n8n hosting

n8n backups: what is kept and how to restore

What an n8n backup must contain, why the encryption key is the part people forget, how a managed instance is backed up and tested, and how a restore works.

By the RinooSol team Published ⁨5⁩ min read

An n8n backup is three things, and most people only take one of them. The workflows are the obvious part. The credentials are the part that makes the workflows work. The encryption key is the part that makes the credentials readable, and a backup without it is a backup of locked boxes. This guide covers what a complete backup holds, how it is done on a managed instance, and what restoring actually looks like.

What an instance holds

  • The database. Workflows, credentials (encrypted), tags, settings, users and the execution history all live in n8n's database. On a production instance that is Postgres.
  • The encryption key. A string n8n generates on first start, stored in its config file or an environment variable. Every credential is encrypted with it. Lose it and every credential in the backup is unreadable; the workflows import, the connections do not.
  • Binary data and files. Workflows that handle attachments may store them on disk, depending on configuration.
  • Community nodes. Installed packages, which are not in the database and have to be reinstalled on a restored instance.

A complete backup is the database plus the key, with the files and the list of community nodes alongside.

How a managed instance is backed up

On a RinooSol instance the backup is part of the plan, not something you set up:

  • Scheduled. The instance's data is backed up on the schedule the plan sets; the comparison table on the pricing page shows the frequency per plan.
  • Off the machine. Backups are written to storage separate from the node that runs the instance, so a disk failure on the node is not a loss of the backup too.
  • With the key. The instance's encryption key is kept with its configuration, so a restore brings credentials back readable, not as locked boxes.
  • Tested. Restores are exercised as part of operating the platform, because a backup that has never been restored is an assumption.
  • Retained for the period the terms of service state, after which old copies are removed.

What is deliberately not included: your execution history beyond the retention n8n itself keeps. It is large, it is rarely needed, and it is not your data in the way workflows and credentials are.

How a self-hosted backup is done

If you run your own instance, the minimum that works:

  1. Dump the database nightly with pg_dump (Postgres) or copy the SQLite file while n8n is stopped, to a location that is not the same disk.
  2. Copy the encryption key. It is in the n8n config file (~/.n8n/config) or the N8N_ENCRYPTION_KEY environment variable. Store it with the dump, somewhere you will find it in a year.
  3. Export workflows as JSON weekly with the CLI (n8n export:workflow --all) as a second, human-readable copy; a JSON file can be imported into any instance, which a database dump of a different n8n version sometimes cannot.
  4. Test a restore once, on a throwaway instance, before you need it. The VPS versus managed guide is honest about how often this step is skipped.

What a restore looks like

On a managed instance, a restore is a support request: tell us what happened and from when you want the data, and we restore the instance to that backup. A workflow you deleted by mistake this morning comes back from last night's copy; a credential you overwrote comes back readable. If only one workflow is needed, we can import it from the backup rather than roll the whole instance back.

On a self-hosted instance, the order is: stop n8n, restore the database dump, put the encryption key back exactly as it was, start n8n, reinstall community nodes, then open each workflow and run it once. If the key does not match, n8n starts but every credential shows an error; that is the symptom of a backup taken without the key.

Export before you leave

Whichever way you run n8n, before you cancel or move, export your workflows and credentials from n8n itself. The migration guide covers the export and the decrypted credentials file. A backup held by a provider is for recovery while you are a customer; your own export is the copy that is yours regardless.

Questions this raises

Can I download the backups of my managed instance? Yes, on request, as a database dump; the workflows and credentials export from n8n itself at any time.

Do backups include my WhatsApp API instance? A managed Evolution API instance is backed up the same way, with its session data, so a restore does not need a new QR scan unless the session itself has expired.

What if I want a backup more often than my plan takes one? Export workflows from n8n on your own schedule as JSON, or move to a plan with a higher frequency; the security page describes how the platform's own copies are kept.

Read next

Tell us what you do by hand today

Book a free consultation and we will design the automation with you.

We reply in Arabic or English

Last updated: