Forum

Hi Jeff,

I checked your server and found exactly what happened — and there's good news: most of your workflows can be recovered by yourself in a few minutes.

What happened: n8n doesn't store workflows inside its own service — it keeps all workflows and credentials in the PostgreSQL service that was deployed together with it from the template (it has been part of your n8n setup from day one, not a separate database you added later). When the volume was unmounted from the PostgreSQL service, that database — including all your workflow definitions — was deleted with it. That's why everything disappeared even though the n8n service itself was untouched.

The good news: your old n8n project from before the server migration is still running, and it still contains your workflows up to April 30 (the YouTube pipeline, EnglishQuiz, and the others). You can recover them yourself:

1. First, re-attach a storage volume to the PostgreSQL service on your new server — it currently has no volume, so anything you create now would be lost again on the next restart. In the dashboard: PostgreSQL service → Volumes → add a volume with mount directory /var/lib/postgresql/data, then restart n8n and set up your owner account again.
2. Log in to your old n8n instance (the richmanlee.zeabur.app one) — all your workflows are there. Open each workflow you want, click the ⋯ menu → Download to save it as a JSON file.
3. In your new n8n, use "Import from File" to bring each workflow back.
4. One limitation: exported JSON files don't include credential secrets, so you'll need to re-enter your API keys (Google Sheets, Groq, Gemini, etc.) once in the new instance.

What can't be recovered: workflows created after the migration (around May 11) — including the recent Reddit RSS one — existed only in the deleted database and there's no backup of it, so those would need to be rebuilt by hand. I'm sorry about that.

About the original crash: your server has 4GB of memory, but n8n is currently allowed to use up to 4GB by itself (NODE_OPTIONS=--max-old-space-size=4096), so running a heavy workflow pushes the machine out of memory and crashes n8n. We recommend lowering that value to 2048, or upgrading to an 8GB server if your workflows need more memory.

If you run into any trouble during the import, reply here and we'll help.

Hi Can,

Thanks for your quick response, and I probably forsee your response.

1. Would it be possible you may help me to migrate the n8 to the dedicate server again? Las time I tried but it doesn't work until your support.
2. I can understand the my work flows after 5/11 stored in the PostgreSQL, however PostgreSQL was not showed as a service. May you guide me why so I can have better knowledge about Zeabur to avoid issues in the future. => The PostgreSQL only show up these 2 days while I tried to create the Reddit RSS one.
3. Regard the original crash, the set up was changed a few times during my interaction with Zeabur's AI agent. Even though it indicate so after a few rounds, but I kind of suspect that as I checked the performance dash board wasn't show in that way. If I lower the value to 2048 would that also cause my workflow would not work?
4. For my workflows I didn't enable much or run in parallel at the same time, so I also a bit not sure why it need 8GB RAM. (i also checked around the performance dash board).

Thanks but seems the Zeabur AI agent wasn't quite helpful enough, I should raise the question directly.

Cheers

Jeff