I've been using unraid for a few years. Super happy with it. Recently migrated from using their normal array to zfs since I got a hold of some enterprise SAS drives.
Been using unraid for a couple of years now also, and really enjoying it.
Previously I was using ESXi and OMV, but I like how complete Unraid feels as a solution in itself.
I like how Unraid has integrated support for spinning up VMs and docker containers, with UI integration for those things.
I also like how Unraid's fuse filesystem lets me build an array from disks of mismatched capacities, and arbitrarily expand it. I'm running two servers so I can mirror data for backup, and it was much more cost effective that I could keep some of the disks I already had rather than buy all-new.
Fuse file system? I've never heard about it. Is it a proper file system or does it work on ZFS/others?
The clue with Unraid is in the name. The goal was all about having a fileserver with many of the benefits of RAID, but without actually using RAID.
For this purpose, Fuse is a virtual filesystem which brings together files from multiple physical disks into a single view.
Each disk in an Unraid system just uses a normal single-disk filesystem on the disk itself, and Unraid distributes new files to whichever disk has space, yet to the user they are presented as a single volume (you can also see raw disk contents and manually move data between disks if you want to - the fused view and raw views are just different mounts in the filesystem)
This is how Unraid allows for easily adding new drives of any size without a rebuild, but still allows for failure of a single disk by having a parity disk - as long as the parity is at least as large as the biggest data disk.
Unraid have also now added ZFS zpool capability and as a user you have the choice over which sort of array you want - Unraid or ZFS.
Unraid is absolutely not targeted at enterprise where a full RAID makes more sense. It's targeted at home-lab type users, where the ease of operation and ability to expand over time are selling points.
Not sure if it's obvious from this comment, but also worth pointing out to folks learning about unraid that it still has parity drives that let you recover from disk failures - it's not just JBOD.
Yup, my comment mentions the parity disk :)
Good to emphasise that a bit more though.
Bruh my reading comprehension....
So it's like SnapRaid and MergerFS. Got it. Thanks
FUSE is a "virtual" filesystem that can be used to make anything look like a local filesystem.
Example, encfs uses it to make an encrypted file tree appear decrypted to the user and performs encryption/decryption as needed on the fly.
Another example, Borg Backup uses it to let you browse backup snapshots as normal files even when they've been compressed, encrypted and/or deduplicated with other snapshots.
The Fuse file system isn’t the actual file system on each disk, it’s like an overlay that brings together the file systems on each disk. Like a super basic setup assuming you had two data drives is that you could have a share called “MyFile Share” and have two files in the share, “File A” located at /mnt/disk1/MyFileShare/FileA and “File B” located at /mnt/disk2/MyFileShare/FileB but if you connect to the file share remotely or do an ls on /mnt/user/MyFileShare/ you’ll see both files A and B. So you create each share and it’ll distribute all your files across disks according to your specifications.
This is the fuse file system, and it’s how UnRAID implements the “RAID-like” features. Because unlike actual RAID your files aren’t striped across the array, each file lives on one disk. So while you can have 1 or 2 parity drives that can rebuild your array in the case of a lost drive, unlike RAID you don’t lose your entire array. If you have one parity drive and a 5 disk array and lose two data drives, your parity can’t rebuild the lost data but the data that’s on the other 3 disks are still accessible.
What RAIDz are you using? How are you feeling it?
raidz1. No issues so far. I've had some prior experience with zfs from work, so moving to it was a no brainer.
I use proxmox with truenas scale. It's a great option, but you just have to make sure to pass the hdd controller PCI device through to the VM. This can either be the SATA controller on the motherboard if you can make that work, or a separate PCIe HBA.
Should I pass the whole controller? I have the Proxmox disk on the same controller, how do I do that?
You would need a separate controller.
Ouch...got it. Thanks
The LSI 9210 8i - IT mode is a great option and can be had for under $50.
What's the advantage to using proxmox and virtualizing TrueNAS in your use case?
I'm looking to setup a TrueNAS box mostly as a file server (I have a bunch of spare drives sitting around, so I can duplicate locally and then backup to a cloud provider), but also as a docker host.
(I'm also researching some setups for friends businesses with the VMware debacle - they have a year to migrate).
I have a lot of services. I use Ansible to manage many of them, so they're all in one VM. I use Home Assistant, which works best when installed as a whole VM or on bare metal. For the remaining services that I have yet to set up with Ansible, I keep the services that need the GPU on one VM, and everything else on another. Finally, I have an LXC container that is my SSH entrypoint and Ansible management system.
I could technically use TrueNAS Scale as a hypervisor for all this, but Proxmox has a lot of quality-of-life features that make it a better hypervisor. I could use Proxmox for ZFS and shares, but TrueNAS has has a lot of quality-of-life features that make it a much better NAS, so I virtualize it.
