testdasi

Sorry but I disagree. LT has always been so often too nice and bowed down to pressure from the minority that happens to scream really loudly about niche issues. e.g. new GUI too big / small and certain people refused to use their browsers' native zoom functionality and demanded very loudly that LT changed everything back to the way they were familiar with - and guess what, LT spent time and resource responding to these GUI supremacists while ignoring the fact that Gigabyte X399 users (e.g. me) have severe lags that to-date still have not been resolved. So it's rather refreshing to hear LT dev(s) get the balls to respond to loudmouth screamers in a different way. Donald Trump and Boris Johnson come to power because people are too nice to scream back at them.

Hard links can's span physical volume. Potentially you have a in 1 disk and b in another disk.

You seem to be pushing hard to justify the X470, in which case, why ask questions? To use the 2nd M.2 slot, the bottom PCIe slot will be disabled. So you are back to the same problem i.e. have no expansion. So is the X470 "enough"? It's certainly not with your requirement to have further future expansion. Also, usually the 2nd M.2 slot is connected to the chipset with means it can't be separated from the other devices without ACS Override - which may or may not cause problems. So even the X570 may not allow you to pass through both NVMe's to VM's. There's no way to ascertain the exact layout and IOMMU grouping without owning the actual board.

Yes it means something. The 1TB and the 840 have not been formatted. I remember there's a box somewhere to format unmountable drives but can't remember - have not had to format anything for quite some time now (since I use the Preclear plugin). You might want to post a question in the General Support forum (alternatively, watch SpaceInvaderOne's video - I'm pretty sure it's in the basic set up guide). Btw, if you have any data that you still want to recover from the 1TB and 840, now is the time.

Just our of curiosity, other than "I don't trust [Insert Cloud Provider]" and "I'm Linus Sebastian / Marques Brownlee", is there any particular reasons that warrant 100TB+ storage server?

My Win10 VM can run on 1 core 2 threads without lag (and my 2990WX has a lower base clock than your 7700K) so something else is definitely going on. Did you isolate the cores for the VM? Where is your vdisk image stored? Also, "mismatched sound as the video progresses" is a typical problem that msi_util will fix. Did you follow SpaceInvaderOne's guide to PCIe passthrough? The msi_util tool floats around the forum a lot. Below is 1 post that SIO attached the file.

Drives don't have to be wiped prior to being added to the array / cache pool. The fact that you have reached Parity Sync suggests the drives have all been formatted. Correct. DON'T assign into the array any drive that you intend to pass through. In fact, if you use the vfio stubbing method in SpaceInvaderOne's video, the NVMe drives should not even show up for you to assign.

Doesn't seem abnormal to me. Total is measured between processes and device and Actual is measured between device and hardware. So things like io cache will cause mismatch. Perhaps your qbittorrent is reading stuff directly from RAM.

The i3-2100 is dual core quad thread 3GHz. So 50%-90% on "some" of its (logical) cores (i.e. threads) does not scream upgrade to me (unless it's like that ALL the time). What VM / docker do you plan to use the server for? Given what you said about your use case (which needs further details), the 2700 is a massive overkill. No Plex = practically anything works so you need to elaborate a bit.

What sort of stuttering are you talking about? In game? In general usage? You haven't mentioned the typical load (e.g. how many Plex streams? how many VMs?) but your config shouldn't be stuttering in most typical home usage scenarios. So sounds to me like a configuration / underlying issue. Also mismatched sound is unlikely to be a CPU problem. If msi_util doesn't fix it then it's unfortunately unfixable without a new GPU.

You are asking a sweeping question on a highly "it depends" matter. For example, let's talk Plex: The typical passmark recommendation is a generous generalisation of how much processing power you need to transcode. Because there are so many possible configurations out there, providing an "overall speed" rec is the least specific way to make sure 99% of users won't be unexpectedly disappointed. Throw in the "right" combination (in my case, PGS subtitle on 4k HEVC mkv) and no amount of passmark will be sufficient because Plex only use 1 single core to transcode. As for VM, it's just like normal barebone PCs. A gaming PC should emphasize more on single-core performance while an AI home-lab prefer more cores and a video-editing workstation needs a carefully balanced middle ground. And that's without considering available instruction sets from different CPU generation / manufacturer. So in short, to answer your question, you need to provide details on what sort of use you are doing. "dockers and VMs" is like saying you are using a car to commute to work. My commute can do with a Nissan hatchback. My friend's dad commute needs a proper Land Rover off-road. Also in terms of hyper-threading, the passmark already take that into account. So when we refer to "cores" we typically mean physical cores. The only context when logical core is relevant is when assigning them to docker / VM which shouldn't be yet relevant to your queries.

