Yes, we'll update the doc

Also.. Thank you for taking a look at it and solving it so quickly, @florent It is a quite nice interaction when reporting bugs that they are looked at and triaged. Not saying everything needs to be a priority 1. But I do appreciate being close to the process. And if I can help make things better by reporting bugs, in a hopefully constructive way. Then I'll continue to contribute my way.

@Andrew no way to fix it without hardware upgrade? I experimented with TCS, try to switch to it from xen clock and disable emulation (vm kernel disable it as it unstable clock source), but no luck

@henri9813 Ahh alright, I understand now! Thanks for clarifying.

@florent I tried to upload it with the "Upload button" and got this error" [2026-02-05T15_03_26.704Z - backup NG.json](Invalid file type. Allowed types are: .png, .jpg, .bmp, .txt, .diff, .patch, .jpeg) " 2026-02-05T15_03_26.704Z - backup NG_json.txt

Ahh excellent news. So indeed the new kernel sees the device USB policy will be improved in the future so it's kept after updates, so doing it once will be enough. Thanks for testing!

@dinhngtu I think it just have been a bad migration or the Win 11 Template did other things to the windows 10 os. Since new test vm did not show the time issues, and a fresh migration of the vim with correct template selected did not have the issues either. I dont know but maybe there is a way to put detect the correct VM template or make note. I didnt think using the win 11 template would have caused issues since using the "wrong" template on vm creation does not have the same effect.

@dinhngtu tried on a Windows Server 2025 VM, UEFI BIOS working well, no reboot needed

@Pilow tags can work and the path to them is much more succinct. Thanks!

@markxc could you tell us what endpoint you used ?

@nvs Hi there, we had a similar issue with on our 5900x boxes. We have about a twenty of these and they have been generally greats hosts for xcp loads. However, one of them was constantly rebooting. While testing the ram the temp jumped to 90/90. They are running in 1U boxes so I expect heat but this seemed silly. We ramped the fans to max but this did not help. We then used bios to drop down various cpu boost profiles and gadgets to lower the heat and re-applied paste to cpu.. so far so good.

Something to document @thomas-dkmt

It's not a problem with XO backups

Hi I tried again and failed with the same error message.. xe vm-export vm=c57a4225-3adb-b422-34ec-d2a23c6223c9 filename=/run/sr-mount/a7f6df31-412f-d5e9-4163-8af75b1df8ad/dlt-sin-mducv8.xva The server failed to handle your request, due to an internal error. The given message may give details useful for debugging the problem. message: Unix.Unix_error(Unix.EIO, "read", "") Hum in this last reproduction, I the issue happened 3hours after I triggered the export... I have a feeling this is an hardware issue here... Anyone any other idea ? Advice where to look at? [EDIT] I have manage to export another vm without issue. Best. Matth.

@olivierlambert so strange though. i have it enabled on 4 pools, that didn't propagte on hosts. still VM boots up when not needed as explained on fist post of the thread. i'll try différent combinations to see what's really going on

To elaborate: it was already in progress, but delayed due to 1. the fact that I underestimated the impact (I thought we'd overwrite the file only when we'd modify it, not every update the package) and 2. lots of other priorities keeping the team busy. But it's clearly the reminder, and the help in better diagnosing it, that allowed to have it re-prioritized and (partly, we've mostly bought time to fix it properly, by simply avoiding to touch the file at all at the moment) fixed so fast.

I am running NVIDIA P40s on a DELL R720/R730 and a NVIDIA A2000 12GB on a DELL Optiplex 9010 in my playlab. Getting the GPUs to be accessible for VMs was quite easy: shutdown the XCP-ng host, install the GPUs, start the host again, activate the GPU in the advanced settings of the host in the PCI Devices section (“Passthrough enabled”). The host will restart at this point after displaying a warning. That's it – the GPU can now be assigned to a VM in the VMs advanced settings. I do use the GPUs for AI workloads under Debian though, so your use case might vary.

An intake is created for review with XO team. Thanks for the feedback @puffymob !