I tried Unraid for the FIRST time in 2024

The moments like 9:09 with UNRAD instead of UNRAID makes me want to yell through my monitor and hope that you hear me 😂

A pro tip: since the license is linked to the USB drive serial number, get a good quality USB drive or even better a known-brand SDcard reader (it will use the serial of the reader, so even if the SDcard dies you don't need to call support to ask a license transfer)

UNRAID is less of a Proxmox competitor and more of a TruNAS one - they have bolted on containers and VMs [containers is more polished for sure].

So uhh, last time I checked (a couple months ago), SSDs are not recommended for array drives because the array does NOT support TRIM. You instead want to use a Pool, where you can use ZFS. The issue is that Unraid will not work without a single device in the array. The people that do this kind of setup either put a single virtual hard disk (if Unraid is virtualized), or a USB drive or some such as a single device in the array and everything else in a pool.

This is the type of content that makes the homelab community better. The good, the bad, the ugly. Been using unraid for a couple years now and love it. Also running a Proxmox server for my business for physical separation. Strengths and weaknesses in both, but great for my needs! Thank you for insights!

Hi! I made the bbergle-jellyfin template! I made it because I could never get hardware transcoding working with any other template. I suggest trying to download that container and try again! It should work lol.

19:00 yeah I agree with that, rather than having to add extra parameters, a checkbox to enable GPU passthrough in containers would be super convienent.

thank you for giving us the all in first time experience, successes plus failures and everything; it's not only me learning things the hard way. this is honestly quite helpful

I've been using unraid since 2017, the main reason for my using it was being able to slowly grow my storage over time instead of needing to buy drives in batches.

Cannot wait for all the upcoming unRaid tutorials. As a user of unRaid for a couple years now, I've not been able to find a use case that unRaid cannot fulfill. Excited to see your progress!

Super happy to see this video, I'm about to build my first server using UNRAID and love your content. Hope to see more!

I love unraid simply for the fact you use different sized drives like a jbod but parity like a normal raid.

The biggest thing that Unraid does (and you didn't mention) is that it has the ability to add drives to the array at a later time, without having to backup/reformat/recreate the array. If you "preclear" the drive before you add it and format it, you don't even need to do a parity rebuild. The other thing is that these drives are not being striped. Every file is intact on a given drive. It's literally JBOD with parity.

Thanks for sharing your experience. I've been running Unraid for a long time now, and I even have it in different remote locations, to do my backup and etc. Like most of your videos, I also tried Truenas, but unraid gives me the flexibility of upgrading my drives (as long as my parity is bigger than my upgrades). With truenas my experience of having all drives the same, feels like I am stuck with the drives, unless I have to upgrade, then I upgrade all the drives to be the same. Oh well, thanks for the demo and video, have a great memorial weekend.

Really like to see this. I did suggest you try unraid in one of your recent videos. Your opinion mostly matches my own. I keep a server running proxmox for VMs but I love my Unraid server for all my media & miscellaneous NAS storage. Its where all of my set and forget homelab stuff lives. Tinkering takes place somewhere else

One of us! One of us!!! Cool....definitely going to be liking seeing your "howto's" on UnRaid like Space Invader One and the like, once you get comfortable with it and start tinkering..plus your constructive critisizm might get more "features" built into the apps that are lacking (like your mentioned "metrics" for the docker containers [there may be an app for that though]) You will be amazed at how much there is...tons of CA (community apps ;) ) to poke thru Keep em coming!!!!

*waits to see what happens when he gets to passing through the GPU*. Basically you have to hit the "Advanced View" toggle in the container config, add --runtime=nvidia to the "Extra Parameters" section, add NVIDIA_VISIBLE_DEVICES key with value of the gpu identifier, and add NVIDIA_DRIVER_CAPABILITIES key with value "all" or "compute" (don't include quotes). After that things should pass through just fine. Then you can go to the container's console and do a nvidia-smi to confirm. The downside to using GPU in containers is that if you try to pass through that same GPU to a VM while containers are using it the system can lock up due to fighting for resources. So each time you want to use a VM that uses that GPU you will have to remove the --runtime=nvidia parameter from each container before booting up the VM. Unless you have more than one GPU of course.

