Wenn man deinen Links glauben darf, dann wird der Hardware-Watchdog direkt vom standardmäßigen "it87_wdt" Kernel-Mdoul unterstützt.
Anhand dessen wie du deinen Beitrag aufgebaut hast, hatte ich damit gerechnet, dass ein "out-of-tree" Kernel-Modul...
Vielleicht als kleine weiterführende Erklärung:
Normalerweise hat der Kernel Code anhand der PCI-ID eines Devices eine sehr genaue Vorstellung welcher Treiber für dieses Device geeignet und/oder vorgesehen ist.
Und wenn ein Treiber anhand der ID...
Wenn du die Grafiktreiber über ".force_probe" zwingen musst, dann weißt du schon wie "gut" die Unterstützung deiner Hardware im aktuellen Kernel ist.
In anderen Worten: Abwarten und Tee trinken.
The quoted part sounds very much like we have another GPU, that does not like to be reset during operation.
There is a specific quirk in the Linux Kernel source code, exactly for these kind of cases.
So, in other words:
You could build a test...
You (both sso and masato) told us nothing about what version of Proxmox you were running and how you configured the VM.
Additionally, nothing about what "Linux" you were booting (ancient kernel version?).
Should we get the good old crystal ball...
The main question is not "will this work", which we cannot judge with certainty, as too many factors are involved.
The real question is, why your statement is:
Technically speaking, your idea might work. With the emphasize on the word "might"...
Try building a test kernel with an added SW-quirk for the ASM1062, like described here:
https://forum.proxmox.com/threads/problems-with-pcie-passthrough-with-two-identical-devices.149003/post-828151
Just replace the device ID mentioned in the...
Your buzzwords are "S4/suspend_to_disk" for the virtualized OS and "hookscripts" (https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_hookscripts) for the Proxmox side. Have fun!
If you want to compare, then please also try 6.14.x on Proxmox, as it is easily installable as well. It might be, that the power-related behavior somehow changed with the kernel update.
Would you care to elaborate on which kernel version(s) you tried when running Ubuntu?
Also what did you do exactly (in the original Proxmox installation and now in the Ubuntu)? Did you tune the settings (e.g. "powertop --auto-tune"), or just...
You pass-through what you need and leave what you need on the host (!) out of the vfio-pci configuration. Should be as easy as that, basically. That means you define your added Intel A310 with it's PCI-ID for vfio-pci, while you just do nothing...
Für einen reinen Fileserver (zum Beispiel) brauche ich keine Virtualisierung, bzw. der Overhead der dafür anfällt ist völlig unnütz.
Also entweder es gibt irgendeine Anforderung, die Virtualisierung wirklich erfordert, oder man macht sich nochmal...
Die Schnittstelle hat mit der Haltbarkeit erstmal gar nix zu tun.
Auch eMMCs gibt es Industriequalität, die sowas locker abkönnten. Nur logischerweise verbaut Ugreen solche nicht, da das niemand zahlen wollen würde.
Worauf es letztlich ankommt...
Ich hätte da eine Theorie:
Laut deinen bisherigen Beiträgen sind "485.3M" übrig/ungenutzt.
Das enspricht nach Umrechnung so ziemlich genau 473 MiB.
Also (( 485.3M * 1000 ) / 1024).
Und siehe da, das Laufwerk meldet eine Größe von "238.47 GiB"...
This is one of the best examples for showcasing the mindset, that you often see from people coming from the Microsoft (paid-software) realm.
You see some sense of entitlement for a certain function, that one is looking for. And if that function...
For the question of distribution for testing:
Due to the huge number of users, the availablity of a mainline linux kernel repository and the fact, that Proxmox actually uses a adjusted Ubuntu kernel, Ubuntu is what I would choose.
Just use the...
As already mentioned by others, passthrough is by no means a trivial topic.
In addition, you are using a consumer GPU, for a task that usually (besides the home tech enthusiast, homelabber or whatever crowd) only used in a professional setting...