You can only ssh using root. In terms of getting kicked out, it may be a connection issue - can you log in from the Unraid console? Check you syslog too, does it show a record of root logging in?

Do you happen to have multiple-drive cache pool? It will only do btrfs on the cache pool if you select more than 1 in the drop down, even if you actually only assign 1 device.

In that case you need X570. Based on spec, the X470 can't handle 2 4x NVMe.

Answers: I understand that I will need to add the Sata & HDD to the system. Does the Sata plug that each connect to matter? Meaning, does the Sata SSD need to be on plug one... ect? Does not matter. One of NVME drives has my current Win10 64 Pro build on it. The other has my wifes VM machine (runs on Vmware Workstation 15 Pro but the FPS in game is just aweful) Will I need to just format both drives and reinstall through UNRAID via the server host software? No need to format. You can pass through the NVMe to the VM as a PCIe device (i.e. just like a GPU - you will need to vfio-stub it - instructions are in SpaceInvaderOne videos) and it should just boot (in which case you don't need a VM vdisk and certainly don't need to format it). (Note: your VM should boot in OVMF mode and not SeaBIOS) Even if it doesn't boot and you need a vdisk, you also don't need to format the NVMe disk if passed through. It should just appear to Windows the exact same way. You can even convert the vmdk vdisk of your wife's VM into qcow2 / raw format to use for Unraid VM (still have to set up your VM template - can't convert that). I vaguely remember vmdk being supported directly, just need to edit the xml tag = vmdk but I have never used it. While talking about format - your HDD in the array and the SSD in the cache pool will be formatted by Unraid so make sure you backup your data. Probably best way is to just copy stuff from the 850 and 1TB to the 2TB and keep the 2TB safe and outside of the case so there is zero chance of accidental format. Also, I have 3 physical computers at my house but they're not networked to mine, though they all connect to the same wireless cable router/modem. What would I have to do to access the server once its running? If they are connected to the same router then shouldn't they be on the same network? What do you mean by "not networked to mine"? The others can speak about the USB stick since I have not had any problem - and I can't possibly do any worse since I use a micro stick AND USB 3.0, all sort of "not recommended" stuff. 😅

Quoting google re SMART error 187 (underline added by me) So that error + you have a corrupted file on the same drive suggest a reasonable cause to ask for an RMA. If I were you, I would ask for an RMA and driving on the point that a corrupted file was found on the drive and no other drives in your server have any issue. Manufacturers typically don't like to argue with customers complaining about lost data so it shouldn't be too much of a fuss.

