This message was deleted.

I think it is safe to say most people are running harvester on bare metal. There are some posts on how to get harvester running on, say amazon’s ec2, but anything in production will be on bare metal machines.

it is pretty easy to run it on equinix metal too

yes, that helps! Equinix seems rather expensive (e.g. more expensive than Google Cloud). Are there more providers people are using Harvester on?

Is there a reason you’re looking to run it on a cloud provider? I could see it being good to run in a colo environment where you have bare metal in someone else’s datacenter, but running in a traditional cloud typically means running it in a VM… So, you’d be double virtualized. Seems inefficient.

I prefer to run harvester bare metal, too. With that being said, I’m looking for solutions that offer (in this order) 1. cost-effective compute 2. easy means of installing Harvester, e.g. via PXE or cloud-init 3. elasticity: ability to automate adding/removing compute. Not sure if this is possible with bare metal

You might want to check out https://www.server4you.com/ I don’t know if their bare metal product has Harvester support, but it’s worth bringing up to their sales team.

Thank you for the pointer! I checked their website and there is nothing about dedicated networks, VLANs, or floating IPs (as needed by Harvester for the VIP or exposed services). I’ve contacted their support to find out more.

The other is to find a colo datacenter and build all of that out yourself.