-backed Ruby library for creating background jobs, placing them on multiple queues, and processing them later.

Features:

• Creating a Job, with container access in order to leverage your Symfony services
• Enqueue a Job with parameters on a given queue
• Creating background worker on a given queue
• A UX to monitor your queues, workers and job statuses
• ability to schedule jobs to run at a specific time or after a number of seconds delay
• ability to auto re-queue failed jobs, with back-off strategies

TODOs:

• Log management
• Job status tracking
• Redis configuration
• Localisation
• Tests

Screenshots

Dashboard

Installation and configuration:

Requirements

Make sure you have redis installed on your machine: http://redis.io/

Get the bundle

Add to your instasent-resque-bundle to your dependencies:

To install, run php composer.phar [update|install].

Add InstasentResqueBundle to your application kernel

Import the routing configuration

Add to your routing.yml:

You can customize the prefix as you wish.

You can now access the dashboard at this url: /resque

To secure the dashboard, you can add the following to your security.yml - provided your administrator role is ROLE_ADMIN

Optional, secure the dashboard behind a role

Add to your security.yml:

Now only users with the role ROLE_ADMIN will be able to access the dashboard at this url: /resque

Optional, set configuration

You may want to add some configuration to your config.yml

See the Auto retry section for more on how to use auto_retry.

Set worker: root_dir: in case job fails to run when worker systems are hosted on separate server/dir from the system creating the queue. When running multiple configured apps for multiple workers, all apps must be able to access by the same root_dir defined in worker: root_dir.

Optional, configure lazy loading

This bundle is prepared for lazy loading in order to make a connection to redis only when its really used. Symfony2 supports that starting with 2.3. To make it work an additional step needs to be done. You need to install a proxy manager to your Symfony2 project. The full documentation for adding the proxy manager can be found in Symfony2's Lazy Service documentation.

Creating a Job

A job is a subclass of the Instasent\ResqueBundle\Job class. You also can use the Instasent\Resque\ContainerAwareJob if you need to leverage the container during job execution. You will be forced to implement the run method that will contain your job logic:

As you can see you get an $args parameter that is the array of arguments of your Job.

Adding a job to a queue

You can get the resque service simply by using the container. From your controller you can do:

Running a worker on a queue

Executing the following commands will create a work on :

• the default queue : app/console instasent:resque:worker-start default
• the q1 and q2 queue : app/console instasent:resque:worker-start q1,q2 (separate name with ,)
• all existing queues : app/console instasent:resque:worker-start "*"

You can also run a worker foreground by adding the --foreground option;

By default VERBOSE environment variable is set when calling php-resque

• --verbose option sets VVERBOSE
• --quiet disables both so no debug output is thrown

See php-resque logging option : https://github.com/chrisboulton/php-resque#logging

Use monolog from your kernel

You can pass a monolog channel to write your logs, see example in supervisor config. Note: Pass Kernel ENV to resque is to instanciate logger, no for sharing kernel between jobs.

Running a worker without forking

Sometimes intensive tasks may lose performance with forking mode. We have added a bin/resque-single binary, you can pass kernel file to be loaded only once at start. Be carefull cause if this Worker fails, then you need to restart manually this worker. We recommend to use supervisor to control this, and superlance with memory plugin to restart worker when hits some memory treshold. Be careful too with your code, shared variables, services instances... etc.

Adding a delayed job to a queue

You can specify that a job is run at a specific time or after a specific delay (in seconds).

From your controller you can do:

You must also run a scheduledworker, which is responsible for taking items out of the special delayed queue and putting them into the originally specified queue.

app/console instasent:resque:scheduledworker-start

Stop it later with app/console instasent:resque:scheduledworker-stop.

Note that when run in background mode it creates a PID file in 'cache//instasent_resque_scheduledworker.pid'. If you clear your cache while the scheduledworker is running you won't be able to stop it with the scheduledworker-stop command.

Alternatively, you can run the scheduledworker in the foreground with the --foreground option.

Note also you should only ever have one scheduledworker running, and if the PID file already exists you will have to use the --force option to start a scheduledworker.

Manage production workers with supervisord

It's probably best to use supervisord (http://supervisord.org) to run the workers in production, rather than re-invent job spawning, monitoring, stopping and restarting.

Here's a sample conf file

(If you use a custom Resque prefix, add an extra environment variable: PREFIX='my-resque-prefix')

Then in Capifony you can do

sudo supervisorctl stop myapp:* before deploying your app and sudo supervisorctl start myapp:* afterwards.

More features

Changing the queue

You can change a job queue just by setting the queue field of the job:

From within the job:

Or from outside the job:

Access the container from inside your job

Just extend the ContainerAwareJob:

Stop a worker

Use the app/console instasent:resque:worker-stop command.

• No argument will display running workers that you can stop.
• Add a worker id to stop it: app/console instasent:resque:worker-stop ubuntu:3949:default
• Add the --all option to stop all the workers.

Auto retry

You can have the bundle auto retry failed jobs by adding retry strategy for either a specific job, or for all jobs in general:

The following will allow Some\Job to retry 3 times.

• right away
• after a 10 second delay
• after a 60 second delay

Setting strategy for all jobs:

With default strategy for all but specific jobs:

The default strategy (if provided) will be applied to all jobs that does not have a specific strategy attached. If not provided these jobs will not have auto retry.

You can disable auto_retry for selected jobs by using an empty array:

Here Some\Job will not have any auto_retry attached.

Please note

To use the auto_retry feature, you must also run the scheduler job:

app/console instasent:resque:scheduledworker-start