Roi

Hi, the situation got much better. I noticed that the setting <cpu mode='host-passthrough' check='none' migratable='off'>from this guide was on again. I disabled it again. I also disabled CPU pinning, so all VMs have all CPU (v)cores now and I let Unraid handle everything. Somewhere in the guide linked above I also noticed that the Fuse filesystem could be a performance thing, too, so I changed the vdisk path from /mnt/user/domains/windows11 to /mnt/cache/domains/windows11. CPU percentage in Idle is now 3-5% and the VM is much more responsive. I was also able to successfully test Fortnite - one of the worst games when it comes to optimization. Regards, Martin

Hi everybody, I am running a Windows 11 VM on Unraid and I am pretty unhappy with the performance. First some screenshots to make myself a bit clearer and below the screenshot the specs and settings. This is how it looks a while after rebooting without anything running except drivers and tools (Mouse, Keyboard, Headset - I also want to use this VM for gaming): This is how it looks, if I move the mouse around on the screen a lot (drawing fast circles, if you like): And here are the processes running and creating this amount of memory and processor consumption: About specs and setup: The machine runs on a Ryzen 5 8600G with 64GB of DDR5 RAM. The virtual machines are on a 1TB NVMe M.2 SSD mirror (the cache device(s)), Windows got 6 (v)cores, the other 6 are for the Linux VM. Btw, the Linux VM, running on EndeavourOS and 24GB of RAM runs totally smooth, complete contrast to the Windows VM. I handed in a Nvidia 1080 graphics card into the Windows VM, also one of the USB controllers of the host. As mentioned, I also want to use it a bit for gaming but I also want to use it physically (display, keyboard, mouse, USB ports for sticks...) when sitting at my desk next to the host machine. I also handed in a SATA SSD drive (Datenträger 0 in the screenshots) for the games, but the performance issues were there also before I did this. First I was using i440fx, which is the default when creating a Windows 11 VM on Unraid. I changed this to Q35 today, but nothing changed when it comes to the performance issues. It is more or like the same. All old hardware is deleted from the Windows Device Manager, I also reinstalled the Virtio Drivers from the ISO file. When starting a bigger programm like a game it takes very very long. Also, when updating games (Steam wanted to update around 30GB today as not used longer) I noticed that the disk drive takes 100% in Windows Task Manager. I followed some tuning guides here in the forum (e.g. about energy management, hibernation, turning off indexing/Bitlocker (Bitlocker was off anyway), network device (changing virtio-net to virtio)...). The basic problems did not change. So what is wrong here? I cannot think of more which I can tune. So any input, ideas, anything, is very appreciated! Thanks, Martin

I experience the same problem, also with Unraid 7.0.1. Did you find anything out, yet? Maybe the mount options? I only use rw.

I recreated the Docker image file today. When I had the problem, I restarted the machine because the Docker system did not work anymore. After restarting the machine everything was fine again.

Thank you. I re-created the Docker image file in the morning and made it bigger (20 to 40GB). I do not have any custom networks for Docker, so no todo for me here. Will try the Appdata backup maybe tomorrow morning and see what's happening.

Thank you. Are you guessing because of what I described or do the diagnostics point to this?

Okay, here are the diagnostics. But it's from after the reboot. Not much sense in it? Or does it bring up information? I am willing to try again, if this is not useful and then I can get diagnostics before rebooting. sol-diagnostics-20250203-2200.zip

I ran into a problem last night when the Docker Engine did not start after Appdata Backup. Posted this into the Docker Engine subforum and now here as it was suggested to post here, too. backup.log

Thank you, okay, I will. But as the plugin only stops and starts the containers, it might not be plugin related. That was the reason I posted it here.

Hello all, I am using the Appdata Backup plugin for quite a while now and never had problems. That changed last night. In the morning I noticed that all containers were not running. I checked the GUI and did not see any Docker containers at all. But Docker was running. I stopped it through settings ("Enable Docker" to No, then Apply) and then enabled it again. On the Docker page I saw "Status: Running" but still no containers. Retried, no change. In syslog I saw something that no memory is left so I stopped two VMs and then also with 20% memory usage I still got the same error. So I rebooted the Unraid machine and everything came back online. The problem is that I also had to bring down the VMs (one of them is OPNsense and another Pi-hole, so no Internet...) and also rebooting and having everything back running takes some time. So that's quite a deal. As everything went offline at 3pm it looked to me already from the beginning of my investigation that it must be the weekly Appdata Backup which starts at 3am on Monday mornings. And yes, the log tells me that as well. The backup itself was okay, but then not all containers were started and it took a long time (one time over one hour) from one container start to the next which failed. I attached the log here. I also got a mail from Appdata Backup that something failed and that I need to check debug log. As I already rebooted (had to get out of the house to work and the services back online...) I do not see any logs anymore and I am also not sure if debug logging is enabled for Appdata Backup, I guess, it's not. I did not find debug in the plugin settings. Anyway, I installed some more Docker containers last week, so something changed here. And I also have lots of data inside these containers for backup, e.g. 5GB+ of zstd compressed MariaDB data. Do I need to adjust some settings? I never did for the Docker Engine. But is that really Docker as all containers get stopped before backup (I set it that way) and started again after backup? Any ideas? I'm glad to bring more data, I would also run a backup again and check debug first before rebooting. At the moment I hope for something simple, maybe a user problem with some setting I should have changed because of more containers, load or whatever. Thanks. backup.log

Ehm... No... Thank you! Temporary fix applied...

I am seeing this for two days now. The container seems to be updates. So this must be a new problem. borgmatic 1.7.5 apprise 1.2.0 crond: crond (busybox 1.35.0) started, log level 8 crond: USER root pid 14 cmd borgmatic -c /etc/borgmatic.d/config.yaml 2>&1 Traceback (most recent call last): File "/usr/local/lib/python3.10/site-packages/borg/archiver.py", line 38, in <module> from . import helpers File "/usr/local/lib/python3.10/site-packages/borg/helpers/__init__.py", line 9, in <module> from .checks import * # NOQA File "/usr/local/lib/python3.10/site-packages/borg/helpers/checks.py", line 4, in <module> from .errors import Error File "/usr/local/lib/python3.10/site-packages/borg/helpers/errors.py", line 3, in <module> import borg.crypto.low_level ImportError: Error loading shared library libcrypto.so.1.1: No such file or directory (needed by /usr/local/lib/python3.10/site-packages/borg/crypto/low_level.cpython-310-x86_64-linux-gnu.so) /etc/borgmatic.d/config.yaml: Error getting local Borg version Command '('borg', '--version')' returned non-zero exit status 2. /etc/borgmatic.d/config.yaml: Error running configuration file summary: /etc/borgmatic.d/config.yaml: Error running configuration file /etc/borgmatic.d/config.yaml: Error getting local Borg version Command '('borg', '--version')' returned non-zero exit status 2. Need some help? https://torsion.org/borgmatic/#issues

Yes. But I cannot say if it is the Unraid upgrade or the reboot since migrating from NerdPack to NerdTools.

I run 6.11.2 since today. NerdTools ist the most current version from November 1st. Yesterday borgbackup still worked on 6.11.1 - maybe there was also another dependency which supplied libffi.so.7 here, too?

Well, borgbackup is included and does not need to be included. It was also included in NerdPack. But now it does not work anymore. I think that is a different topic then.