I am trying to teach myself new and better means to deploy and control oVirt and HCI
deployment.
Does anyone have notes, experience and or examples.. website etc.. where someone managed
deployment and admin of HCI deployment (CentOS + Cockpit + gluster + oVirt) and instead of
using shell and manaual / gui deployment, did this "Infrastructure as code"
Ex Step 1:
1) Foundation with kickstart. three nodes with "minimal install" all with ssh
keyless password login for user "ansible"
2) Keys exchanged between nodes for keyless login for root and ansible user
3) Using ansible publish HCI deployment playbook.
-> Define drive for each host to use
-> Define name and host / ip of engine
-> install HCI engine with gluster
-> Add SSH keys for ansible and access so future oVirt interaction can be done via
ansible
I am just starting to research this.. but hoping others have blazed this path and I can
leverage their process.
Concerns:
1) How to get HCI to deploy only engine layer on (in my case... 512GB SSD .. with VDO and
avoid destroying VMS not related to engine).. aka if I have to repair engine again will it
keep nuking the volumes "data" and "vmstore" volumes which I think the
HCI deployment wants to do if it has to redeploy fresh.
2) If above is always to be an issue .. can I make default VM deployment be a dedicated
gluster volume (my case is 1TB SSD with VDO in each server) such that rebuild of engine
layer can just slurp back in the VMs from that volume.