Add support for Flatcar OS

[dalees]

Fetch os_distro metadata and pass to helm chart.

Depends on Helm chart flatcar support:
https://github.com/stackhpc/capi-helm-charts/pull/53

[mkjpryor]

I'd like to wait to merge this until we know what is happening with Magnum driver selection. Because I would like to move back to using os_distro: ubuntu for the Ubuntu-based images, which would make this patch more consistent.

At the moment, for Ubuntu-based images we would get capi-kubeadm-cloudinit passed to the Helm chart as osDistro, which isn't actually correct and could have undefined behaviour in the Helm charts (although I think it works now as the stuff for ubuntu happens if osDistro != flatcar...).

[JohnGarbutt]

I'd like to wait to merge this until we know what is happening with Magnum driver selection. Because I would like to move back to using os_distro: ubuntu for the Ubuntu-based images, which would make this patch more consistent.

At the moment, for Ubuntu-based images we would get capi-kubeadm-cloudinit passed to the Helm chart as osDistro, which isn't actually correct and could have undefined behaviour in the Helm charts (although I think it works now as the stuff for ubuntu happens if osDistro != flatcar...).

I don't like using "ubuntu" because its not a valid value for Nova... and when you use a valid value Nova has a habit of doing the wrong thing.

[mkjpryor]

I don't like using "ubuntu" because its not a valid value for Nova... and when you use a valid value Nova has a habit of doing the wrong thing.

I didn't realise that - but I'm also not keen on what we are doing now. We should use whatever value Nova like for Ubuntu Jammy then.

[JohnGarbutt]

I don't like using "ubuntu" because its not a valid value for Nova... and when you use a valid value Nova has a habit of doing the wrong thing.

I didn't realise that - but I'm also not keen on what we are doing now. We should use whatever value Nova like for Ubuntu Jammy then.

The problem with that is that Nova then does strange things with it. I would rather we stop using os_distro all together, as I mentioned at the last few meetings I attended. Sorry I missed this weeks one 🤦

[JohnGarbutt]

OK, lets go with what you are proposing here.

Sorry, I merged the test refactor that conflicts with this, I should have done it the other way around, sorry! 🤦

dismissing the previous request changes after discussion with MattP and seeing the upstream spec.

[JohnGarbutt]

Ok, to help move this forward I have proposed: #27 and #28

[JohnGarbutt]

PS Sorry it took me so long to notice these!

[JohnGarbutt]

Thank you, this looks great now.