Resque Bundle Laravel Package

Getting Started

1. Install and check constraints first. composer require mcfedr/resque-bundle — before you commit, open the bundle's composer.json and confirm its Symfony constraint matches your app. This is a low-traffic package (9 stars, no dependents), so assume it targets an older Symfony (2.x/3.x/early 4.x) until proven otherwise.

2. Register the bundle. Classic Symfony: add new Mcfedr\ResqueBundle\McfedrResqueBundle() to AppKernel. Flex-era: add it to config/bundles.php.

3. Point it at Redis. The bundle's config lives under its extension alias (typically mcfedr_resque); read DependencyInjection/Configuration.php for the exact keys:

   mcfedr_resque:
       redis:
           host: '%env(REDIS_HOST)%'
           port: 6379
           database: 0
           namespace: 'resque'
   

4. First use case — enqueue one job. Create a job class and fire it from a controller or service:

   $resque = $this->get('mcfedr_resque'); // verify service id in Resources/config/services.xml
   $resque->enqueue('default', SendWelcomeEmail::class, [$userId]);
   

5. Start a worker. Run bin/console list resque to discover the actual command names (worker/queue commands live in the bundle's Command/ directory), then start one against your queue.

Where to look first: DependencyInjection/Configuration.php for all config keys, Command/ for worker behaviour, and the chrisboulton/php-resque README — this bundle is a thin Symfony wrapper around that library, so upstream docs explain 80% of behaviour.

Implementation Patterns

• Job classes as services. The day-to-day value of the bundle is container-aware job execution. Register job classes as services and let the bundle's job factory resolve them, so perform() gets constructor injection (repositories, mailers, loggers) instead of new $class. If the bundle version you have doesn't wire this, subclass/override the job factory — that's the main extension point.
• Enqueue from services, not controllers. Wrap enqueuing in an application service (JobDispatcher) so queues are named in one place and are easy to assert in tests. Keep controller code to $dispatcher->sendWelcomeEmail($user).
• Queue-per-concern, consumed in priority order. Use mail, imports, default; run one worker per queue under Supervisor/systemd rather than one worker for everything. A single worker taking mail,default serialises unrelated failures.
• Deploy-safe job contracts. Job args are JSON; pass scalars or a DTO that explicitly implements toArray()/fromArray(). Include a jobVersion field in payloads so you can reject or migrate stale jobs after a deploy.
• Explicit retries. php-resque does not retry automatically. Catch in perform(), re-enqueue with an incremented attempt count, and after N attempts push to a dead-letter queue or the failure backend. Do it once, in a base job class.
• Failure handling. Use php-resque's Redis failure backend plus (if present) the bundle's events to route failures into your existing logger/Sentry/alerting. Without this, failures are invisible until someone inspects Redis.
• Testing. Swap the mcfedr_resque service for a fake in the test container that records enqueues; test the job's perform() directly with real dependencies. Add one integration test against a real Redis to catch serialization regressions.

Gotchas and Tips

• Maintenance risk is the headline. Check last commit/release and PHP version support. php-resque predates PHP 8's stricter semantics and may throw deprecations or break on modern Symfony. Budget for a fork or a migration path before betting production work on it.
• pcntl/posix required. Workers fork; this does not work on Windows and often fails on restricted shared hosting. Verify extension_loaded('pcntl') in the target environment.
• Namespace/database mismatch is the #1 silent failure. Enqueuer and worker must share the same Redis host, database and namespace. If a job "vanishes", check redis-cli -n 0 keys 'resque*' first.
• Worker code is loaded once. After every deploy, gracefully restart workers (SIGTERM/SIGQUIT to let the current job finish; don't SIGKILL mid-job). Stale workers executing old class definitions cause the weirdest bugs.
• FQCN is part of the wire format. Renaming/moving a job class breaks in-flight jobs. Keep a stable alias map in the job factory if you refactor often.
• No scheduler, uniqueness, or delayed jobs out of the box. You'll bolt these on with cron + resque-scheduler or your own enqueueAt. Don't assume Horizon-style features.
• Debugging toolkit: redis-cli for resque:queues, resque:queue:<name>, resque:failed; worker verbosity flags for stdout; pipe worker output into your log shipper since php-resque only logs to stdout.
• Compare before committing. For new work, Symfony Messenger (or Laravel Queues/Horizon, if you're coming from that world) gives you retries, failure transport, DI, and monitoring for free. Choose this bundle only when you specifically need Resque interop with existing Ruby/Node consumers or legacy infrastructure.