Production deployment guide

This guide covers production defaults for running Laravel queue workers with RabbitMQ and native ext-amqp.

Recommended production baseline

QUEUE_CONNECTION=rabbitmq
RABBITMQ_CONSUME_MODE=consume
RABBITMQ_HEARTBEAT_CONNECTION=30
RABBITMQ_CONNECT_TIMEOUT=10
RABBITMQ_READ_TIMEOUT=30
RABBITMQ_WRITE_TIMEOUT=30
RABBITMQ_MAX_RETRIES=3
RABBITMQ_RETRY_DELAY=1000
RABBITMQ_HEALTH_CHECK_ENABLED=true
RABBITMQ_HEALTH_CHECK_INTERVAL=30

Use consume mode for hot queues and poll mode when you want the most Laravel-like worker lifecycle.

Supervisor

[program:laravel-rabbitmq]
process_name=%(program_name)s_%(process_num)02d
command=php /var/www/html/artisan rabbitmq:consume --queue=default --consume-mode=consume --memory=256 --tries=3 --timeout=60 --num-processes=1
autostart=true
autorestart=true
stopasgroup=true
killasgroup=true
user=www-data
numprocs=4
redirect_stderr=true
stdout_logfile=/var/log/supervisor/laravel-rabbitmq.log
stopwaitsecs=3600

Scale with numprocs, containers, or separate deployments. Prefer one hot queue per worker group when using consume mode.

systemd

[Unit]
Description=Laravel RabbitMQ worker
After=network.target

[Service]
User=www-data
Group=www-data
Restart=always
WorkingDirectory=/var/www/html
ExecStart=/usr/bin/php artisan rabbitmq:consume --queue=default --consume-mode=consume --memory=256 --tries=3 --timeout=60
TimeoutStopSec=3600
KillSignal=SIGTERM

[Install]
WantedBy=multi-user.target

Docker Compose worker

services:
  worker:
    build: .
    command: php artisan rabbitmq:consume --queue=default --consume-mode=consume --memory=256 --tries=3 --timeout=60
    restart: unless-stopped
    depends_on:
      rabbitmq:
        condition: service_healthy

  rabbitmq:
    image: rabbitmq:3.13-management
    ports:
      - "5672:5672"
      - "15672:15672"
    healthcheck:
      test: ["CMD", "rabbitmq-diagnostics", "ping"]
      interval: 10s
      timeout: 5s
      retries: 5

Kubernetes worker

apiVersion: apps/v1
kind: Deployment
metadata:
  name: laravel-rabbitmq-worker
spec:
  replicas: 3
  selector:
    matchLabels:
      app: laravel-rabbitmq-worker
  template:
    metadata:
      labels:
        app: laravel-rabbitmq-worker
    spec:
      containers:
        - name: worker
          image: your-app:latest
          command: ["php", "artisan", "rabbitmq:consume"]
          args: ["--queue=default", "--consume-mode=consume", "--memory=256", "--tries=3", "--timeout=60"]
          envFrom:
            - secretRef:
                name: laravel-env
          resources:
            requests:
              cpu: 100m
              memory: 256Mi
            limits:
              cpu: 1000m
              memory: 512Mi

Prefetch and concurrency

Start conservative:

'options' => [
    'queue' => [
        'qos' => [
            'prefetch_size' => 0,
            'prefetch_count' => 10,
            'global' => false,
        ],
    ],
],

Increase prefetch_count only after measuring processing time, memory usage, and retry behavior.

Publisher confirms

Enable publisher confirms for workflows where message persistence must be acknowledged by RabbitMQ:

RABBITMQ_PUBLISHER_CONFIRMS_ENABLED=true
RABBITMQ_PUBLISHER_CONFIRMS_TIMEOUT=5

This improves delivery confidence at the cost of extra publish latency.

Quorum queues

Use quorum queues for stronger queue durability in RabbitMQ clusters:

RABBITMQ_QUEUE_QUORUM=true

Do not combine quorum queues with priority queues.

Dead-letter routing

RABBITMQ_REROUTE_FAILED=true
RABBITMQ_FAILED_EXCHANGE=failed.jobs
RABBITMQ_FAILED_ROUTING_KEY=%s.failed

Create dashboards and alerts around failed exchanges and DLQs.

Operational checklist

• Build PHP images with ext-amqp installed and verified.
• Set heartbeat and timeout values explicitly.
• Use Supervisor, systemd, or orchestrator restarts.
• Restart workers during deploys.
• Monitor worker memory and restart before memory pressure.
• Use separate worker groups for high-priority queues.
• Enable publisher confirms where message loss is unacceptable.
• Use DLQs for failed-message inspection.
• Document queue topology in version control.