- What does huangdijia/laravel-horizon-restart do?
- It is a lightweight Laravel package that exposes Horizon's worker restart capability programmatically or from an Artisan command. It sends Horizon's restart signal so workers reload gracefully without manual SSH intervention. It is a convenience control layer for Horizon, not a replacement for the Horizon runtime.
- How do I install laravel-horizon-restart?
- Install it with Composer using composer require huangdijia/laravel-horizon-restart. It is a standard Laravel package with auto-discovery, so no migrations or long-running process are required. Typical integration is to publish any config if needed and gate the restart route behind authentication middleware.
- Does laravel-horizon-restart support Laravel 12 and PHP 8.4?
- Yes, v3.1.0 is a pure compatibility release adding Laravel 12.x and PHP 8.4 support, validated via Laravel Shift. It is a minor version bump with no breaking changes, new features, or deprecations announced. The v3.0.0 line covered the Laravel 11 era, giving a bridge across the upgrade window.
- Do I need Laravel Horizon to use this package?
- Yes, this is a plugin for Laravel Horizon and depends on an existing Horizon commitment. If your application uses a database queue with queue:work under Supervisor instead of Horizon, the package is irrelevant. Horizon remains the runtime; this package only provides a control-plane surface for restarting its workers.
- How do I restart Horizon workers with this package?
- The package exposes Horizon's restart capability through an application-facing surface, such as an Artisan command or a programmatic call, wrapping Horizon's first-party restart Artisan signature. The v3.1.0 release does not add new APIs, so the exact route and command surface should be verified in the repository. In production, invoke it only through authenticated and rate-limited paths.
- Is the laravel-horizon-restart endpoint secure by default?
- The evaluation warns that an unauthenticated or under-authenticated restart endpoint is a denial-of-service primitive against queue throughput. You should gate it with middleware, an IP allowlist, a signed token, and rate limiting. The package's out-of-the-box authentication primitives are not confirmed in the provided evaluation, so review authorization before production use.
- Will restarting Horizon lose in-flight jobs?
- Terminating Horizon interrupts in-flight job processing, so jobs mid-execution can be lost or retried depending on tries, timeout, retry_after, and idempotency. That data-correctness risk is owned by the application, not the package. The package makes the restart action easier to trigger, which can increase restart frequency if not governed.
- Does one restart call restart all Horizon supervisors on every node?
- Not necessarily. Horizon runs per host, and a single HTTP call reaches one node unless the request fans out to the rest of the fleet. Operators may believe restart restarted the whole fleet when it restarted only one supervisor, so plan fan-out explicitly for multi-node deployments.
- Is laravel-horizon-restart actively maintained and tested for Laravel 12?
- The project is a small repository with one primary maintainer and low commit volume, so maintenance continuity is a procurement concern for a production control path. v3.1.0 shows validation against Laravel 12 and PHP 8.4 via Laravel Shift, but the provided evaluation does not publish a CI matrix or test coverage details. Review the repository's tests and CI before adoption.
- What are alternatives to laravel-horizon-restart?
- Laravel Horizon's built-in horizon:terminate or horizon:restart Artisan commands are the direct alternative. You can also create a custom guarded deployment hook or route for the same purpose. For non-Horizon queues, this package does not apply; use queue:work under Supervisor instead.