testdasi

First check to make sure your vdisk file only exists under ONE of these possible paths: /mnt/cache/domains/001 - Windows 10/ /mnt/disk1/domains/001 - Windows 10/ /mnt/disk2/domains/001 - Windows 10/ ... and so on to the last disk Just ls each of the path. Does the vdisk exists under more than one path?

There are 2 things that were reported: High IOWait and/or server lockup when copying large files to BTRFS-formatted (mostly) Samsung SSD Inconsistent / low write speed when copying large files to BTRFS-formatted (mostly) Samsung SSD I have never had any of those issues and I have run 970 Evo, 860 Evo, 850 Evo and SM951 as (btrfs) cache at one point or another.

To make log persists between boot, go to Settings -> Syslog Server and turn on mirroring to flash. Also do a check disk of your USB stick.

What is your vdisk type? qcow2? raw? Are you sure you haven't changed vdisk type in the process when you changed the path? That is the most often issue with new user when changing vdisk on the VM template i.e. they don't know / check what vdisk type they are using.

I am running 860 Evo and 850 Evo with BTRFS and no issue. Dropped an 80GB archive on it and... still no issue.

LOL my script is more of a "Hulk! Smash!" approach. virsh shutdown VM1 wait 10s virsh start VM2 wait 10s virsh start VM2 wait 10s virsh start VM2 Since VM1 and VM2 has the same passed-through GPU, it would be impossible to start both VM at the same time. So if the 1st virsh start is before VM1 has been fully shutdown, it would just error out without doing anything and then the 2nd virsh start after 10s and then the 3rd one after another 10s. If VM2 already started, any subsequent virsh start would also error out so they don't do anything.

An enterprise build has features that you won't find on consumer build e.g. PSU monitoring, IPMI, intruder protection, encryption, etc. These features will be available on enterprise motherboards, which require enterprise CPU (e.g. Intel Xeon and AMD EPYC). Even when you compare an Intel Xeon and an Intel i7 of exactly the same feature-set, the Xeon CPU is more expensive just because of market segmentation i.e. you have less money than [insert the name of a corporation]. As for the motherboard, yes, go for the other 2 examples instead of the Aorus Elite as long as the feature set match your requirements. Picking a Gigabyte motherboard with 2 x16 slots is defeating the purpose. (X570) mobo with 3 x16 slots will have the 3rd slot running at x4 speed (i.e. slower). So if required, you can put a low-end GPU in this slower 3rd slot and save the 2x fast slots for other things. Mobo with 2 x16 slots only have fast slots. A vbios is likely required to pass through a single GPU to a VM. There are 2 methods to get a vbios (both methods require some Hex editing to work (watch SpaceInvader One Youtube videos) but that is beside the point): Dump your own, which requires a 2nd GPU to boot Unraid with before dumping Download from Techpowerup The vbios must match the exact brand, model and revision. That means the version on Techpowerup may not match your GPU exactly. Dumping your own is the only way to ensure it's the right one. Once you have dumped the vbios and create a VM that works with the passed through GPU, you may be able to remove the low-end one and run your server with just a single fast GPU. It depends on the exact hardware so it's impossible to tell in advance. So IF you need a low-end GPU for it to work, having the ability to use a slow PCIe slot for the low-end GPU for Unraid to boot with (and reserve the fast slot for the fast GPU that would be used for the VM), in addition to NOT needing to physically swap GPU slots, makes life easier. And that's why I said you can borrow the low-end from a friend but budget for "the possibility" of needing it. Also, you asked about "can Unraid boot". Unraid will boot with a single GPU. We are talking about passing that GPU to a VM. On a side note, that's also why it's usually easier to pass through a GPU on an Intel build with iGPU. The iGPU serves as the low-end so dumping vbios is trivial.

It depends on how risk averse you are. The fastest way requires that you are comfortable with being temporarily unprotected. Start new config (Tools -> New Config), and reassign all disks + assign the 10TB as a data disk (i.e. no parity i.e. not protected). The 10TB is likely to require to be formatted. Copy the content of each to-be-removed small HDD over to the 10TB and the existing 8TB. (since you said 1 new 8TB gives you sufficient space to do that, 1 new 10TB should definitely give you sufficient space). This should be faster due to no parity but don't do multiple parallel writes to the same HDD. Verify everything is copied correctly and remove the small HDDs and connect the new 12TB + new 8TB Start new config and reassign the old 8TB + 10TB as data. Preclear the new 12TB + new 8TB. Once preclear is done, assuming no error, stop array, assign new 12TB to parity and new 8TB to data. Parity should be rebuilt automatically when array starts.

Your math is wonky or your concept of significant is wonky. Even assuming worst case of 20 W = 0.02 kW = 0.02 kWh/hr = 0.48 kWh/day = 175.2 kWh / year = $21.02 / year Given your current server is at worst 8-year-old, you need at least 3 upgrade cycles to break-even (based on the CPU prices I can google). Having said that, if enterprise-level stability is a paramount concern to you then of course get the Xeon build.

