0

I will be getting VMs at work for installing a Kubernetes cluster. It will be used by the development team; but we need to test the scaling too during development, so single node is not really an option.

I won't have rights to create or re-spawn the VMs from known images. So I need to be able to cleanup a cluster if it gets damaged beyond repair and do it without asking for a VM refresh.

To sandbox ordinary software I would generally use containers. I wanted to do the same for the Kubernetes installation.

I only found kind and k3d. But both are meant to simulate multi-node clusters while using a single node. Whereas I want to have a real multi-node solution that can run in docker or its alternatives.

Dojo
  • 5,374
  • 4
  • 49
  • 79
  • 1
    When you install your entire system on a btrfs formated filesystem, you could use the btrfs snapshot function, which allows you to do snapshots of your entire system, equivalent to a virtual machine snapshot function. It can be highly customized (read only master snapshot, directories not being target of snapshot, raid, etc.). I can really recommend this. When using btrfs for the first time, you should take your time and learn about the filesystem, as it provides many options and chances, but also some dangers (mostly caused by untrained personal which is using that filesystem), if not known. – paladin Apr 27 '22 at 06:00
  • @paladin btrfs snapshots sound interesting will check it out of curiosity. But I work at a very conservative company. We use known VM images, I don't think I can play around with filesystems. Looking for a distro level solution here. – Dojo Apr 27 '22 at 11:07
  • SUSE Linux has a mature built-in btrfs-snapshot functionality, ready for business use. --> https://www.suse.com – paladin Apr 27 '22 at 13:18
  • PS SUSE Linux might also be a good distro for Kubernetes in general. – paladin Apr 27 '22 at 13:24

0 Answers0