Hi all!
I’m in the process of migrating my home server from Unraid to TrueNAS with a ZFS pool, as well as upping storage from 2 6TB drives to 4. Unfortunately, because of either my bad luck or incompetence, it seems like one of the drives has died. So, here’s my question. I’ve read up a bit on resilvering and I know that if I replace the dead drive with a larger drive, the pool will be unable to use that extra space until the remaining drives are upgraded, but would there be any other drawbacks? Especially if the pool was left running in this configuration for an extended period.
I definitely see myself upgrading the pool to larger drives in the future, and it would be nice to save myself buying an extra drive that may end up getting replaced before the end of its life. (Note: I’m aware that resilvering isn’t the safest way of upgrading a pool, but the data on the pool is either backed up or non-essential, so I’m fine with the risks)
You must log in or # to comment.
I have mismatched drives. Mostly 8TB, and 2 10TB. The 10TB are simply wasted money, but no other downsides.
I haven’t seen any deleterious effects from what you’re proposing and I haven’t heard of any either. Were it me, I’d go for it.
Agreed. It’s not a safety issue it just won’t use the full drives capacity until they replace the other smaller drives
Not an expert, but as far as I know running with mismatched drives isn’t bad - it’s just a pointless long-term configuration.
Are you planning on mirroring the drives? If you upgrade to 4 drives in the future you could just do 2 mirror vdevs with mismatched drive sizes per vdev. Also I’m surprised you’re bothering with 6TB drives - you can get 14TB very cheap nowadays (recertified).
I looked at doing two vdevs but was put off by the lower usable storage. At a certain point, maybe that’s not as important as I think though.
Yeah, the choice for 6TB wasn’t my best. I got the two older drives a few years back on a Newegg flash sale, and it seemed like plenty, especially considering Unraid’s model of 1 parity drive and 100% usable storage on the data drive(s). Then, when I decided to upgrade, I was too cheap to go buy 4 whole new drives, so I just went with more of what I already had (to add insult to injury, they’re all WD Red drives…).
Yeah lower usable storage sucks, but it’s an okay trade at the moment given its flexibility. raidz expansion is always “right around the corner” but never actually arrives - if it’s merged a raidz array might make more sense for a smaller NAS.
Personally I’ve found that if you’re really budget-conscious or have a low number of disks, MergerFS+SnapRAID backed by BTRFS disks is a better choice for flexibility (very similar to an Unraid setup), since you can add/remove at any time, and they can be mismatched drive sizes.
Can I ask why you’re migrating to TrueNAS? I have been running unraid for a couple years and have been happy enough. I’m no expert so I mostly use basic features. Am I missing out on something TrueNAS has to offer?
Part of me is starting to wonder, honestly. I will say that the web UI for TrueNAS Scale is leagues better than Unraid’s, which to me always felt confusing and hacked together. ZFS is also really nice, although Unraid did recently add support.
One pain point I’ve run into with TNS is that access to Docker or Kubernetes seems to be intentionally locked down from access anywhere but the built in apps catalog. As someone who works with Docker and various orchestration engines professionally, I much prefer being able to define and stand up my own services to using a list of predefined templates. There are obviously ways of getting around the restrictions in TNS, but with Unraid, I could install something like Portainer or simply drop into the terminal and run docker commands myself. Not having that is frustrating.
Overall though, I think TrueNAS is a much cleaner and more modern user experience, so long as you stay on their rails. Which I suppose is the point.
Truenas is great for creating the zfs pool but I hated that I couldn’t also use the machine as a computer without making a vm, so I ended up exporting my zfs pool and then importing it on a Linux system with cockpit installed to keep a Web interface, definitely recommend looking into it
That sounds interesting. Thanks for the recommendation!
I run my Truenas Scale with 5 mirror vdevs. This is sort of like raid 10 (I don’t need the differences explained to me). This means that I get 50% of the raw storage as usable but, it means that to upgrade space, I only need to upgrade two drives at a time. It also means that replacing a failed drive is fast, much faster than replacing a drive from a raidz* vdev. As you move to Truenas, this is something to consider. Given that you’re going to have 4 drives total, I don’t think you’d be wasting any additional space as you shouldn’t consider raidz safe (same problem as raid 5, high risk of second drive failure during rebuild) which leaves you at raidz2.
