I have a workflow that takes a webhook, then splits the flow based on keywords held in the body of the request. The flow uses multiple nodes, including CoinGecko, Twitter, and Respond to Webhooks. When the Twitter node is an active part of the flow, I get an error before the workflow even begins:
“Problem executing workflow There was a problem executing the workflow: “One or more of the used credentials are not accessible.””
This error triggers regardless of whether the webhook path splits towards the twitter node or not. The thing is, when I clone the twitter node, detach it from the workflow, disable the twitter node within the flow and execute the cloned node by itself, it sends tweets just fine.
I have also tried replacing the dynamic content for the tweet with placeholder temporary text to be sure that it’s not the formatting of the webhook content causing the problem. Additionally, all other nodes work fine as long as I detach the twitter node from the rest of the workflow. Finally, I have other twitter workflows that do not have this problem, and I’m really scratching my head here.
Running on Version 0.178.2 of n8n via docker-compose on ubuntu 20 with digital ocean.
Thought I should also mention I’m a huge fan of n8n, and I appreciate what you all are doing!
Hey @shalalalaw, welcome to the community and sorry for the trouble.
This sounds like it could be a permission problem. Do you have multiple users on your n8n instance? Does the problem persist if you remove the Twitter node and add a new one from scratch (not copied from another workflow)?
Thanks @MutedJam , I do have multiple users on the n8n instance. This error is occurring when logged in to the admin level account, and the error does persist if I recreate the node from scratch. Again, it has the same thing going on where the node executing in isolation works, but I get the same credential error before the execution begins when the node is part of the flow.
@MutedJam I was able to get the workflow working, though the problem wasn’t actually ‘solved’. I created a brand new workflow and copied over the entirety of my original flow (as pasted above) into the new flow. Voila, it started working.
My theory, for bug-hunting purposes, is that this happened because the workflow is about a week old, and the particular twitter credentials being used were only a day or two old. Perhaps the workflows cache the availability of credentials?
I hope this helps, and thanks for your time!
Hey @shalalalaw, thanks so much for confirming, I am glad to hear it’s working. I played around with this for quite a while but could ultimately not reproduce this problem. Did you by any chance upgrade n8n between creating the workflow and creating/setting the credentials? And would you happen to remember the upgrade path?
@MutedJam it’s likely that I upgraded between the creation of the workflow and setting the new credentials, I cannot confirm what versions I moved between if I did. I update n8n every week or two, and that workflow was about a week and a half old. When I update n8n, I use docker-compose pull > down > up.
I can also tell you that after I had exhausted all the steps I could think of for resolving the issue, I upgraded from 0.178.2 to 0.179.0. This did not fix the issue, but when I copy-pasted the workflow to get it working again, it was in the latter version.
I hope this helps!
Once again thanks so much for sharing!