Laravel 5.5队列优化：如何按需处理队列并检查运行状态？

Absolutely! Laravel gives you plenty of tools to handle queues on-demand and check their status—let's break this down step by step:

1. Manually Trigger Queue Processing #

Instead of relying on Supervisor's fixed polling, you can trigger queue jobs on-demand using either Artisan commands or direct code calls:

Using Artisan Commands #

The queue:work --once command is perfect for this—it processes one pending job and then exits immediately. You can tweak it with additional parameters:

• php artisan queue:work --once --queue=your-queue-name: Target a specific queue
• php artisan queue:work --once --max-jobs=5: Process up to 5 jobs before exiting
• php artisan queue:work --once --timeout=60: Set a 60-second timeout for job execution

Using Code (Programmatic Trigger) #

You can also call queue processing directly from your controllers, custom commands, or service classes using the Queue facade:

use Illuminate\Support\Facades\Queue;

// Process the next job in the default queue
Queue::work('default', ['maxJobs' => 1]);

// Or process all pending jobs (use cautiously to avoid overload)
Queue::work('default', ['maxJobs' => null]);

2. Check if a Queue Worker is Running or Jobs are Being Processed #

Laravel doesn't have a single "check worker status" function out of the box, but there are reliable ways to track this:

Option 1: Use queue:status (Laravel 8+) #

Laravel 8 introduced the queue:status Artisan command, which returns details about your queues including pending jobs, delayed jobs, and running workers. You can call this programmatically:

use Illuminate\Support\Facades\Artisan;

Artisan::call('queue:status');
$statusOutput = Artisan::output();

// Parse the output to check for running workers or pending jobs
// Example output might look like: "Default queue: 3 pending, 0 delayed, 1 running worker"
if (str_contains($statusOutput, 'running worker')) {
    // A worker is already active
}

Option 2: Track Worker State with Cache/Database #

You can listen to Laravel's queue events to mark when a worker starts/stops, then check that state later:

1. Register the event listeners in your App\Providers\AppServiceProvider:

use Illuminate\Queue\Events\WorkerStarted;
use Illuminate\Queue\Events\WorkerStopped;
use Illuminate\Support\Facades\Cache;
use Illuminate\Support\Facades\Event;

public function boot()
{
    // Mark worker as running when it starts
    Event::listen(WorkerStarted::class, function () {
        // Set a 5-minute expiration to handle unexpected crashes
        Cache::put('queue_worker_active', true, now()->addMinutes(5));
    });

    // Clear the mark when worker stops
    Event::listen(WorkerStopped::class, function () {
        Cache::forget('queue_worker_active');
    });
}

2. Then check the state wherever you need:

if (!Cache::has('queue_worker_active')) {
    // No active worker—safe to trigger on-demand processing
}

Option 3: Check Supervisor Status (If You're Still Using It) #

If you're keeping Supervisor for some workers, you can run the Supervisor CLI command to check status:

// Replace "your-worker-group" with your actual Supervisor worker name
$status = shell_exec('supervisorctl status your-worker-group');

if (str_contains($status, 'RUNNING')) {
    // Supervisor-managed worker is active
}

Note: Make sure your web/server user has permission to run supervisorctl for this to work.

3. Example: On-Demand Queue Workflow #

Putting it all together, here's how you might implement an on-demand handler:

use Illuminate\Support\Facades\Cache;
use Illuminate\Support\Facades\Queue;

public function handleQueueOnDemand()
{
    // Check if no worker is running
    if (!Cache::has('queue_worker_active')) {
        // Check if there are pending jobs (optional)
        $pendingJobs = \Illuminate\Support\Facades\DB::table('jobs')->count();
        
        if ($pendingJobs > 0) {
            // Process up to 10 jobs before exiting
            Queue::work('default', ['maxJobs' => 10]);
        }
    }
}

4. Bonus: Ditch Polling Entirely (If Possible) #

If you're using the database queue driver (which requires polling), consider switching to a driver like Redis or Beanstalkd. These drivers support real-time job pushing—workers will wake up immediately when a job is added, no polling required. This is far more efficient than fixed-interval polling.

内容的提问来源于stack exchange，提问作者Pankaj