Honestly I would get a pcie sata controller card. It is fairly inexpensive and doesn't tie up the controller on the motherboard
Running a Debian Bookworm hypervisor using KVM/QEMU with virt-manager for vms + Incus for lxc containers gives you a lot of freedom with how you use it.
edit: It also means you build your own hypervisor from parts - kinda like installing postfix/dovecot/mariadb/spamassassin instead of a packaged solution like mail-in-a-box. It takes more time and effort but I find I understand the underlying technologies better afterwards.
That's something I like, just plain Debian with KVM. How is virt-manager compared to the Proxmox UI? Does it needs to be installed on a separate PC (Windows?)?
No problem running virt-manager on the hypervisor itself. You can also use https://cockpit-project.org with the cockpit-virtual-machines addon to manage kvm vms from https://machineip:9090
Cool! I think that I'll give it a go! Does the cockpit needs to be installed directly on the host OS or...?
Yeah, you need to install cockpit on any linux you wanna manage using it, then you can use the ssh keys to setup so your cockpit session on the hypervisor gives you access to your vms too.
I recently switched from Proxmox to Debian Bookworm with Incus(LXD fork) as my primary setup, it's been a pleasant experience. I also like the idea of using something like Cockpit to manage VMs though haven't come to a need yet for a VM over a container. I'll also point out that Incus can handle VMs as well.
Stéphane Graber, Project leader of Linux Containers is also on the fediverse and responds to questions often.
For the hypervisor I recommend either Proxmox or XCP-ng. XCP-ng is technically a better hypervisor, but I personally use Proxmox because I like the UI.
For the NAS OS, I use and recommend TrueNAS Scale. You can run Docker containers on it. All this being said, I've never used Unsaid so I don't know how they compare.
Out of curiosity, why move from OMV? I was thinking about trying it out for a second NAS.
Out of curiosity, why move from OMV? I was thinking about trying it out for a second NAS.
It's probably a problem of not dedicating enough time to learn how it works, but I've installed a couple of time in the past years, but...I don't like it much. It seems complicated to me (still, I probably don't dedicate it enough time).
For the NAS OS, I use and recommend TrueNAS Scale. You can run Docker containers on it
I'll check it, thanks
Your reason to move from OMV is the same as mine for XCP-ng. It is supposed to be a better hypervisor, but I just did not like the UI at all. It doesn't help that you have to host a small VM just to have a webGUI.
I had to build that small VM from source on a different machine because my XCP-ng install was refusing to set it up. Then I was able to move it over to the XCP-ng machine to self host.
I liked the UI and honestly I like the VM that keeps the web app separate from the hypervisor. But then it started looking like a massive pain to hotplug USB devices to the VMs, so I bounced.
I am trying Incus/LXD right now and enjoying it so far.
If you can - separate host and storage. Run what ever hyper visor you like - Xcp-ng is also good. Any nas is good
While you can do a lot of the stuff that PM does via Unraid and other tools, it's all there in one spot. I love taking snapshots before upgrades, migrating machines between nodes live while I upgrade the nodes, having HA for my OPNsense and other important boxes, and the PBS backup system. I know you could do all this with other tools, but it's damn convenient in PM and "just works".
You can install a NAS vm in PM, just give it raw access to the disks you're looking to use for data, and back them up independently. Don't try to do something like overlay ZFS on ZFS.
Yeah I experimented with Truenas in a VM, it randomly dropped the pool. Do not do this.
Nixos ❄️
I recently tried out Cockpit on top of plain old Debian and it was really nice. You can manage VMs and whatnot, but it's quite a bit more lightweight than Proxmox IMO.
You also could try OpenStack
Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I've seen in this thread:
| Fewer Letters | More Letters |
|---|---|
| ESXi | VMWare virtual machine hypervisor |
| HA | Home Assistant automation software |
| ~ | High Availability |
| LXC | Linux Containers |
| NAS | Network-Attached Storage |
| PCIe | Peripheral Component Interconnect Express |
| RAID | Redundant Array of Independent Disks for mass storage |
| SATA | Serial AT Attachment interface for mass storage |
| SSH | Secure Shell for remote terminal access |
| ZFS | Solaris/Linux filesystem focusing on data integrity |
9 acronyms in this thread; the most compressed thread commented on today has 15 acronyms.
[Thread #502 for this sub, first seen 11th Feb 2024, 09:35] [FAQ] [Full list] [Contact] [Source code]
What's your thoughts on ZFS? If that's something you want to do you could go buy a pcie Sata card and then pass it though to a TrueNAS VM.
Have you considered TrueNAS Scale?
A place to share alternatives to popular online services that can be self-hosted without giving up privacy or locking you into a service you don't control.
Rules:
Be civil: we're here to support and learn from one another. Insults won't be tolerated. Flame wars are frowned upon.
No spam posting.
Posts have to be centered around self-hosting. There are other communities for discussing hardware or home computing. If it's not obvious why your post topic revolves around selfhosting, please include details to make it clear.
Don't duplicate the full text of your blog or github here. Just post the link for folks to click.
Submission headline should match the article title (don’t cherry-pick information from the title to fit your agenda).
No trolling.
Resources:
Any issues on the community? Report it using the report flag.
Questions? DM the mods!