Skip to content
algos crm.

algos crm.

CRM Documentation

The same product docs available in the dashboard—guides for setup, architecture, modules, and day-to-day use.

Getting Started · Deployment
docs/getting-started/deployment.md PDF Download all

Deployment

Production deployment guide for the multi-tenant CRM.

Architecture in production

flowchart LR
  User[Browser users] --> Web[Web server / PHP-FPM]
  Meta[Meta / external webhooks] --> Web
  Web --> App[Laravel app]
  App --> DB[(MySQL / DB)]
  Worker[Queue worker] --> DB
  Worker --> App
  Cron[Scheduler cron] --> App

You need three long-running concerns:

  1. HTTP — PHP-FPM / Octane / shared hosting entrypoint
  2. Queue worker — channels + default
  3. Scheduler — php artisan schedule:run every minute

Pre-deploy checklist

  • APP_ENV=production, APP_DEBUG=false
  • Strong APP_KEY (never commit)
  • APP_URL is the public HTTPS URL
  • SESSION_SECURE_COOKIE=true behind HTTPS
  • Real mailer (smtp / SES / Postmark / Resend) — not log
  • QUEUE_CONNECTION=database (or Redis if you configure it)
  • META_APP_SECRET set if using Facebook Lead Ads / WhatsApp
  • Database backups configured
  • SSL certificate valid for webhook callbacks

Deploy steps (typical VPS)

cd /var/www/crm
git pull origin main

composer install --no-dev --optimize-autoloader
npm ci && npm run build

php artisan migrate --force
php artisan permissions:sync
php artisan config:cache
php artisan route:cache
php artisan view:cache
php artisan event:cache   # if applicable

# restart PHP-FPM / Octane / queue workers after deploy

Always run migrations before relying on new features. For MySQL, ensure prior channel migrations completed cleanly (partial tables from failed runs must be cleaned up — see Troubleshooting).

Queue worker (required)

Channel webhooks enqueue ProcessChannelWebhookJob on the channels queue.

php artisan queue:work --queue=channels,default --sleep=1 --tries=5 --timeout=90

Recommended: supervise with Supervisor / systemd so the worker restarts on failure and after deploys.

Example Supervisor program:

[program:crm-worker]
process_name=%(program_name)s_%(process_num)02d
command=php /var/www/crm/artisan queue:work --queue=channels,default --sleep=1 --tries=5 --timeout=90
autostart=true
autorestart=true
user=www-data
numprocs=1
redirect_stderr=true
stdout_logfile=/var/www/crm/storage/logs/worker.log

After code deploys: php artisan queue:restart.

Details: Queues.

Scheduler (required)

Add a system cron entry:

* * * * * cd /var/www/crm && php artisan schedule:run >> /dev/null 2>&1

This drives lead/task reminders, trial notices, pruning, and platform alert jobs.

Details: Scheduler.

Web server notes

  • Document root: public/
  • Allow POST (and GET for Meta verify) to:
    • /webhooks/leads/website
    • /webhooks/channels/{uuid}
  • These paths are CSRF-exempt in bootstrap/app.php.
  • Prefer HTTPS end-to-end; Meta requires a public HTTPS callback for WhatsApp / Lead Ads.

Zero-downtime tips

  1. Put app in maintenance: php artisan down (optional).
  2. Deploy code + composer install --no-dev.
  3. Migrate.
  4. Rebuild caches.
  5. php artisan queue:restart.
  6. php artisan up.

Health checks

Check How
App up GET /up (Laravel health)
Scheduler alive Super Admin platform settings / scheduler_last_run_at heartbeat (every 5 minutes)
Queues draining jobs table not growing unbounded; worker logs show processed jobs
Webhooks Channel detail → Recent events; Meta webhook delivery logs

Security

  • Never commit .env or secrets shared in chat — rotate Meta tokens/App Secret if exposed.
  • Restrict Super Admin accounts tightly.
  • Keep WEBSITE_LEAD_WEBHOOK_SECRET server-side only.
  • Review failed webhook events and activity logs regularly.

Related