n8n 3.0 is scheduled for October 2026, according to n8n's v3.0 breaking changes page. It is mostly a clean-up release: tighter security defaults and the removal of features that were already deprecated. If you self-host n8n, some workflows will stop working after the update unless you prepare.
As of 30 September, the stable line on n8n's GitHub releases page was still 2.41, so there is time. n8n says it will add full migration guides as the release gets closer, so check its page again before you update.
1. Check how n8n is installed
This is the biggest change. n8n 3.0 will no longer support installations run with npm or npx n8n, and self-hosted n8n will need Docker. If your n8n runs on a VPS through npm and a process manager like PM2, plan that move first.
n8n's Docker guide mounts a named volume at /home/node/.n8n, and n8n notes that this folder holds your encryption key, among other things. Without that key your saved credentials can't be read, so copy the existing ~/.n8n folder into the volume before the first start.
docker volume create n8n_data
docker run -it --rm \
--name n8n \
-p 5678:5678 \
-e GENERIC_TIMEZONE="Asia/Kolkata" \
-e TZ="Asia/Kolkata" \
-e N8N_ENFORCE_SETTINGS_FILE_PERMISSIONS=true \
-v n8n_data:/home/node/.n8n \
n8nio/n8nThis is the basic command from n8n's Docker guide with the timezone set for India. For a production server, n8n says Docker Compose is expected to be the easiest path.
2. Open the migration report
n8n's notes point to a Settings > Migration Report screen that lists the nodes and settings in your instance affected by 3.0. Run it on your current version and save a copy of what it shows.
3. Replace removed nodes
A long list of legacy nodes goes away. These are the ones most likely to appear in older small-business workflows:
| Removed in 3.0 | What to use instead |
|---|---|
| Function, Function Item | Code node |
| Cron, Interval | Schedule Trigger |
| Item Lists | Split Out, Aggregate, Sort, Limit or Remove Duplicates |
| HTML Extract | HTML node |
| Read PDF | Extract From File |
| Read/Write Binary File | Read/Write Files from Disk |
| Legacy OpenAI node, OpenAI Assistant | The current OpenAI node, or a chat model attached to an AI Agent |
AI Transform nodes are converted to Code nodes automatically on upgrade. For everything else, rebuild the step and check it with a real test run, because field names and output shapes can differ.
4. Update old AI Agent nodes
Version 1 of the AI Agent node is removed, along with its Conversational, ReAct, Plan and Execute, OpenAI Functions and SQL agent modes. Workflows already on the Tools Agent keep working once updated, so open each agent workflow and check which version it uses.
5. Review settings that changed
- Task runner timeout:
N8N_RUNNERS_TASK_TIMEOUTdrops from 300 to 60 seconds. Long Code node jobs, like big CSV clean-ups, may now time out. Split them or set the value yourself. - Storage folder:
~/.n8n/binaryDatabecomes~/.n8n/storage, and in-memory binary data mode is removed. - Community nodes:
N8N_UNVERIFIED_PACKAGES_ENABLEDnow defaults to false, so unverified community packages need an explicit opt-in. - Private network calls: the SSRF block list now includes
100.64.0.0/10. Carrier-grade NAT and Tailscale both use that range, so HTTP Request nodes that call machines on those addresses may be blocked. - Chat: Chat Hub is off by default, and Chat Trigger WebSocket messages become JSON objects with a
typefield. Embedded chat widgets need@n8n/chat1.31.0 or later.
6. Test on a copy, then update
- Back up the database and the
.n8nfolder, and export your workflows. - Start a copy of your instance on the new version with the same data.
- Run your most important workflows by hand: lead capture, WhatsApp replies, invoices.
- Update production in a quiet hour and watch the executions list for the first day.
If you'd rather not do this yourself, our n8n automation services include upgrades and workflow fixes.
n8n workflows wait for a trigger you define. On Friday we look at a different kind of automation that arrived in the last two weeks: always-on AI agents from OpenAI, Manus and Google that keep working between conversations, and what they mean for small teams.
Need help before then? Book a free 30-minute automation audit.
Want to know what you could automate?
Book a free 30-minute automation audit. We'll look at one process with you and tell you honestly whether automating it is worth it. See how we work or browse our services.
Book a free auditGet posts like this in your inbox once a week. Subscribe to NXT Weekly.
Keep reading
Voice AI Agent Latency: Where the Milliseconds Go
A phone call runs on a tighter clock than chat. Where the time goes in one turn of a voice AI agent, cascade versus speech-to-speech designs, and what actually cuts the delay.
Always-On AI Agents Are Here: What Changed in Two Weeks
OpenAI, Manus and Google launched agents that keep working between conversations. What they do, the safety problems that surfaced the same fortnight and how a small business should try them.