I ended up on shucking 4x the 14 TB WD Elements Desktop. They contain helium drives, the WD140EDGZ in my case, and are about a third cheaper than 4x the 12 TB WD Red Plus drives (which are air-filled). The shucking was easier than I expected too, and the performance seems very comparable. The warranty is a definite downside (European, so no Magnuson-Moss), but I think I can even get them back in their enclosure should they fail during the 2-year warranty period.

I've put some second hand 256 GB M.2 SSDs in there as boot drives. It was a bit of a struggle to get it to work in a way that failure of one of the drives doesn't hold up booting, combined with LUKS, TPM keys and ZFS on root. Learned a lot about systemd-boot which I have never used before, but feels a lot saner to me than grub ever was. So now I have a large script which debootstraps a Debian based NAS into being.

I noticed that there are a lot of ZFS myths and cargo culting. For example TFA mentions ECC RAM, which in some circles is a must-have because ZFS would wreck your pool during a scrub otherwise, which is a myth. It's also very expensive, especially this year. You also don't need much RAM for ZFS, L2ARC doesn't use much RAM at all, to name a few others.

Still doubting about setting `dnodesize=auto` (which is the default), because there are some horror stories about that [1]. And it seems impossible to find a cloud storage provider with reasonable prices that supports `zfs send`. Rsync.net upped their minimum order to 10 TiB recently, which is far too much for my use case.

[1] https://github.com/openzfs/zfs/issues/11353

[2] https://www.rsync.net/products/zfsintro.html

Install avahi-daemon. Samba will automatically register with it to advertise SMB/CIFS to macOS and Linux clients over DNS-SD.

Install wsdd2 so that your server will be auto-discovered by Explorer on Windows 10+ clients with SMB 1.0 disabled, too.

Your Linux hostname is probably lower-case, but by default, Samba publishes a capitalized rendering of the hostname to NetBIOS and Avahi. If this bothers you, set “host-name=something” in the [server] section of /etc/avahi/avahi-daemon.conf, and set “mdns name = mdns” in the [global] section of /etc/samba/smb.conf.

If you have macOS clients, you should enable vfs_fruit in your Samba configuration: https://www.samba.org/samba/docs/current/man-html/vfs_fruit..... There are some compatibility reasons to do this, but mostly it means you can set the “fruit:model” so that your server has a fun icon in the Finder sidebar.

To avoid the creation .DS_Store files, you can disallow them: https://ryanoberto.github.io/blog/2015/04/01/disabling-the-c.... I think you can also set “fruit:resource = xattr” to store Finder preferences in xattrs, but I haven't tried it.

Although macOS deprecated AFP in favour of SMB years ago (and are slated to remove AFP client support altogether in the upcoming macOS 27), SMB client support in macOS is still pretty miserable. The upcoming macOS 27 is set to drop AFP support, but until then I will continue to run Netatalk side-by-side with Samba. Netatalk also registers itself with Avahi, and macOS will (tellingly) use AFP preferentially to SMB, so clients will talk to the right daemon automatically.

And that stuff is often not just more expensive, but uglier and noisier. I ended up making my own “enclosure”

Sure, I traded the convenience of kitchen sink (or Swiss army knife for the more charitable souls) ZFS for some initial pain, but I'm very happy with my choice today.

https://world-playground-deceit.net/blog/2025/06/nas-setup-l...

Is here okay?

I want to have physical storage over here, and logical storage over there, and I want to control the mapping from one to the other. I want to talk about encryption, replication, latency. Make "time machine" available on this logical storage. And then I buy some new physical storage, and it joins the story. This physical storage is over at my friend's house for backup. This physical storage is slow and archival. This is a device for writing archival media, and there's a brand new media in it, go ahead and write to it. Show me the health report on all of the physical media, and show me what you've done to protect the logical storage. Graph my usage and make suggestions about when to add more physical storage.

Buying physical storage with power and wifi, and configuring it with a QR code that's on an e-ink display - seems like it should be the most obvious thing in the world, that we should all be really used to doing by now.

What am I missing?

I try to keep my network configuration restrictive by default, so I'm not too concerned about possible security arguments running it from my main machine. I've probably committed some great sin here, but is plain NFSv3 and secondarily Samba (for compatibility) really not enough?

(And I know I have to do that, because when the disk fails it beeps and lights a led near the bad disk)

It’s easy to build a NAS such as the one described in this article, but in the long run, data loss is significantly more likely.

