Taako

Okay cool, my current VM GPU is already bound to VFIO, just wanted to make sure that would be okay.

Currently i have just one Nvidia GPU on my unraid server and it's passed through to my windows VM. Tomorrow i'll be getting a second Nvidia GPU to add to the server which will be used for dockers that need a GPU for acceleration (such as ML, transcoding, etc). I see on the Nvidia-Driver plugin is it tells you "ATTENTION: If you plan to pass through your card to a VM don't install this plugin!" How do i handle this situation where i want 1 GPU passed through to a VM and 1 to be managed by Nvidia-Driver plugin for my dockers?

i think the issue is we aren't sure what other symptoms are happening that we aren't noticing. The pop-up may be indicative of *some* issue that we aren't seeing immediately. It would be nice to address it Also as an aside, Backblaze has a brand new Personal Backup tool coming out soon with (version 9.0), will this docker support that?

for what it's worth the pop-up doesn't seem to prevent it from backing up

maybe a permissions issue, make sure your user has read access to that contents. maybe SSH into your unraid server and check the permissions

i already have that step done and am still seeing the popup

im still seeing this pop-up frequently. Anyone know how to resolve? backups seem to be working for my D drive

I just restarted when it hung and it let me past EDIT: It seems it let me past but it did not work properly to back up. Instead i re-installed the container (deleting the appdata) and did not do the symbolic link until after it had gotten to the back up page. then i did the symbolic link and restarted the container. It took some time to scan the new drive so i left it running over night, now in the morning its backing up

It seems there is still *some* issue and im not sure where if its going to be problemtic or not. Whenever i start the server i get this error about permissions issue for C:/ProgramData/Backblaze/bzdata/bzreports Does anyone know how to resolve this? Docker run command is that same as i posted in my earlier comment

I went into the docker container itself and executed it. Click on the container icon, then click "Console". Execute within the window that pops up

I'm hitting the exact same issue... My docker run command is also for some reason its showing as drive Z not D EDIT: figured it out, i missed the installation step to do `ln -s /drive_d/ /config/wine/dosdevices/d:` in the docker. Now it sees drive D map directly!

[Backblaze Personal] Is there a way to mount different shares to different dives? I'd like each share to have its own drive letter in Personal Backup. Otherwise i have to just mount `/mnt/user/` to `drive_d` and there are shares i dont want to back up

Hi all, I'm tired of blackouts causing my unraid server to poweroff unexpectedly. I'm growing increasingly worried about data corruption too. I want to invest in a UPS (budget ~$250 or less) that will not only allow me to gracefully shut down on power loss, but gracefully auto-boot when power is restored (and ideally, but not required, the UPS has at least X% battery). I know you can use the BIOS to auto-boot when power is applied but I want to make sure the UPS can handle that. Ideally it would be great if the UPS only applied power until it has some set battery %. I was looking at UPSes and the Eaton 5S1500LCD looked like a winner (Eaton seems to get amazing reviews, not horror stories like with CyberPower or APC) but then i stumbled on this video and around 13:08 the reviewer says that if the computer attempts to boot the machine right after the ups regains power it will cause the computer to power cycle. Are there UPSes that handle auto-boot like i want? Do you have your machines auto-boot when power is restored? What UPS do you recommend? One thing that might work for me is just a raspberry pi to send Wake-On-Lan some time after power is restored. But i'd need to make sure the UPS is providing power to the machine by then. Do all UPSes have the ability to turn themselves back on and then provide power to the connected devices immediately when power comes back on?

So here's the story, I wanted to start using my i5-10400 iGPU for Jellyfin transcoding. I have a GTX 960 that is passed-through in the VFIO so i can use it with my Windows VM. Until now, my BIOS had its primary graphics adapter set to "auto", but after installing intel GPU Top and still not seeing /dev/dri i discovered that i need to set the iGPU as my primary GPU in the bios. So i went and changed it from auto to "onboard" and suddenly my unraid server kept bootlooping. After enough fiddling i discovered that I need to have my mobo hdmi port connected to a monitor when it boots in order for unraid to start, otherwise it bootloops. When my BIOS primary GPU was set to "auto" i was able to boot the server just fine with no monitors plugged in - in fact, i had to boot with no monitors plugged in, otherwise if i had a monitor plugged into my 960 when i booted, UnRaid would use that for the CLI and i could no longer use the 960 for my windows VM. What's the deal here? Why is unRaid bootlooping if i don't have a monitor in the iGPU slot now that it's the primary bios gpu? Is it my motherboards issue? I am using an AsRock B560M Pro4/ac mobo. Not sure if the issue is that the mobo wigs out when no monitor is detected, but it never had an issue with having no monitor back when it was set to auto (presumably using the 960). Is the solution here to buy an hdmi dummy?

