Queues
- Configuration
- Basic Usage
- More Dispatch Methods
- Queueing Closures
- Running The Queue Listener
- Daemon Queue Worker
- Failed Jobs
Configuration
Lumen utilizes Laravel's queue component to provide a unified API across a variety of different queue services. Queues allow you to defer the processing of a time consuming task, such as sending an e-mail, until a later time, thus drastically speeding up the web requests to your application.
Lumen and Laravel provide support for database, Beanstalkd, IronMQ, Amazon SQS, Redis, null, and synchronous (for local use) queue drivers. The null
queue driver simply discards queued jobs so they are never run.
The QUEUE_DRIVER
option in your .env
file determines the queue "driver" that will be used by your application.
Queue Database Table
In order to use the database
queue driver, you will need a database table to hold the jobs. The table schema should look like the following:
Schema::create('jobs', function(Blueprint $table)
{
$table->bigIncrements('id');
$table->string('queue');
$table->text('payload');
$table->tinyInteger('attempts')->unsigned();
$table->tinyInteger('reserved')->unsigned();
$table->unsignedInteger('reserved_at')->nullable();
$table->unsignedInteger('available_at');
$table->unsignedInteger('created_at');
});
Other Queue Dependencies
The following dependencies are needed for the listed queue drivers:
- Amazon SQS:
aws/aws-sdk-php
- Beanstalkd:
pda/pheanstalk ~3.0
- IronMQ:
iron-io/iron_mq ~1.5
- Redis:
predis/predis ~1.0
Basic Usage
Pushing A Job Onto The Queue
All of the queueable jobs for your application are stored in the App\Jobs
directory. The base App\Job
class may serve as a base class for the rest of your jobs.
Note: If you intend to use the
Queue
facade, be sure to uncomment the$app->withFacades()
call in yourbootstrap/app.php
file.
To push a new job onto the queue, use the Queue::push
method:
Queue::push(new SendEmail($message));
You may also use the dispatch
method from a Closure route or a controller:
$this->dispatch(new SendEmail($message));
The job's handle
method will be called when the job is executed by the queue. You may type-hint any dependencies you need on the handle
method and the service container will automatically inject them:
public function handle(UserRepository $users)
{
//
}
Specifying The Queue / Tube For A Job
You may also specify the queue / tube a job should be sent to:
Queue::pushOn('emails', new SendEmail($message));
Passing The Same Payload To Multiple Jobs
If you need to pass the same data to several queue jobs, you may use the Queue::bulk
method:
Queue::bulk([new SendEmail($message), new AnotherCommand]);
Delaying The Execution Of A Job
Sometimes you may wish to delay the execution of a queued job. For instance, you may wish to queue a job that sends a customer an e-mail 15 minutes after sign-up. You can accomplish this using the Queue::later
method:
$date = Carbon::now()->addMinutes(15);
Queue::later($date, new SendEmail($message));
In this example, we're using the Carbon date library to specify the delay we wish to assign to the job. Alternatively, you may pass the number of seconds you wish to delay as an integer.
Note: The Amazon SQS service has a delay limit of 900 seconds (15 minutes).
Queues And Eloquent Models
If your queued job accepts an Eloquent model in its constructor, only the identifier for the model will be serialized onto the queue. When the job is actually handled, the queue system will automatically re-retrieve the full model instance from the database. It's all totally transparent to your application and prevents issues that can arise from serializing full Eloquent model instances.
Deleting A Processed Job
Once you have processed a job, it must be deleted from the queue. If no exception is thrown during the execution of your job, this will be done automatically.
If you would like to delete
or release
the job manually, the Illuminate\Queue\InteractsWithQueue
trait provides access to the queue job release
and delete
methods. The release
method accepts a single value: the number of seconds you wish to wait until the job is made available again.
public function handle(SendEmail $job)
{
if (true) {
$this->release(30);
}
}
Releasing A Job Back Onto The Queue
IF an exception is thrown while the job is being processed, it will automatically be released back onto the queue so it may be attempted again. The job will continue to be released until it has been attempted the maximum number of times allowed by your application. The number of maximum attempts is defined by the --tries
switch used on the queue:listen
or queue:work
Artisan jobs.
Checking The Number Of Run Attempts
If an exception occurs while the job is being processed, it will automatically be released back onto the queue. You may check the number of attempts that have been made to run the job using the attempts
method:
if ($this->attempts() > 3) {
//
}
Note: Your job / handler must use the
Illuminate\Queue\InteractsWithQueue
trait in order to call this method.
More Dispatch Methods
Mapping Command Properties From Requests
It is very common to map HTTP request variables into jobs. So, instead of forcing you to do this manually for each request, Lumen provides some helper methods to make it a cinch. Let's take a look at the dispatchFrom
method available from Closure routes and controller methods:
$this->dispatchFrom('Command\Class\Name', $request);
This method will examine the constructor of the job class it is given, and then extract variables from the HTTP request (or any other ArrayAccess
object) to fill the needed constructor parameters of the job. So, if our job class accepts a firstName
variable in its constructor, the job bus will attempt to pull the firstName
parameter from the HTTP request.
