laravel-horizon

When to use

Use this skill for anything related to Laravel Horizon:

• Queue worker configuration and supervision
• Horizon dashboard setup and access control
• Job metrics, throughput, and failure monitoring
• Balancing strategies and scaling workers
• Production tuning and deployment

For writing queue jobs themselves, see jobs-events.

Procedure: Configure Horizon

1. Inspect current queue setup — Read config/queue.php, config/horizon.php (if present), and composer.json to identify the current driver, supervisors, and queues in use.
2. Pick environments and balance strategy — Decide per-env maxProcesses, balance, tries, timeout based on workload; review the config block below.
3. Apply configuration — Edit config/horizon.php, register supervisors and queue priority.
4. Verify — Run php artisan horizon locally, hit /horizon dashboard, confirm jobs flow and metrics appear.

config/horizon.php

'environments' => [
    'production' => [
        'supervisor-1' => [
            'maxProcesses' => 10,
            'balanceMaxShift' => 1,
            'balanceCooldown' => 3,
            'connection' => 'redis',
            'queue' => ['default', 'high', 'low'],
            'balance' => 'auto',  // auto, simple, or false
            'tries' => 3,
            'timeout' => 60,
            'maxTime' => 3600,
            'maxJobs' => 1000,
            'memory' => 128,
        ],
    ],
    'local' => [
        'supervisor-1' => [
            'maxProcesses' => 3,
            'connection' => 'redis',
            'queue' => ['default', 'high', 'low'],
            'balance' => 'auto',
            'tries' => 3,
            'timeout' => 60,
        ],
    ],
],

Queue priority

// Higher priority queues are listed first
'queue' => ['high', 'default', 'low'],

Running Horizon

# Start Horizon (foreground)
php artisan horizon

# Pause / Continue
php artisan horizon:pause
php artisan horizon:continue

# Terminate gracefully (for deployments)
php artisan horizon:terminate

# Check status
php artisan horizon:status

Deployment

Always terminate and restart Horizon after deploying new code:

php artisan horizon:terminate
# Supervisor will auto-restart Horizon

Supervisor config (production)

[program:horizon]
process_name=%(program_name)s
command=php /path/to/artisan horizon
autostart=true
autorestart=true
user=www-data
redirect_stderr=true
stdout_logfile=/var/log/horizon.log
stopwaitsecs=3600

Dashboard access

// HorizonServiceProvider
protected function gate(): void
{
    Gate::define('viewHorizon', function (User $user): bool {
        return $user->isSuperuser();
    });
}

Balancing strategies

Strategy	Behavior
auto	Distributes workers based on queue workload (recommended)
simple	Round-robin across queues
false	Fixed worker count per queue

Metrics and monitoring

• Throughput — jobs processed per minute per queue
• Runtime — average job execution time
• Wait time — time jobs spend waiting in queue
• Failed jobs — track and retry from dashboard
• Tags — auto-tagged by Eloquent models, custom tags via tags() method

// Custom tags on a job
public function tags(): array
{
    return ['customer:' . $this->customer->getId(), 'report'];
}

Core rules

• Always use Supervisor in production — Horizon is a long-running process.
• Terminate on deploy — horizon:terminate ensures workers pick up new code.
• Use auto balancing — it adapts to workload automatically.
• Set maxTime — prevents workers from running forever (memory leaks).
• Set maxJobs — recycles workers after N jobs to prevent memory bloat.
• Tag jobs — makes debugging and filtering in the dashboard much easier.

Output format

1. Updated Horizon configuration with supervisor and queue settings
2. Environment-specific balancing strategy rationale

Auto-trigger keywords

• Horizon
• queue worker
• queue dashboard
• job monitoring
• supervisor
• queue balancing

Gotcha

• Horizon config changes require php artisan horizon:terminate and restart — they don't hot-reload.
• Don't set maxProcesses too high — each process holds a DB connection. Monitor your connection pool.
• The model forgets that Horizon only works with Redis queues — not database or SQS.

Do NOT

• Do NOT run Horizon without Supervisor/systemd in production.
• Do NOT expose the Horizon dashboard without access control.