Also, any guide like this that doesn’t guide you through “disk 3 failed, this is how you safely replace it” is imho incomplete, even if it doesn’t go through telling you how you know a disk has failed.

It always gets me how the world of self-hosting is usually introduced from claiming that you can start giving a second life to a Raspberry Pi or a forgotten laptop, and suddenly the next blog you read calls "minimal" a beast machine meant for racks and semi- or professional environments.

Bought a ThinkCentre M910q with an internal SSD and 16 GB of RAM for €200 a couple years ago... Right now I got it chugging along with TrueNAS + 2 USB disks in ZFS Mirror (sitting in a closed cupboard so no chance of cable disconnections).

For me, "minimal home server" means a small computer that fits on a cabinet on the living room, is practically silent, and has a very small power consumption profile (less than a decent Hetzner otherwise the cost wouldn't be worth it). I have a mini-PC in mind, but people think of Dell PowerEdges. Even if given for free, I would never install at my home a PowerEdge for a home server.

I guess it must be the difference between living on a flat vs a 2 story house :-)

Building a NAS from scratch is really fun! A small hinderance, but definitely part of the fun as well, is the lack of a "complete resource" on the topic covering how to do every single thing you need to do. Part of the point of my blog post is actually to bring to the internet yet another opinionated NAS setup "guide" (eventhough I would hesitate to call it a guide, but if I ever had to do the same thing again I would definitely read my own post first).

> I am creating a RAIDZ1 (RAID 5) zpool. That means 1 drive redundancy in-case of failure

A friend once told me that RAID5 has a high latency cost, because every Write requires a Read to update the stripes across all drives, and while this made sense when drives were expensive, nowadays you might as well do a RAID10 instead, and trade space for latency.

Is this still true with ZFS RAIDZ1?

I am thinking of buying a USB bay with 5 SSD slots in it and then 3 HDD drive. My use case is very, very cold. It is mostly just readonly data with some rclone sync every week.

Does anyone have a suggestion? I am pretty much relying on ZFS to do all the redundancy for me.

Caveat: LVM does not have built-in snapshots like ZFS has.

I don't know much about ZFS, but it sounds like I need to learn. Docker may have conquered the world, but I plan to stay with LXD for services.

The one thing I take issue with: an appliance like this runs 24/7. It should be low power and fanless. A processor like the N100 seems like the obvious choice.

Used mergerfs and snapraid, and a simple NFS share. Absolutely perfect for Proxmox backups, our pictures, media, etc, etc. No fuss, easy replacement of drives without needing to keep drives of the same size around.

I'm good with ZFS, have been for years now managing storage for $COMPANY. And I still freaking love using zfs send/receive with proxmox ;-)

But for most at home stuff: mergerfs and snapraid are just more logical.

I had a cloud backup (Backblaze B2, using rclone and encryption) for my home NAS. Some unlucky drive failures later, I was getting ready to recover from my Backblaze backup... and then I couldn't find/remember where I saved the rclone encryption backup.

I lost all data in that NAS, including irreplaceable personal/family photos, due to that mistake. My lesson to share: please verify you can indeed recover from backups.

It's a set and forget OS that will run for years without requiring your attention. But these days it has decent container support for hosting services on.

Also, are there neal.* besides .fun and .computer?

I'll check all the other TLDs real quick...

For some reason people insist on doing truenas on top of proxmox and then introducing a networking layer between everything they do. Noooo…

unraid.net

or proxmox ( especially community edition) is fucked or zfs is still unstable, surprisingly with different drives on different interfaces I obtained same results, didnt swap proxmox for clean os, that might being some changes since modules and options would be different i guess. ( drives firmware etc are fine and performing well when inspected with proprietary windows tools)

just search zfs nvme pcie or usb problema or disconnect and you see similar stability problem for different users in different cases (os, drives etc), but also unraid and others, maybe somewhere is rithe right combo of options/glitches but didnt find it yet

Running ZFS on anything but Solaris/Illumos/FreeBSD is asinine.

ZFS is a permanent second-class citizen on Linux (due to usual open-source politics). This will never resolve.

I don't want to trust my data to some half-assed out-of-tree solution that may or may not break in a week.

FreeBSD ZFS support has matured and is outstanding. Quality-wise it has reached parity with Illumos.

If you can afford Solaris then you're probably not building your own NAS from parts of lesser computers.