Great take on Unraid! I really appreciate that you recognized the target user/use case for it and didn't bash on it because it didn't have all the "enterprise" features. A great follow up video would be testing out the ZFS implementation. I love the snapshot/replication ZFS adds.

When you start using Unraid you have to throw away some preconceptions about how things work with other systems. Once you learn the tricks, it becomes more apparent. Passing through a gpu to a VM or Docker, you have to install the driver (add the nvidia runtime plugin) and then also use the serial number of the card given by the plugin. Passing through a GPU to a VM isn't as hard as it seems either, if you have a motherboard with an onboard graphics (or cpu graphics) you can boot on that so the system itself has something, and then you pass through the (additional) GPU you installed separately itself, which is why IOMMU is so important. Up until the driver plugins were released it was a LOT harder. I have 4 unraid servers, 1 is a 48 bay 200+TB slicestor 1440 with a 32core EPYC server board with 8tb NVMe, 256 gb ram, and a P2000 for Plex. It's a monster and runs great. Another mITX 5TB of all-nvme storage AMD 3400g "test" server, one with a 6600K and 8 bays, and another 12-bay with an AMD 3900x, 6tb nvme, with nvidia P400 card. Two of those servers run on 10Gbit, 1 on 2.5, and the others on 1gbit. Pooling NICs is done in the software, not hardware. Nvidia, Intel, and AMD have different requirements to pass through GPU acceleration for Plex/Emby, but I actually have gotten it working with all combinations: Quicksync on the 6600k, AMD's gpu acceleration (yes, it does work), and 2 boxes using Nvidia. I will say: EMBY was a pain getting acceleration going. Plex is much easier. I have learned a LOT about Linux from Unraid, and I'm a Windows admin by day. I knew just enough to be dangerous with Linux before Unraid. Hint: Support configuration for HW accelleration depends on your video card type, and the nomenclature to access it is different. Nvidia: driver install, go to plugins, find the nvidia driver, and find the GPU-#### string, copy that and that is your "nvidia visible devices" number you need to paste in the docker (you may need to add it the Container variable NVIDIA_VISIBLE_DEVICES on Emby), and then use the --runtime=nvidia flag and restart the docker. That should fix your problem (it does with Plex). I gave up on Emby because it's still not quite user friendly to me on the encoding side. To get AMD and Intel quicksync working with Plex (and presumably Emby), you use a different flag: --device=\dev\dri, and you have to install the appropriate driver for the platform (Intel vs AMD vs Nvidia drivers), and create a device entry in the docker as well with a value that points to that "\dev\dri" even though you have it as a runtime option, you have to create the device for the runtime option to use. I think there's also another one instead of \dev\dri it's \dev\renderD128 that works (or worked in the past, I don't use Emby so I can't remember what one I used to get it to work). I was like you at the beginning, but after I did my research and played around with it, I was able to figure these things out and started treating Unraid like more of a compartmentalized system, and so passing hardware through is one of the most difficult things at first because you have to think of the OS it boots on, the dockers, and the VMs, and drivers/plugins being all separate entities from each other, and getting them talking is the part that gets people confused. Remember that Plugins will work whether docker or VMs are running or not (that's why drivers are plugins). I'm now a few years in and still learning all sorts of things as I go. Yes, Docker & VM snapshot/backup is done via CA plugins. CA = Community Applications, Dynamix = that refers to UNRAID themselves. Look into the following apps: Scrutiny, Unassigned Devices (THIS IS A MUST HAVE), Appdata Backup, and Netdata, and any of the CA & Dynamix plugins that may apply to you. BTW, as someone else has mentioned, go back and look at 9:09 and realize your mistake is there to see.

Nice video Tim! I bought a license a while back to eventually try UNRAID. I just haven't had the time to do so. My go to is Proxmox with a virtualized TrueNAS Scale VM.