The nodes don’t show. This happens for all the workflows without exception.
I updated from 1.94.1 to 1.106.3 yesterday and hadn’t noticed this issue before. I tried to rollback, thinking it was related to the upgrade, but the issue persists in 1.94.1.
The reason seems to be linked to the call to GET /rest/community-node-types not responding with any data.
When I stop the server while the “Loading” shows, it hides once the server is cut, and if I navigate using “browser previous” back to the workflows and open a workflow then it shows the nodes, but nothing is clicable, and refreshing the page restarts the “Loading” issue forever.
I don’t explain the root of the issue, but the consequence is clear: We can’t view/update any of our existing workflows, although they seem to be running just fine.
Things were working fine on July 28th as another coworker had done things and recorded them for me, so I can tell it’s an issue that happened since then.
There is a call to n8n’s api to get available community nodes so that it can be displayed in the node list (at least thats my understanding, if not that its something similar). It is that external service that was down not an error in n8ns actual code.
Seems like it was accidentally made as a blocking request, meaning the ui wont load until the api call responds. So the community node service was down so every version that relies on it will be effected
It seems to be working now without the workaround though
Note that some of my workflows now display empty instead of display their actual content.
Navigating between tabs somehow helps in making them show the workflow’s content again.
But nodes cannot be opened/edited nor it is possible to display the past Executions that happened.
I thought our self-hosted instance was not linked to n8n servers except for the authentication/licensing stuff, but I guess I was wrong and there are a lot more APIs being shared under the hood.
I’m growing growingly concerned about this issue, I haven’t received feedback, as if everything was working perfectly fine, while the reality is that our instance is partially unusable at the moment.
It’s still impossible to edit any node, although I have access to the history of executions and can see our workflows running, editing them is not possible anymore.
To be more precise, what happens is the following:
When first navigating to a workflow, in any project, and opening any node, the “edit node” modal doesn’t show
But, if I switch project and open another node then edit modal shows properly
And, if I go back to the first opened project and open the same node as before, now if opens the edit node modal properly
So, there seem to be a bug that is mostly visual and affects the first attempt of opening the “edit modal” of a node, navigating in n8n and opening any other workflow/node fixes the problem, until the page is refreshed, and we have to apply the same trick again.
Note that you don’t have to switch project either, just switching workflows works as well.
That’s very counterintuitive and obviously a bug, which doesn’t seem to be affecting everyone.
We’re on self-hosted with an NGO/Enterprise license.
I’m running 1.107.2 (Self hosted, Postgres, Cloudron) and I’ve also noticed situations where some running workflows are blocking the n8n interface from loading. This thread has provided some interesting insights into why. Glad to hear it is being worked on.
The situation seems to be fixed, noticed yesterday.
But I must say again, I find this concerning, the fact that some change at n8n can completely break my self-hosted instance out of the blue without any insights or support.
This particular issue was not even related to updating versions on our instance, which I first assumed to be.
The problem went away as it appeared: silently and without notice.
