I currently have a personal nas running ubuntu server, but I’m considering moving it to opensuse leap. I’ve dabbled a bit with leap inside of virtual machines, but maybe someone more experienced with it can give me a more complete opinion. Also, is btrfs worth getting into, or can I just use ext4 and loose out on nothing (except snapshots)?
Thank you for the answer! Maybe I should look into btrfs a bit more, as I am honestly interested in learning new things (thus wanting to switch), but I’m also scared easily when the amount of things to learn increases beyond my level of comfort. The NAS is supposed to serve as the host for a couple virtual machines as well, like a pihole-VM for my home network and another VM that serves as the entry point from the internet so that I can access my nas on the go. All of that is already set up and running with the nas being on ubuntu, but I have new hardware ordered and want to try something new to go along it while keeping the functionality that’s currently covered.
I like to mess around with things, but not always and I do need to work on my patience if stuff doesn’t work and I dont understand why. That’s why I figured leap would be a good ubuntu-replacement, as it’s not as involved as something like arch would be, while also being different enough from ubuntu for me to gain more experience. Your btrfs-raid is still done through mdadm though, right? Btrfs does not come with integrated raid-like features like zfs does?
As for the reason to switch: that’s something I can get behind, although you could also just slap Proxmox on it an do all your experimenting in VMs; at least that keeps the server itself running as smoothly as possible, while not limiting you in your learning experiences.
As for btrfs: it most certainly does have RAID functionality. RAID5/6 is considered unstable (although I’ve heard/read from plenty of people who have great experiences with it, provided you don’t run into the edge cases), but I’m sticking with RAID1 because I don’t need to run the risk, and I’m not sure if waiting for a checksum calculation whenever something does go pear-shaped is going to do a whole lot of good for me.
Anyway, as for my setup: an HP Microserver (an oldie, a Gen8 with a Xeon switched in) running Leap, powering a few VMs, a collection of Docker containers, and a few “native” services (nginx, PHP, stuff like that). The root fs is a single SSD (btrfs SINGLE with some directories having a flag to disable COW), and there’s a data pool of 3 spinning disks (2x2 and 1x4GB), 4GB effective, that contains “data”. Most of it is setup with Ansible these days, hence no real use for YaST on that machine for me.
Okay, you got me convinced to look into it more!
Out of curiousity though: You have a RAID1 setup with 3 Disks that amounts to 4GB total?
That’s correct. Btrfs will simply divide your disks in 1GB chunks, and when writing, always ensure that a bit of data is always stored in 2 chunks on two different disks. You can also do 1C3 or 1C4 if your data is truly that critical, which means data is always stored in 3 or 4 chunks (on different disks), respectively. Of course, that also requires at least they amount of drives.
This chunking is also the reason why the sizes of the drives don’t have to match, as long as it’s possible to divide it evenly you won’t lose space as unused. Simply put, make sure your largest drive is not larger than your other drives combined and you should be fine.
In my case, data will always see one copy on the 4gb drive, and another on either of the 2gb drives.