PsychoWards

Hey, Not sure if your 170Z is a typo and should be an AMD 1700 CPU. If it is, there is a bug concerning AMD Ryzen 1000-3000 CPUs which is causing exactly those random freezes. You have to disable all the C-states in the BIOS to stop this from happening. You could use something like uptime kuma to easily see if the server is still running or not, but you would need to install it on another server/pc and have it running there to get proper tracking.

What would be your intended use case for this? Opening the deck website, going to the apps section and clicking the launch button is not a way I imagine you would want to visit immich. 🙂 Sth like Heimdall would offer you this flexibility (and more) already.

Would be great if it would be like Truenas, you just export your config and during (or after) installation you just reload this config and it's all back and configured to how it was before. Maybe even give an advanced tab where you can select which categories/settings/apps you want to restore and which not. This backup could even be done automagically upon a settings change and keep a couple of versions, and be stored on one of the pools and Hexos might be looking for it during or after the setup and give you a choice of which version (automatic or manual backup and from which date/time) to use. Pool import is already working now, you don't have to create a new pool anymore during the setup, you import them afterwards.

This is not foreseen to be done in Truenas and therefor Hexos. Hexos is an appliance and you are not installing anything on the OS. The drivers are in the kernel and you cannot easily update them. Newer Versions might support your NIC in the future though.

Hey, Just something to consider a CPU with a TDP of 200W doesn't necessarily consume more then a CPU with 52W TDP when doing the same work. All it means is that it can consume more power during work (but therefore typically being faster than a lower TDP CPU of the same gen), but since the NAS is running at idle most of the time, due to advanced power saving features, there often is no or very little difference in idle power draw. Also, the Intel iGPUs (12th Gen Core i CPUS or newer) are very good at decoding/encoding for streaming with Jellyfin/Plex. If that's all you want to do, you cannot beat intel iGPUs, because a dedicated GPU adds additional (idle) power consumption. If however you want to do AI or Machine Learning things you do need a GPU with lots of VRAM, depending on what and how many things you what to run. You can also mix and match, take a dedicated GPU for the ML/AI things and the iGPU for streaming.

In addition to what @ubergeek and @Mobius said, never run RAM in single Channel (1 Module) always go for dual Channel (2 modules) or else you are leaving a lot of performance on the table. If you want to have 32GB RAM go for 2x16GB but not 1x32GB 🙂

Looks like a solid system. No dramatically oversized boot drive, Intel NIC, CMR HDDs (if you stay with the Red Plus variant, absolutely avoid SMR drives) Depends on what your needs are, but as a NAS with only plex and not to many streams in parallel, this should be up to the task. You can edit a post 5-10min after you posted it, after that you can no longer edit it.

You do have similar priced 8 TB CMR drives, although a bit more expensive: https://www.bestbuy.ca/en-ca/product/seagate-ironwolf-8tb-3-5-5400rpm-sata-desktop-internal-hard-drive-st8000vnz04/19186378 https://www.bestbuy.ca/en-ca/product/wd-red-plus-8tb-5640rpm-sata-internal-nas-hard-drive-wd80efpx/17935390 You might not notice anything now, but once SMR drives hit a certain capacity threshold you will notice it. You can make it work with SMR drives, but expect some weird performance issues down the road, just something to keep in mind. 🙂

Hey, Seems like a decent build and you got some nice value out of it. 🙂 Just a couple of remarks to help with some pitfalls: For Ryzen 1000-3000 CPUs you need to disable all the C States in the BIOS to get the server stable, with C states enabled it will freeze at random. Your Motherboard has a Realtek NIC, they are not supported by Hexos due to bad linux drivers, causing stability issues and causing data corruption. Get yourself an Intel NIC which is supported and working out of the box. Return the HDDs and get yourself some CMR drives, the Barracuda drives are SMR drives, which you really shouldn't use in a Raid. 🙂 It looks like your motherboard has 2x m.2 slots and only the 2nd one (M2_2) is sharing bandwidth with a SATA port. As long as you use only 1 M.2 SSD in the M2_1 slot, you can use all 4 SATA ports. If you start with 2 HDDs, Hexos will create a mirror which cannot be extended. If you want to be able to extend the pool in the future you need at least 3 HDDs now to create a RaidZ1.

Mirror: you have 2 drives, the data is stored in those 2 drives, they are exact replications of one another and you can lose 1 drive, you have the capacity of 1 drive. RaidZ1: you have at least 3 drives and data is stored on multiple drives, while 1 drive is being used for parity.You can lose 1 drive and have the capacity of all drives combined -1 drive RaidZ2/3: Similar to RaidZ1, but you can lose 2/3 drives and your capacity is that of all drives -2/3 drives Stripped: You have only 1 drive and if it dies everything is gone.

Please don't use an external USB enclosure, this is just asking for trouble, because very often they are not forwarding the drives serial number but are just giving a generic one, which means that Hexos cannot differentiate between drives and will not be able to create a pool or use those drives. Only stick with internal SATA ports and SATA expansion cards to avoid any trouble. A lot of people in here are using Jonsbo cases, which exists in various sizes to accommodate varying amount of drives. A couple of remarks to keep in mind: You need a dedicated drive just for the OS and you cannot use it for anything else. Don't waste a 1TB SSD on this, 16GB is all you need, so go for a low capacity one. With 2 drives Hexos will create a mirrored Pool which cannot be extended, if you want to extended your Pool later on you need 3 drives now to create a RaidZ1 which is extendable. If you go for the 2 drives now, 4 drives in total later, you will have 2 mirrored drives and only the capacity of 2 drives in total Stay away from SMR HDDs. If you decide to go custom built, please post your specs in here as there are some pitfalls which we can help you avoid. 🙂

Hey, With Ryzen 1000-3000 CPUs you have to disable all the C-states or else the server is constantly freezing, just something to keep in mind

That's impressive, glad that it's working as you expected it =D Is your pool 2 now RaidZ2? 🙂 How do you intend to copy the files over? If the old server is Hexos/Truenas I would recommend replication task via network or just chuck the old drives in the new server and do a replication tasks from one pool to the other. If it's none Hexos/Truenas then good luck 😁 nah just kidding rsync is also awesome and if you leave it running you should be done in a week 🙂 Do you have an UPS for your setup?

Hey, I'm not sure if I understand you question, but yes you will find a link to download the OS which you then need to install as a new OS. During the installation process you can choose to skip the creation of a new pool if you want to import an existing pool. This however has then to be done in the truenas GUI. As for Hexos, it only supports Truenas 24.10.2.X. 25.04 is currently not supported and will break some functions/displays in Hexos.

Hey, For Plex/Jellyfin the recommendation is generally to go with Intel CPUs (at least 12th gen) because Intel Quick Sync is really great for decoding/encoding, much better then AMD in fact. Some people would say to avoid 13th & 14th gen and to go for either 12th or Core Ultra (or how are they called again, the latest gen ?) due to the issues which 13 and 14th gen had/have. i5 or i7 are a good option for Intel. Also if you plan to use Proxmox, use it as the OS and not Hexos. You can host Hexos as a VM in Proxmox but you definitely should not host Proxmox as a VM in Hexos. This is a decision which you should take and already deploy now, because changing it afterwards is a lot of work and hassle.