I am building a new server and am questioning whether it is worth it to run raid 1 or a ZFS mirror for boot drives.
I haven’t had the opportunity to work with ZFS yet and have never done any raid systems on boot drives before. How much work is required to implement this on ubuntu/ubuntu server? And would it be worth doing for my homelab use case? (amount of work vs the benefit)
I am planning on creating ansible playbooks for all of my configuration and having daily consistent backups. I’m thinking worst case without redundant boot drives I would simply get a new boot drive, install the OS, run the playbooks which should take be back to the state I was at before the boot drive failed.
There’s /boot/efi which is the fat32 ESP partition that has some kind of bootloader. That needs to be mirrored as simply as possible e.g. using mdraid.
Then there’s your kernel and initrd in /boot, I haven’t tried ZFS, but btrfs mirror works there.
You can do whatever you want from there for root /, because you can stuff whatever drivers you want into initrd.
Just give your setup a try in a VM if you don’t have ipmi/ilo/idrac/tinypilot/pikvm - from my experience it takes a few times to get it right - and to make sure it boots correctly with drives disconnected… and drives reconnected after the fact.