If you’re like me and use n8nio/n8n:latest in a dockerfile - you probably know this already.
I broke it
I was lazy, and didn’t update my primary n8n.io instance until this morning. To all those that may use my repository, sorry! I hope you figured it out.
I fixed it
TLDR is that n8n now runs as the node user, instead of root, which is a good thing for a lot of reasons. It’s a bad thing for my Dockerfiles, which assumed differently. This has been corrected in my custom-images repository. Which, given the method the images take, ‘custom’ is kind of a stretch.
What changed?
Many things are better now, check out the official migration guide here: https://docs.n8n.io/1-0-migration-checklist/
What bit me was this simple change:
When using Docker-based deployments, the n8n process is now run by the user node instead of root. This change increases security.
If permission errors appear in your n8n container logs when starting n8n, you may need to update the permissions by executing the following command on the Docker host: …..
And I’m sure due to this, the directories for custom nodes moved around a bit:
n8n will no longer load custom nodes from its global node_modules directory. Instead, you must install (or link) them to ~/.n8n/custom (or a directory defined by CUSTOM_EXTENSION_ENV). Custom nodes that are npm packages will be located in ~/.n8n/nodes. If you have custom nodes that were linked using npm link into the global node_modules directory, you need to link them again, into ~/.n8n/nodes instead.
Resulting simple dockerfile to include a custom node on startup and rebuild:
1
2
FROM n8nio/n8n:latest
RUN mkdir ~/.n8n/nodes && cd ~/.n8n/nodes && npm install n8n-nodes-<node-name>
Since we’re here..
What else changed?
Execution order
The branching execution order is now a bit more logical- each branch executes top to bottom completely before moving to the next branch.
Data retention defaults
Starting from n8n 1.0, all successful, failed, and manual workflow executions will be saved by default.
… the EXECUTIONS_DATA_PRUNE setting will be enabled by default, with EXECUTIONS_DATA_PRUNE_MAX_COUNT set to 10,000
Good!
Python support
I know a lot of people were looking forward to this. Release the pandas!
There’s more, but these are impactful to me.
That’s all for now, hope you found that interesting!
Comments powered by Disqus.