This message was deleted.

There was talks internally about making Rancher on Harvester a first class citizen. I would use k3s on a separate node or small form factor.

I’ve been using vcluster since 1.3 - works great. It might be replaced but certainly not anytime soon.

Thank you guys! @thousands-advantage-10804 when you mean you would install a k3s on a separate node, you mean, you would have a VM on another Virtualization platform and/or a SLE + K3s running rancher manager on top?

or a small bare metal node?

We just followed the official documentation which says "Rancher manager should be installed on a different node". However this become a can of worms when you think about "making it right" and end up having like a proxmox to run a rancher manager in HA, to manage your Harvester(s).. We are running SLE + Harvester and MicroOS + k3s + rancher and while evaluating - for a different project - the cisco solution using harvester, we saw that they have everything in one box.. by the way, @thousands-advantage-10804 the guys in my team are your fan 🧑‍🚒

oh cool. tell them I said HI!. it is a can of worms. You need to balance complexity with simplicity. Ask the team how much failure your “system” can live with. Then engineer for that. Meaning, if you need absolute uptime. I would have at least 9 servers. 3 for JUST rancher in HA mode. And 6 for harvester. Harvester will automatically make the first 3 has controllers. I would have no less than 20gb networking between all the nodes and SSDs.

Thank you @thousands-advantage-10804 ! That was definitely helpful. Harvester works great, but unfortunately, it lacks good, up-to-date documentation. Your videos are always incredibly helpful and kind of fill in the gaps. Interestingly, part of my team consists of firefighters too, so I guess there’s some natural connection between Kubernetes and firefighting 🧯

we talked to them about the documentation. tell your firefighters to stay safe!

@freezing-ability-583 what Cisco solution are you looking at?