You may also pass an array as the third argument to the dispatchFrom
method. This array will be used to fill any constructor parameters that are not available on the request:
$this->dispatchFrom('Command\Class\Name', $request, [
'firstName' => 'Taylor',
]);
Queueing Closures
Note: Before queueing Closures, you will need to add the
jeremeamia/superclosure
(~2.0) dependency to yourcomposer.json
file.
You may also push a Closure onto the queue. This is very convenient for quick, simple tasks that need to be queued:
Pushing A Closure Onto The Queue
Queue::push(function($job) use ($id) {
Account::delete($id);
$job->delete();
});
Note: Instead of making objects available to queued Closures via the
use
directive, consider passing primary keys and re-pulling the associated models from within your queue job. This often avoids unexpected serialization behavior.
When using Iron.io push queues, you should take extra precaution queueing Closures. The end-point that receives your queue messages should check for a token to verify that the request is actually from Iron.io. For example, your push queue end-point should be something like: https://yourapp.com/queue/receive?token=SecretToken
. You may then check the value of the secret token in your application before marshalling the queue request.
Running The Queue Listener
Lumen, like Laravel, includes an Artisan task that will run new jobs as they are pushed onto the queue. You may run this task using the queue:listen
job:
Starting The Queue Listener
php artisan queue:listen
You may also specify which queue connection the listener should utilize:
php artisan queue:listen connection
Note that once this task has started, it will continue to run until it is manually stopped. You may use a process monitor such as Supervisor to ensure that the queue listener does not stop running.
You may pass a comma-delimited list of queue connections to the listen
job to set queue priorities:
php artisan queue:listen --queue=high,low
In this example, jobs on the high-connection
will always be processed before moving onto jobs from the low-connection
.
Specifying The Job Timeout Parameter
You may also set the length of time (in seconds) each job should be allowed to run:
php artisan queue:listen --timeout=60
Specifying Queue Sleep Duration
In addition, you may specify the number of seconds to wait before polling for new jobs:
php artisan queue:listen --sleep=5
Note that the queue only "sleeps" if no jobs are on the queue. If more jobs are available, the queue will continue to work them without sleeping.
Processing The First Job On The Queue
To process only the first job on the queue, you may use the queue:work
job:
php artisan queue:work
Daemon Queue Worker
The queue:work
also includes a --daemon
option for forcing the queue worker to continue processing jobs without ever re-booting the framework. This results in a significant reduction of CPU usage when compared to the queue:listen
job.
To start a queue worker in daemon mode, use the --daemon
flag:
php artisan queue:work connection --daemon
php artisan queue:work connection --daemon --sleep=3
php artisan queue:work connection --daemon --sleep=3 --tries=3
As you can see, the queue:work
job supports most of the same options available to queue:listen
. You may use the php artisan help queue:work
job to view all of the available options.
Deploying With Daemon Queue Workers
The simplest way to deploy an application using daemon queue workers is to put the application in maintenance mode at the beginning of your deployment. This can be done using the php artisan down
job. Once the application is in maintenance mode, Lumen and Laravel will not accept any new jobs off of the queue, but will continue to process existing jobs.
The easiest way to restart your workers is to include the following job in your deployment script:
php artisan queue:restart
This job will instruct all queue workers to restart after they finish processing their current job.
Note: This job relies on the cache system to schedule the restart. By default, APCu does not work for CLI jobs. If you are using APCu, add
apc.enable_cli=1
to your APCu configuration.
Coding For Daemon Queue Workers
Daemon queue workers do not restart the framework before processing each job. Therefore, you should be careful to free any heavy resources before your job finishes. For example, if you are doing image manipulation with the GD library, you should free the memory with imagedestroy
when you are done.
Similarly, your database connection may disconnect when being used by long-running daemon. You may use the DB::reconnect
method to ensure you have a fresh connection.
Failed Jobs
Since things don't always go as planned, sometimes your queued jobs will fail. Don't worry, it happens to the best of us! Lumen and Laravel include a convenient way to specify the maximum number of times a job should be attempted. After a job has exceeded this amount of attempts, it will be inserted into a failed_jobs
table.
The failed_jobs
table should have a schema like the following:
Schema::create('failed_jobs', function(Blueprint $table)
{
$table->increments('id');
$table->text('connection');
$table->text('queue');
$table->text('payload');
$table->timestamp('failed_at');
});
You can specify the maximum number of times a job should be attempted using the --tries
switch on the queue:listen
job:
php artisan queue:listen connection-name --tries=3
If you would like to register an event that will be called when a queue job fails, you may use the Queue::failing
method. This event is a great opportunity to notify your team via e-mail or HipChat.
Queue::failing(function($connection, $job, $data) {
//
});
You may also define a failed
method directly on a queue job class, allowing you to perform job specific actions when a failure occurs:
public function failed()
{
// Called when the job is failing...
}
Retrying Failed Jobs
To view all of your failed jobs, you may use the queue:failed
Artisan job:
php artisan queue:failed
The queue:failed
job will list the job ID, connection, queue, and failure time. The job ID may be used to retry the failed job. For instance, to retry a failed job that has an ID of 5, the following job should be issued:
php artisan queue:retry 5
If you would like to delete a failed job, you may use the queue:forget
job:
php artisan queue:forget 5
To delete all of your failed jobs, you may use the queue:flush
job:
php artisan queue:flush