netlogix/jobqueue-scheduled

Allows for scheduled jobs and deduplicates

1.2.4 2023-06-26 13:53 UTC

Requires

• php: ^7.4 || ^8.0 || ^8.1
• doctrine/dbal: ^2.9
• doctrine/orm: ^2.6
• flowpack/jobqueue-common: ^3.0
• neos/flow: ^5.3 || ^6.3 || ^7.3 || ^8.0

Suggests

• netlogix/supervisor: Create supervisord programs that run the 'scheduler:pollforincomingjobs' command

MIT

README

This package provides a PDO based scheduler for JobQueue jobs.

There are two main goals:

1. Schedule a Flowpack JobQueue job for later execution.
2. Have a single job only scheduled once.

Installation

composer require netlogix/jobqueue-scheduled

Schedule jobs

The first goal of this package is to have a way to put any kind of JobQueue job on hold and mark it for later execution. In contrast to e.g. "Defer" annotated methods or default behavior of JobQueue jobs, there is a time aspect to it. Scheduled jobs are not executed immediately but at a time specified while scheduling.

Regular job queue jobs need to be serializable. That's just an implementation detail of how Flowpack FakeQueue and t3n RabbitQueue work.

To schedule an existing Flowpack job, just wrap it in a Scheduled Job and pass it to the scheduler.

use Netlogix\JobQueue\Scheduled\Domain\Model\ScheduledJob;
use Netlogix\JobQueue\Scheduled\Domain\Scheduler;
use Flowpack\JobQueue\Common\Job\JobInterface;

assert($scheduler instanceof Scheduler);
assert($jobqueueJob instanceof JobInterface);

$dueDate = new \DateTimeImmutable('now + 1 minute');
$queueName = 'default';

$schedulerJob = new ScheduledJob(
    $jobqueueJob,
    $queueName,
    $dueDate
);
$scheduler->schedule($schedulerJob);

Jobs with unique identifiers

The second goal of this package is to avoid duplicate schedules.

Certain jobs don't calculate a specific computation but just "run to the end". An example of those are the catchup jobs of the Neos.EventSourcing package. Queuing triggers the catchup call of some event listeners.

Queuing another catchup job while the first one is running is good because there might be additonal changes.

Queueing another catchup job while the previous one has is not even started is unnecessary because the first one will already catch up to the end.

use Netlogix\JobQueue\Scheduled\Domain\Model\ScheduledJob;
use Netlogix\JobQueue\Scheduled\Domain\Scheduler;
use Flowpack\JobQueue\Common\Job\JobInterface;

assert($scheduler instanceof Scheduler);
assert($jobqueueJob instanceof JobInterface);

$dueDate = new \DateTimeImmutable('now + 1 minute');
$queueName = 'default';

$jobIdentifier = 'event-sourcing-catchup-' . $eventListenerName;

$schedulerJob = new ScheduledJob(
    $jobqueueJob,
    $queueName,
    $dueDate,
    $jobIdentifier
);
$scheduler->schedule($schedulerJob);

The "schedule" will only schedule a new job if the specified identifier is not already scheduled. If there are conflicts between the existing due date and the one provided by the new job the earliest value is taken.

Queue scheduled jobs

A scheduled job lives in the database and is not processed any further until queueing happens.

This is currently done via cronjobs.

* * * * *   ./flow scheduler:queueduejobs

The internal scheduling mechanism will make sure only those jobs are passed from the scheduler to the job queue which are "due" according to their individual due date values.