Production Checklist for Zerops

Keywords

production, checklist, ha, high availability, minContainers, mailpit, smtp, adminer, volatile, sessions, object storage, deploy production, go-live, launch

TL;DR

Before going to production: (1) databases to HA mode, (2) minContainers: 2 on app services, (3) replace Mailpit with real SMTP, (4) remove Adminer, (5) use Object Storage for uploads, (6) use Redis/Valkey for sessions.

Database

Item	Dev	Production
Mode	NON_HA	HA (must recreate)
Backups	Optional	Enabled
Connection	Single primary	Primary + read replicas

HA is immutable — cannot switch after creation. Delete and recreate with mode: HA.

Application Services

Item	Dev	Production
minContainers	1	2+
Health checks	Optional	Enabled
Logging	Console/debug	Structured (syslog)
Debug mode	Enabled	Disabled

# Production app service
- hostname: app
  type: nodejs@22
  minContainers: 2
  maxContainers: 4

Dev Services to Remove

Mailpit → Production SMTP

# REMOVE for production:
- hostname: mailpit
  type: go@1
  buildFromGit: https://github.com/zeropsio/recipe-mailpit

# REPLACE with production SMTP env vars:
envVariables:
  SMTP_HOST: smtp.sendgrid.net
  SMTP_PORT: "587"
envSecrets:
  SMTP_PASSWORD: your-production-key

Adminer → Remove or Restrict

Remove entirely or disable enableSubdomainAccess. Use VPN + pgAdmin/DBeaver locally.

File Storage

Container filesystem survives restarts but is replaced on every deploy — files stored on disk persist through reload/restart/stop+start but are lost on deploy or container replacement (scale-up/down).

Use case	Solution
User uploads	Object Storage (S3)
Media files	Object Storage (S3)
Temp files	Container disk (OK)
Build artifacts	Deploy via zerops.yaml

# Add Object Storage for persistent files
- hostname: storage
  type: object-storage
  objectStorageSize: 2
  objectStoragePolicy: public-read

Sessions & Cache

File-based sessions break with multiple containers and are lost on deploy.

Use case	Solution
PHP sessions	Redis/Valkey
Laravel sessions	Redis driver
Django sessions	Redis backend
Express sessions	Redis store

# Add Valkey for sessions/cache
- hostname: cache
  type: valkey@7.2
  mode: NON_HA  # HA for production

Framework-Specific Production Settings

PHP/Laravel

• APP_ENV: production, APP_DEBUG: "false"
• Trusted proxies: TRUSTED_PROXIES: 127.0.0.1,10.0.0.0/8
• Sessions in Redis, not files
• Optimize: php artisan config:cache && route:cache && view:cache

PHP/Symfony

• APP_ENV: prod
• TRUSTED_PROXIES: 127.0.0.1,10.0.0.0/8
• Logging via Monolog SyslogHandler

Python/Django

• DEBUG: "false"
• CSRF_TRUSTED_ORIGINS: https://your-domain.com
• ALLOWED_HOSTS: .zerops.app,your-domain.com
• Static files via collectstatic

Node.js

• NODE_ENV: production
• HOST: 0.0.0.0
• Health check endpoint at /status or /health

Java/Spring

• server.address: 0.0.0.0 (required — default binds localhost)
• Actuator health endpoints enabled
• JVM memory flags: -Xmx512m (match container limits)

Elixir/Phoenix

• PHX_SERVER: "true" (required to start server in releases)
• SECRET_KEY_BASE generated via preprocessor
• PHX_HOST set to domain

HA Checklist

Item	Recommendation
Core package	Serious Core for production (better SLA, dedicated resources)
CPU mode	cpuMode: DEDICATED for consistent performance under load
Environment separation	Separate projects for dev/staging/prod
Stateless design	Sessions in Valkey, uploads in Object Storage — no local state
Database mode	mode: HA for all managed services (immutable — plan before creation)
Min containers	minContainers: 2 on all app services for zero-downtime deploys

Health Check Pattern

Combined readiness + runtime health check for production services:

zerops:
  - setup: app
    deploy:
      readinessCheck:
        httpGet:
          port: 3000
          path: /health
    run:
      healthCheck:
        httpGet:
          port: 3000
          path: /health
      start: node server.js

Readiness check gates traffic during deploy. Health check runs continuously — unhealthy containers are restarted after 5-minute retry window.

Gotchas

1. HA is immutable: Must delete and recreate service to switch modes
2. Container filesystem survives restarts but is replaced on every deploy: use external storage for persistent data
3. File sessions break with scaling: Multiple containers don't share filesystem
4. Mailpit is not production SMTP: Only for dev — no delivery guarantees
5. Debug mode leaks secrets: Disable APP_DEBUG in production
6. Missing health checks: Load balancer can't route around unhealthy containers

See Also

• zerops://themes/core — import.yml patterns
• zerops://guides/scaling
• zerops://guides/backup