Tried to install SeedSync but container wont start, only thing in the docker logs is this Traceback (most recent call last): File "/app/python/seedsync.py", line 378, in <module> seedsync = Seedsync() File "/app/python/seedsync.py", line 60, in __init__ config.to_file(self.config_path) File "/app/python/common/persist.py", line 53, in to_file with open(file_path, "w") as f: PermissionError: [Errno 13] Permission denied: '/config/settings.cfg' During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/app/python/seedsync.py", line 386, in <module> Seedsync.logger.exception("Caught exception") AttributeError: 'NoneType' object has no attribute 'exception' Traceback (most recent call last): File "/app/python/seedsync.py", line 378, in <module> seedsync = Seedsync() File "/app/python/seedsync.py", line 60, in __init__ config.to_file(self.config_path) File "/app/python/common/persist.py", line 53, in to_file with open(file_path, "w") as f: PermissionError: [Errno 13] Permission denied: '/config/settings.cfg' During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/app/python/seedsync.py", line 386, in <module> Seedsync.logger.exception("Caught exception") AttributeError: 'NoneType' object has no attribute 'exception' any help?

Got it! I was able to rescan and it worked. Now i have this issue: https://github.com/advplyr/audiobookshelf/issues/999

Absolutely amazing app, big thanks to the dev! I just set this up and restructured my unraid server audiobook directory structure in the format audiobookshelf wants (Author, Series, Book number with name). A bunch of my books are showing a red missing label, i cant click or hover over it to see what exact is missing. I open the details for it and use [quick-match to fill in what it can, but it still shows "missing" in red. What do i need to do to get rid of the missing label? How can i get it to tell me what it thinks is missing?

Can someone explain to a database noob why one might choose Redis vs Postgres? Does it matter, is one faster? What are the tradeeoffs? Which one is more likely to be used by other containers? I currently dont run any apps that require databases and want to install ghostfolio, but if i'm going to install a database i may as well install one thats supported by other containers that i may want to use in the future (like maybe HomeAssistant?) EDIT: It looks like i might want to install both as they seem to do slightly different things (cacheing vs database storage)

Well I got this USB controller: Inateck PCI-E to USB 3.0 (5 Ports) PCI Express Card and 15-Pin Power Connector https://a.co/d/8z92cz8 and I bound it to VFIO and passed it through to the and it did the trick! You were on the money with your conclusion because after it plugged the mayflash adapter in it worked flawlessly, no weird ghost inputs or any errors from the Dolphin Log Thanks a ton for the help! It's great having a dedicated USB controller too, makes hotplugging stuff dead simple!

Thanks! You've been a great help. I'll start looking for a USB pcie card and bind the whole thing to vfio. Maybe a separate question but do you think CPU pinning/isolation would help here or is that unlikely since our believe it's a host side issue emulating the device. I have the 4 cores (8threads) pinned for the VM (the other two cores (4 threads) are pinned to docker so they don't overlap. But I didn't isolate those 4 cores fro unRAID. So if I understand pinning/isolation correctly it means the VM is only using those 4 cores but the host can use all of the cores on the machine unless I isolate them

Just to confirm, since my knowledge of VMs and how they work is severely lacking. Will emulating the PCI/PCIe slot be less resource intensive than emulating the controller? I just want to avoid these dropouts Btw im using Q35-5.1 right now, i was using i440fx-5.1 when i initially created the VM but i switched to Q35 to see if it helped, it didn't. Not sure if i should switch back, everything seems to be working the exact same with Q35 and the machine type is (supposedly) newer. but that doesn't necessarily mean better. I just don't really know the pros/cons of each.

thanks for the info, i have an i5-10400 and am giving the VM 4 cores (8 threads), leaving 2 cores (4 threads) for unraid itself. The system has 48GB of DDR4-3200 RAM and only 16GB is going to the VM. I guess i can try reduce the number of cores/threads the VM gets but i dont wanna hamper it too much. If i get a PCIE usb controller and manage to bind it to the VFIO at boot (like i'm doing with the GPU) does this mean the KVM won't have to emulate that USB controller any longer?

I am trying to pass through a Mayflash 4-port Gamecube adapter to my windows 10 VM and it is working but im noticing the controllers sometime register a downward press on the joystick when playing games via the Dolphin emulator. This doesnt happen on other games that use the mayflash adapter (Slap City, via steam) and it doesnt happen on my windows 10 laptop with the same version of Dolphin. I did some searching and it seems like this is an issue caused when "something else is interfering with Dolphin, stealing the adapter, which makes Dolphin temporarily unable to read inputs." Right now i am not passing the entire USB controller since it seems like i cannot select it to bind to VFIO, and ACS override wont split it up even on "both" setting; so instead i just select the mayflash adapater USB device to pass to the windows VM. I am wondering if passing the device this way is causing it to not be fully owned by the windows VM and thus leading to the issue where something is interfering with it very momentarily, causing dolphin to not read in the inputs. Which apparently manifests as a "down" on the joystick... Do you think if i purchased a PCIE USB controller (so i could pass the whole thing to the windows vm via VFIO binding) that would resolve it?

I tried "both" in the ACS override options (as well as Downstream and Multifunction) but they still aren't separated out 😭 Is there any other way to get it separated?

I have a windows VM where the GPU is passed through. I selected it and bound to VFIO at boot. I rebooted my server and when it booted up it started using the monitor connected to the VFIO-bound GPU to display its startup logs. Then when i tried to start the VM it was stuck at the TianoCore logo. I shut down the server and unplugged the HDMI cable, then rebooted it, started the VM, and then plugged in the HDMI cable and it works normally. If my GPU is bound to VFIO at boot why does unraid see it and use it to display startup logs? It seems like thats causing it to be unusable for the VM