n8n with Postgres instead of SQLite: when and how to switch

• Signs you should switch
• A Compose setup with Postgres
• The encryption key is critical
• Migrating from SQLite
• Prune execution data
• Webhook URL
• Backups
• Scaling later
• Common errors after switching

n8n ships with SQLite by default, which is perfect for trying it out. Once workflows matter to your business, PostgreSQL is the better backend: it handles concurrent executions, larger execution history and proper backups much more gracefully.

§Signs you should switch

• Workflows run frequently and execution history grows large
• You see “database is locked” errors under load
• You want queue mode with separate worker containers
• You need reliable, consistent backups while n8n is running

§A Compose setup with Postgres

services:
  postgres:
    image: postgres:16
    environment:
      POSTGRES_USER: n8n
      POSTGRES_PASSWORD: change-me
      POSTGRES_DB: n8n
    volumes: [pg_data:/var/lib/postgresql/data]
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U n8n"]
      interval: 10s
  n8n:
    image: n8nio/n8n:latest
    environment:
      DB_TYPE: postgresdb
      DB_POSTGRESDB_HOST: postgres
      DB_POSTGRESDB_DATABASE: n8n
      DB_POSTGRESDB_USER: n8n
      DB_POSTGRESDB_PASSWORD: change-me
      N8N_ENCRYPTION_KEY: a-long-random-string
      WEBHOOK_URL: https://n8n.example.com/
    depends_on:
      postgres: { condition: service_healthy }
    volumes: [n8n_data:/home/node/.n8n]
volumes:
  pg_data:
  n8n_data:

§The encryption key is critical

n8n encrypts stored credentials with N8N_ENCRYPTION_KEY. If you lose it or it changes, every saved credential becomes unreadable. Set it explicitly, store it in a password manager, and use the same value when migrating.

§Migrating from SQLite

1. Export workflows and credentials using the n8n CLI (n8n export:workflow --all and n8n export:credentials --all) from the old instance.
2. Note the existing encryption key from the old .n8n/config file.
3. Start the new instance with Postgres and the same encryption key.
4. Import with n8n import:workflow and n8n import:credentials.
5. Test a few workflows, then switch the webhook domain.

§Prune execution data

Execution history grows fast. Enable pruning:

EXECUTIONS_DATA_PRUNE=true
EXECUTIONS_DATA_MAX_AGE=168

This keeps a week of history and stops the database from ballooning.

§Webhook URL

Set WEBHOOK_URL to the public HTTPS address. Otherwise n8n generates webhook URLs with internal hostnames and integrations fail to call back.

§Backups

Back up Postgres with pg_dump nightly and keep the encryption key separately. Without the key, a database backup restores workflows but not credentials.

§Scaling later

With Postgres in place, you can enable queue mode with Redis and run separate worker containers when a single process is no longer enough.

§Common errors after switching

• Connection refused on start: n8n started before Postgres was ready. The health check and depends_on condition above fix this.
• Credentials show as invalid: the encryption key differs from the old instance. Restore the original key rather than re-entering every credential.
• Webhooks return 404: the workflow is not active, or the proxy strips the path. Check that test and production webhook URLs both point at the public domain.
• Slow editor with big executions: large binary data is stored in the database. Set N8N_DEFAULT_BINARY_DATA_MODE=filesystem so files go to the volume instead.

A step by step deployment including domain and SSL is in this guide to deploying n8n with PostgreSQL.

Disclosure: I work on Peon, which has an n8n template, but the configuration above is plain Docker Compose.