Not sure if this is a question or a feature request.
We’ve had various timeout errors on workflows that when we debug them show there was some kind of error with credentials on the last notifying node (rocket.chat) in our case.
The credentials issues seem to be coming from this being created by a sub-user when executed by the main user. Nevertheless, what I am not sure if is the desired behaviour is that even if that last rocket.chat node has a “Continue on errors” the whole workflow fails for that reason.
We’ve now tried to solve it, passing it to a new workflow that is fully dedicate to rocket.chat notifications.
Just wondering if “continue on errors” should apply to any kind of errors. I have the feeling that n8n reviews all credential nodes before execution, but I even so don’t understand why we get a timeout instead of an error, even if the workflow has an error capture.
Just wondering if this is really the intended behaviour.