A little confused about the supported ways of getting cloud-init data into RancherOS

Hey, everyone.

tldr; Can someone point me to the docs for feeding cloud-init/config data into RancherOS when not running on AWS and thus no user-data service to query?

In my particular case I’m attempting to run a couple of RancherOS instances on top of KVM (virt-manager is also in the mix). I’ve found the various cloud-config docs and some forum threads about config disks and such but I’ve yet to find docs bridging the two topics.

Much appreciated.

if you install RancherOS into disk, thing is really simple. Just boot from iso, scp clound-init config into your node, then run a command similar like this:

ros install -c cloud-config -d /dev/vda 

After this, both cloud-config and RancherOS is persistent in your harddisk.

Thank, Liyi.

I was making it too hard…trying to start with lots of stuff in a config-disk as opposed to little more than an SSH key to get started.

config-disk is less obvious to me. If I would like to automate deployment a lot of nodes, I prefer to put a cloud-config into a dir, and run a simple python web server there, e.g.
python3 -m http.server 8000 &

I then just indicate in my ipxe boot file to download the file from there.

It’s a fair idea. At current I’m just getting familiar with RancherOS and Kubernetes and only need a few VMs for testing. Production will either be deployed into AWS or GCE and thus should have some sort of user-data service available.