Getting n8n to open in a browser is the beginning of a self-hosted deployment. The harder question is whether you can recover the automation when its server disappears. Use this worksheet before moving an important workflow onto a VPS; it is a planning aid, not a tested installation recipe.
Choose an operating arrangement
| Your situation | Sensible next step | Why | Caveat |
|---|---|---|---|
| Learning Docker with a disposable workflow | Build an isolated practice instance | You can learn without interrupting a business process | Keep real credentials and important data out of the exercise |
| Comfortable administering servers, with time for maintenance | Evaluate a self-hosted deployment | You can own its configuration and recovery process | Budget operating time as well as the server bill |
| A critical workflow with nobody available to maintain its server | Evaluate managed hosting or assign an operator first | Responsibility needs a named owner before launch | Compare actual terms, export options and costs; “managed” is not a recovery guarantee |
These are recommendations, not eligibility rules. n8n itself recommends self-hosting for experienced operators and identifies server management, scaling, security and application configuration as required knowledge. Its Docker Compose guide describes a Linux deployment with these storage locations:
n8n_data, mounted at/home/node/.n8n, contains the SQLite database and encryption key.traefik_datacontains the proxy's TLS certificate data../local-files, mounted at/files, makes a host directory available to workflows.
Those names describe the documented example. Inventory your actual deployment before designing its backup; a different database or mount arrangement needs a different recovery plan.
Three hypothetical decisions
A learning project. Lina has an experimental workflow that she can recreate in an afternoon. She chooses a disposable instance and records what she learned instead of presenting it as a production service. When an experiment breaks, rebuilding is an acceptable outcome for this particular project.
A daily business workflow. Omar's report must reach a colleague each morning. He selects self-hosting only after naming a second person who can respond when he is unavailable. His fallback is a documented manual report, so a server incident does not leave the recipient wondering whether the business process ran.
A growing integration. Maya's workflow starts reading uploaded files. Her existing recovery plan covers only application state. Before enabling that change, she adds the external files to her inventory and tests with a harmless sample. The lesson is that a new workflow dependency can change recovery requirements even when the VPS itself stays the same.
One path from trial to launch
- Define the consequence of failure. Record the longest acceptable interruption and how much recent work you can afford to lose. These are your requirements, not limits supplied by n8n.
- Follow the current installation documentation. Keep the chosen version and configuration recorded privately. Prefer a small, non-critical workflow for the first deployment.
- Check the public address. For a reverse proxy, n8n documents
N8N_WEBHOOK_URL,N8N_PROXY_HOPS=1, and forwardingX-Forwarded-For,X-Forwarded-HostandX-Forwarded-Protoon the last proxy.N8N_WEBHOOK_URLreplaces the deprecatedWEBHOOK_URLfrom version 2.35.0. Consult the current proxy instructions for the applicable configuration rather than mixing examples from different versions. - Design and rehearse recovery. Use a backup method appropriate to your database and storage. Restore into an isolated environment and prevent scheduled workflows or external actions from running unexpectedly. This worksheet does not provide a database-consistent backup command.
- Record the result and decide. Compare the rehearsal against the requirements from step one. Keep failures visible and resolve them before using the instance for an important process.
Verification checklist
- A harmless test produces the expected result, not merely a successful HTTP response.
- Restored credentials can perform their intended test operation without exposing their values in the report.
- The recovery rehearsal did not send duplicate messages, transactions or other real-world actions.
- Someone other than the original installer can find the recovery instructions and explain the remaining limitations.
Costs that a VPS price omits
| Cost to estimate | What to record |
|---|---|
| Storage and transfers | Backup size, retention and any provider charges |
| Maintenance | Time reserved for reviewing changes and testing updates |
| Incident handling | Who responds, when they are available and the cost of interruption |
| External services | Charges for the APIs and destinations used by the workflow |
Enter actual supplier prices and your own time assumptions. A small server bill alone does not establish that self-hosting is the cheaper operating choice.
Start with a workload whose failure you understand. Expand only after the recovery exercise gives you evidence that the arrangement fits its purpose.
Prepared with AI assistance and linked official sources; this is not hands-on deployment testing. Report an error.