The executable and configuration files are in the USB stick which it will boot from. Then things get copied into RAM and run out of RAM. The USB stick only provide the functionalities, the "whole system" isn't on it e.g. the storage array / cache pool, VM disks, docker image etc. are not on the USB stick. In Windows terms, it's like booting into Windows on a USB stick - everything else, including apps, will have to be stored somewhere else. See below for longer answer See below for longer answer So what you need to do, I'll break it into stages for you. Stage 1 - buy stuff Get USB stick (with unique GUID - so don't buy cheap), preferably USB 2.0 stick but USB 3.0 is fine. You don't state you have something for storage other than the 2 NVMe drives so get a HDD to be used for the array. You need at least 1 drive in the array, even if you are using your server purely for VM i.e. no NAS storage requirement. Ideally you want 1 HDD in the array and 1 SSD (256GB SATA would be the minimum in my opinion) in the cache pool. 1 more dedicated GPU - you only listed 1 dedicated GPU and you would want 2 for your 2 VMs. While it is possible in some cases to pass through the iGPU to a VM, the process is more complicated so my general recommendation is avoid it if possible. You MAY need 1 more USB PCIe card - depending on what your IOMMU device layout is. Stage 2 - feasibility assessment i.e. should things work for your use case in theory? Make sure all the necessary devices are plugged in - sounds redundant but I once spent an hour figuring out why my 970 isn't showing up when it was on the shelf behind me. Create the USB drive using UNRAID USB Flash Device creator. Don't enable UEFI boot - see below Go into BIOS, enable everything that is necessary (e.g. VT-x, VT-d, IOMMU etc.) Make sure to set your BIOS to boot in Legacy Mode (i.e. not UEFI). This will save you a lot of troubles in Stage 3 with Nvidia driver Error 43. Make sure to set your BIOS to use the iGPU as the primary GPU (and connect your (boot) monitor to the motherboard port to be absolutely sure Unraid is using the iGPU to boot. Boot into the USB stick, watch for the boot selection screen and select GUI mode (i.e. no need a 2nd PC in the network to configure stuff!). The default timer will boot into non-GUI mode. Preferably, use the USB 2.0 port on your motherboard if available. Go to Tools -> System Devices and review the PCI Devices and IOMMU Groups section to see if devices are in their own IOMMU groups. If you are not sure what to look for, you can copy-paste the section here - we usually can help have a look to see if things are more or less ok Please use the "Code" functionality on the forum (the </> button) when copy-pasting stuff from the GUI / code. It makes it a lot easier on the eyes to look for things. Stage 3 - doing it - assuming it looks like things should work in theory Watch SpaceInvaderOne videos on Youtube. They are incredibly useful and you won't need to spend more than an hour. For your use case, pay attention to his basic configuration + VM set up + PCIe pass-through + NVMe boot videos. I have never had any luck booting VM directly from NVMe drives but others have reported success so it should be possible. I still recommend creating a vdisk (100GB is more than sufficient) for the VM to boot from and use the NVMe for apps and games. (Hence, I said ideally you want an SSD in the cache pool to store the vdisk image). It just saves you unnecessary issues. Boot into GUI mode and do the configuration, following the Youtube instructions as closely as possible. If you have problems, post on the forum and ask for help.

It's GUI update timing - this has been like that for a while. I start 2 VMs and I can watch status outside of the GUI and everything is fine. The GUI just reads the status too soon. Also only affects "heavy" VM. My feather weight MS-DOS VM doesn't have this issue.

You need to use a VNC viewer app for full copy-paste support.

Did you edit the template using the GUI or the xml? If via GUI, did you accidentally change the format of your vdisk? (e.g. the GUI has the tendency to change to raw by default IIRC - if you use any other format e.g. qcow2, vmdk etc. then it can cause unexpected issue).

I don't think that's possible. You can't hot-switch a VM.

I was able to achieve what you are getting with only numatune set to strict e.g. <numatune> <memory mode='strict' nodeset='2'/> </numatune> I was able to do THIS every boot without numad 😅. My VM core assignment spreads across 2 nodes so consistent 50-50 split is the most ideal. Per-node process memory usage (in MBs) for PID 31136 (qemu-system-x86) Node 0 Node 2 Total --------------- --------------- --------------- Heap 0.02 0.04 0.05 Stack 0.02 0.02 0.05 Private 24412.36 24810.51 49222.87 ---------------- --------------- --------------- --------------- Total 24412.40 24810.57 49222.96

No I don't use SAS controller unfortunately so can't test that. I fully embrace the cloud now so don't need a big array anymore.

Do you happen to use ACS Override? I had severe lags in my VM moving from 6.5.3 to 6.6 that went away after I switched off ACS Override.

Yes, that should work. A few things to keep an eye for Make sure your 2-bay dock is mounting each disk as individual devices and not some weird JBOD or RAID nonsense. It shouldn't happen but check the manual just to be sure. I do cp just because I have always used cp and I know it works every time. Don't use Krusader (and mc) for large migration. I have found these more advanced tools to cause issues (e.g. Krusader grinding to a halt after a few hundred GBs, mc causing fragmentation etc.).