Trying to figure out how to get Fedora CoreOS VMs provisioned on my #ovirt cluster using #terraform. I can't figure out how to do the network configuration using ignition 😓
I spent all day yesterday trying to get #terraform to provision some #fedora CoreOS and #RHCOS VMs, but I think I'm going to give up on getting Terraform to generate the ignition files, and injecting them through oVirt. I think the oVirt Ignition support is still a bit buggy.
Instead I'm going to just use terraform to build some blank VMs, and PXE boot them and point them to the ignition scripts.
Finally! got a FCOS VM built. It's an ugly process, but it works for now.
Now to repeat a few more times, and then try kubespray'ing it
Kubespray playbook is doing its thing. I'll be quite happy if this works without having to do much troubleshooting.
@mhamzahkhan just out of interest, how do you handle backups in your clusters? It's currently one of my biggest concerns especially since I don't want to develop an own backup mechanism for each application in the cluster.
@sheogorath I'm using https://velero.io/ to backup K8S resources, which drops them into a S3 bucket. It's supposed to be able to backup persistent volume as well, but as I'm using TrueNAS for my persistent storage, I just create a ZFS snapshot of all the PVs, and send the snapshots to another machine.
Yay, kubespray worked without any issues, other than having to reboot the nodes once due to how rpm-ostree installs packages.
Shiny new cluster is running. Using cri-o as well, instead of docker. :)