I'm surprised by these questions given your 1k+ posts. Create vdisk on the mounted UD. UD mounts are under /mnt/disks/[mount name]. Mount name can be found (and changed) on the UD screen on the Main page. Don't preclear SSD. Poor thing just wasted write cycles.

Settings -> Syslog Server -> turn on mirroring to flash. Then next time it hangs, reboot and attach the syslog saved to flash. That's the only way to see the log of what happens shortly before the crash. PS: Avoid taking screenshots where possible e.g. you are better off copy-paste stuff to notepad and attach the text files to the forum post instead of taking screenshot.

Intel monopolised the CPU market, both enterprise and consumer, for decades until AMD came back with Ryzen. So that's why Intel is already generally more expensive than AMD. Then you are comparing AMD consumer CPU vs Intel Xeon line (i.e. enterprise clientele), of course the Xeon build is (a lot) more expensive. The reason why AMD has risen was because Ryzen offers better performance than Intel for cheaper. Also for your X570 motherboard, don't pick the one with only 2 x16 slots (i.e. the long ones). Go for the one with 3 x16 slots. It would be defeating the purpose to get Gigabyte motherboard for the Initial Display Output functionality but then cut off 1 of the possible 3 x16 slots (i.e. one that can be used for a GPU). And budget about $30-$50 more for the possibility of needing a low-end GPU for Unraid to boot with (or borrow it from a friend) so you can dump the vbios.

Watch SpaceInvader One vid. The link to the msi fix is in the vid description. Tha

Just thinking out side of the box, I remember on the other topic you said your GT710 works fine in a Windows VM. Volumeio github (https://github.com/volumio/Volumio2) says it has VirtualBox / VMWare VM image pre-built. Perhaps you can attempt a nested virtualisation? E.g. install VMWare Workstation Player on a Windows VM and try to make the VMWare VM image work?

What is the issue?

USB storage devices shouldn't be used on an on-going basis. Your unresponsiveness could be IOWait as the CPU waits for the USB device to respond and if the device is not working well, it would lead to very long wait. Given you are writing to it 24/7, the USB controller could very well have overheated and is on its way out (if not already effectively dead).

Tools -> Diagnostics -> attach zip file.

Don't bother trying to compile mergerfs-tools. Trapexit created the docker for Unraid just so we can compile the mergerfs binary precisely because it wasn't possible to compile the binary out of Unraid. What you can try is to just reboot. That should rebuild the latest mergerfs binary and hopefully it will resolve your problem.

I would suggest you keep the Drobo as the backup of what is on the new server. Given you have 16TB over presumably many years, you definitely want to have a backup. Just as a guide, remember the "Four-O": Original copy Online copy to protect against, for example, a drive failure Offline copy to protect against, for example, a power surge or a cryptovirus infection Offsite copy to protect against, for example, a house fire The diff between Offline and Online is whether the storage device is off + disconnected most of the time (e.g. an external HDD) or on / connected most of the time (e.g. an always-on NAS server). Offsite is generally the cloud but it could simply be an external HDD kept at a relative's place. So generally, your most critical data should have at least "Two-O" First O should be either (1) or (2). Second O (the backup) should be either (3) or (4). You only have "One-O" of your most critical data. That's not recommended. I have almost 5TB of data I consider most critical and they have the full "Four-O". In terms of costs for cloud backup, it's actually not that expensive e.g. Backblaze costs $60/year last I checked. Also, you don't have to be concerned about privacy. Data is encrypted or can be encrypted. Backblaze, for example, even offers Private Encryption Key, which makes it impossible for anyone but you to decrypt the files. Note: I'm not endorsing Backblaze in anyway and personally use Google (Business) Drive. I just use Backblaze as an example since they are a major service provider.

Use 4.2. 4.1 is buggy.

I read your post as "Hi, I'm Johnny English..." 🤣 I would pick option 3. I would recommend getting a Gigabyte motherboard instead of Asus. Gigabyte BIOS has the Initial Display Output setting which gives you flexibility picking which GPU Unraid boots with, which in turns makes life easier to pass through the GTX. Now I don't understand your point about needing 2 servers for redundancy. What form of redundancy are you after here? Does your Drobo has a USB out (i.e. to use it like a USB external drive)? If so, that would be the fastest way to get data off it, assuming you are doing some forms of RAID with the Drobo.

I can understand the point about not wanting to build a rack unit. But what's wrong with 12TB drives? HDD fails in statistical patterns so you should be looking to go big, go few. The more drives you have, the more likely for you to have a failed drive. For a gaming build, especially a build that only has ONE VM, and even more especially with a high-end GPU, I would recommend to go Intel (single-die CPU). You will get more consistent frame rate with Intel than AMD due to the Zen architecture inherent properties (i.e. the CCX/CCD design).

It is impossible to say without knowing what your script is doing.

Because of licensing issues?

It's on the AMD website. You don't need an AMD GPU to download the driver. https://www.amd.com/en/support/previous-drivers/graphics/radeon-500-series/radeon-rx-500-series/radeon-rx-580