mleczakm / swoole-bundle-scheduler

v0.1.0 2026-09-03 09:53 UTC

Requires

• php: ^8.3
• ext-swoole: *
• psr/cache: ^2.0 || ^3.0
• psr/clock: ^1.0
• psr/event-dispatcher: ^1.0
• psr/log: ^3.0
• swoole-bundle/swoole-bundle: ^0.32
• symfony/cache: ^7.4 || ^8.0
• symfony/clock: ^7.4 || ^8.0
• symfony/config: ^7.4 || ^8.0
• symfony/dependency-injection: ^7.4 || ^8.0
• symfony/http-kernel: ^7.4 || ^8.0
• symfony/messenger: ^7.4 || ^8.0
• symfony/scheduler: ^7.4 || ^8.0

Requires (Dev)

• friendsofphp/php-cs-fixer: ^3.94
• macpaw/symfony-health-check-bundle: ^2.1
• phpstan/extension-installer: ^1.4
• phpstan/phpstan: ^2.1
• phpstan/phpstan-phpunit: ^2.0
• phpunit/phpunit: ^11.5 || ^12.5
• symfony/framework-bundle: ^7.4 || ^8.0
• symfony/lock: ^7.4 || ^8.0

Suggests

• macpaw/symfony-health-check-bundle: Enables the /health scheduler heartbeat check (swoole_bundle_scheduler.health_check)
• symfony/lock: Enables the opt-in cross-process tick lock (swoole_bundle_scheduler.lock)

MIT

• Michał Mleczko

• symfony
• cron
• scheduler
• Messenger
• swoole
• openswoole
• swoole-bundle

Published on Packagist.org on 2026-09-03 09:56 UTC

README

Run Symfony Scheduler inside the Swoole HTTP server, on a Swoole\Timer::tick() loop - no separate messenger:consume scheduler_* worker process to run, supervise, restart, or pay memory for.

A companion package to swoole-bundle/swoole-bundle.

Install

composer require mleczakm/swoole-bundle-scheduler

// config/bundles.php
SwooleBundle\Scheduler\SwooleBundleSchedulerBundle::class => ['all' => true],

# config/packages/swoole_bundle_scheduler.yaml
swoole_bundle_scheduler:
    enabled: true
    interval: 60

Define a schedule the normal Symfony way (#[AsSchedule] + ScheduleProviderInterface) and it is polled automatically. Restart the Swoole server to pick it up.

What you get

• Polls every #[AsSchedule] provider off the Swoole event loop, dispatching due messages through the message bus with the same PreRunEvent / PostRunEvent / FailureEvent as Symfony's own worker.
• A failed tick is logged and retried next interval - it never crashes the server.
• Overlapping ticks are skipped, not queued.
• Each tick resets the pooled stateful services it touches, via the swoole-bundle's own coroutine boundary machinery.

Opt-in production hardening

All disabled by default; enable per app when you hit the failure mode:

Option	Defends against
lock	The tick firing in a second OS process (Swoole manager during reload) and double-dispatching
watchdog	A hung Scheduler::run() wedging the scheduler forever behind a never-freed semaphore
heartbeat + health_check	A wedged scheduler staying invisible while /health keeps passing
pre_run / after_tick	App-specific readiness checks / state resets around each tick

Docs

• docs/usage.md - walkthrough, full config reference, caveats
• docs/production-hardening.md - when and how to enable each hardening option
• examples/ - copy-paste schedule, message, handler, config, custom scheduler

Requirements

• PHP >= 8.3, ext-swoole
• swoole-bundle/swoole-bundle ^0.32 with coroutines enabled
• symfony/scheduler + symfony/messenger
• symfony/lock (only for the lock option)
• macpaw/symfony-health-check-bundle (only for the health_check option)

License

MIT