How to handle Ranchers Continuous Delivery?

Hi, I am kinda new to rancher. I have a test environment with rancher and rke2. I have tested a few things and like it so far, but I am a little confused by the continuous delivery part.
I have created a gitlab repo and added it to rancher CD. These are the things I observed:

  1. When I add a path in rancher in the config under “Paths”, everything works fine and rancher grabs only those file in that subpaths in git and applies them to my cluster.
  2. When I don’t add any paths, rancher seems to grab everything in root path and all subpaths but it does not grab the cluster so it does not apply the kubernetes objects anywhere.
  3. After 1, when I clone the repo from 1 with a different (sub)path, rancher also does not grab the cluster so those files are also not applied.

Actually I don’t mind 2, since using root path is not that clever, but with 3 I do have a problem.
When I want to install like different apps in my cluster, where each of them has a couple of resources (deployment, service, ingress…) I would put yml files for each of those apps in a subpath in my gitlab repo and add a repo in rancher CD pointing to that subpath, now everything is grouped for the first app and the app is installed in my cluster. But when I clone that repo in rancher CD (using “Clone” in rancher UI) and change the path, pointing it to the second app it never gets installed in my cluster because rancher does not grab my cluster a second time.

Is this as designed? or is this a bug? I kinda don’t want to add a second path to the first repo in rancher CD, because then they would not be grouped for each app and if I wanted to uninstall one of those apps it would be difficult if possible at all.

Hmm I just checked again. Now it does work, maybe there is a bug somewhere and it is not stable so it got confused with 2 so it failed with 3 afterwards…
I just deleted all repos in rancher CD, created a new one with a subpath, waited until everything was deployed and then I created another repo using “create”, not “clone” and now it does grab my cluster a second time ¯_(ツ)_/¯
I created a bug report: