System requirements

The hardware, software, and network you need to self-host Quackback. This applies to the open-source edition only; Quackback Cloud needs none of it.

Minimum requirements

Software requirements

PostgreSQL

PostgreSQL is the only datastore Quackback needs. It holds your data, the background job queue, caches, rate limits, and realtime events. You do not need Redis, Dragonfly, or the pg_cron extension.

Quackback creates both extensions on startup. If the database user can't create extensions, ask a superuser to run these once:

CREATE EXTENSION IF NOT EXISTS vector;
CREATE EXTENSION IF NOT EXISTS pg_trgm;

Warning: Realtime updates use PostgreSQL LISTEN/NOTIFY, which doesn't work through a transaction-mode pooler (for example PgBouncer in transaction mode, or a hosted pooler on port 6543). Point DATABASE_URL at the database directly or at a session-mode pooler.

Before it runs migrations, Quackback checks the server version, both extensions, and the TEMPORARY privilege. If anything is missing, it stops without changing the database and names what to fix.

Object storage

File uploads need S3-compatible storage: AWS S3, Cloudflare R2, Backblaze B2, a Railway bucket, or the bundled PGSTY Silo (a maintained MinIO fork) in the Docker Compose stack. Without storage, uploads are disabled.

Docker (recommended)

The published image is multi-arch (linux/amd64 and linux/arm64), so it runs natively on both x86-64 and ARM hosts (Apple Silicon, AWS Graviton, Ampere, Raspberry Pi).

Bun (alternative)

If you run without Docker, you need Bun 1.4.0 or newer.

Network requirements

Ports

Domains

You need:

• A domain or subdomain (for example feedback.yourcompany.com). One instance serves every module (feedback portal, help center, status page, widget) on that domain.
• A TLS certificate (Let's Encrypt works well)

Outbound connections

Quackback needs to reach:

• Your email provider (SMTP, Amazon SES, or Resend)
• OAuth providers you enable (GitHub, Google, and so on)
• Integration services you connect (Slack and others)
• Your AI endpoint, if you configure AI features

Database requirements

PostgreSQL configuration

Recommended settings for production:

# Memory
shared_buffers = 256MB
work_mem = 16MB
maintenance_work_mem = 128MB

# Connections
max_connections = 100

# WAL
wal_level = replica

Storage estimates

Add headroom for attachments, vector embeddings (for AI features), logs, and backups.

Backups

Warning: Plan for regular backups to prevent data loss. Point-in-time recovery is recommended for production deployments.

• Daily full backups
• Point-in-time recovery (recommended)
• Off-site backup storage

Cloud provider options

AWS

Minimum instance: t3.small

Google Cloud

Minimum machine: e2-small

DigitalOcean

Minimum droplet: 2 GB RAM

Railway, Render, Fly.io

All support Quackback with managed databases. Railway has a ready-made template: see Deploy to Railway.

VPS providers

Security requirements

TLS

• Required for production
• Use Let's Encrypt (free) or a commercial certificate
• Minimum TLS 1.2

Firewall

Allow only:

• 443 (HTTPS)
• 80 (HTTP, redirect to HTTPS)
• 22 (SSH, restricted IPs)

Block direct database access and public access to the application port.

Secrets

SECRET_KEY is required and must be at least 32 characters. Generate one with openssl rand -base64 32 and keep it the same across upgrades: it signs sessions and encrypts stored credentials.

Warning: Manage all secrets securely, including SECRET_KEY, database credentials, OAuth client secrets, and integration tokens. Use environment variables or a secrets manager.

Plan for scaling

• Single server: suitable for up to about 10,000 users and moderate traffic.
• Scale horizontally: for higher loads, split HTTP serving from background processing across replicas. See Scale with multiple replicas.
• Scale the database: vertical scaling, read replicas for read-heavy loads, or connection pooling in session mode (PgBouncer pool_mode = session).

Set up monitoring

• Application: the health endpoints, error tracking, request latency
• Database: connection count, query performance, disk usage
• Infrastructure: CPU, memory, disk I/O, network throughput

Related articles

• Deploy with Docker
• Install without Docker
• Scale with multiple replicas
• Environment variables
• Self-hosting FAQ