Moin! Maybe a very junior question -> Does it make sense to "export" my harvester resource yaml files to be able to rebuild a cluster, potentially through something like terraform/ansible/scripts or should I document and handroll a cluster everytime? I do not think I need to do that in prod in real life, but for the test setup it might be helpful to speed things up.

How often are you rebuilding the nodes? You can PXE boot it with a config. https://docs.harvesterhci.io/v1.5/install/pxe-boot-install

maybe 4-5 times within the next week or so, nothing permanent. I'll have a look, thank you!

PXE all the things. let me know if you want my config. Here is a vid I did a while back on it :

Do you know if I can export the yaml of my current harvester to not guess how the config should look like?

If you look in the

/oem

directory there is a yaml that will get you very close to what you are looking for.

No, you can’t rebuild Harvester using only the manifests. This is because some resources aren’t fully defined in YAML—for example, Longhorn backing images and volumes are not just interpreted from manifests.

I noticed that too, but the config is already enough a start point to create a very similar machine in case I break the single node test harvester