This message was deleted.

Why 6 nodes? Trying to run Rancher HA I use three nodes as control-plane but also leave them as workers too as you don't really want to put a lot of other load on a Rancher cluster anyway. If you follow that sort of pattern, you can only scale to odd numbers as the ETCD on your control plane nodes needs to have an odd number for voting. Also remember that scaling to a certain point gives you diminishing retuns with the nodes having to sync with each other and adding more nodes later isn't a big deal. So I'd recommend that you start with three nodes and add more if you need. If you're using cert-manager with Rancher, it's in the same cluster hosting Rancher. Last time I used Rancher I don't recall external-dns and I'm not familiar with fluxcd. If you want fluxcd to help manage all the clusters with Rancher then you probably need to install it to the same cluster as Rancher, however you should keep such things to a minimum. Though I suppose if you're cramming the Rancher cluster with other workloads, that might be a reason to have dedicated worker nodes too.

What I ended up doing is installed external-dns, metallb, cert-manager and nginx-ingress with fluxCD - and right as I am writing this I am installing rancher helmchart with fluxcd.