Win11 VM opening many tabs at once crashes proxmox host
- Thread starter Snuupy
- Start date
I just had my first crash with this custom CPU we made. Yep, it's way less often, but the issue still exists.
current bios is v1.09 (latest @ https://www.minisforum.com/new/support?lang=en#/support/page/download/79), but there have been known to be BIOS issues in the past for this chipset (not just this particular minipc model).
I looked up the microcode files to make sure there are no updates:
Code:
# grep microcode /proc/cpuinfo | uniq
microcode : 0xa704104
root@SnuUM780:/# cat /proc/cpuinfo
processor : 0
vendor_id : AuthenticAMD
cpu family : 25
model : 116
model name : AMD Ryzen 7 7840HS w/ Radeon 780M Graphics
stepping : 1
microcode : 0xa704104
- `vendor_id`: AuthenticAMD / 0x8
- `cpu family`: 25 / 0x19
- `model`: 116 / 0x74
- `stepping`: 0x1there is no file named "xxx74F01" so there is currently no available microupdate for the 7840 cpu. (following https://github.com/platomav/CPUMicrocodes/tree/master/AMD from https://www.reddit.com/r/linux/comments/15xvpfg/updating_your_amd_microcode_in_linux/)
That is why when installing
amd64-microcode (via wget http://ftp.us.debian.org/debian/pool/non-free-firmware/a/amd64-microcode/amd64-microcode_3.20231019.1_amd64.deb) there is no entry in dmesg saying that the microcode has loaded, because there is no firmwire file for this model, i.e. there are no updates even after following https://cyrusyip.org/en/post/2023/01/31/install-microcode-on-proxmox/ .Maybe there will be a microcode update in the future, but I don't see one for now.
hopefully with software updates in the future it can become more stable, I'm considering reverting back to a custom CPU of x86-64-v4 base + svm
I run Proxmox on MINISFORUM NPB7 (13th Gen Core i7-13700H(14C/20T) Mini Computer 32GB RAM DDR5) - almost 5 months - not even a hiccup.
Yes (slightly) more expensive, but hate to disappoint you their official TDP's are similar. See here & here.
However we all know that TDP's are one thing and real-world use; quite another.
But let me tell you that my system idles around 10-11 watts. 2 LXCs 2 VMs usually on. In my experience that's hard to beat.
Last edited:
Well that's exactly what I have been doing for 1 month now. Not a single crash.
There's definitely an issue with this CPU when we use `host`
Which is kinda bad since I need GPU passthrough
Just to add another data point to this issue - I have a Thinkpad (P16v Gen1) with a 7940HS, running Proxmox (latest non-subscription version 8.1.4) and was experiencing the same issue - random resets every once a while. I have a single Windows VM running on this machine with other VMs being Linux.
Because of the issue, I decided to have the motherboard replaced since I thought it might be a hardware issue. While waiting for the tech to arrive, I found this post! So, the motherboard has been replaced (likely not needed), and I changed the Windows VM's CPU type to x86-64-v4. No random resets anymore. Changing two things at a time isn't the best troubleshooting method, but regardless, I think the issue was, indeed, related to the CPU type being set to "host" on this machine.
Because of the issue, I decided to have the motherboard replaced since I thought it might be a hardware issue. While waiting for the tech to arrive, I found this post! So, the motherboard has been replaced (likely not needed), and I changed the Windows VM's CPU type to x86-64-v4. No random resets anymore. Changing two things at a time isn't the best troubleshooting method, but regardless, I think the issue was, indeed, related to the CPU type being set to "host" on this machine.
I'm imagining this was under warranty (computer is under a year old). You cost Lenovo some bucks! They/you could test the machine under a native bare metal OS to see if it suffers the same random resets.
Just ranting .....
Last edited:
Yes, under their premium warranty. I've given enough money to Lenovo that I don't exactly feel bad - plus they were willing to do it. I didn't have time to deal with testing, honestly, much less waiting, sometimes days, for it to reset.
how did you (or you guys in general using those exact miniPC model) design your network ? you directly connect hosts using 2,5GbE interfaces (physically like switch stack) and use USB-Lan-1Gbps/2.5Gbps additional adapters for VM traffic? or with switches and one network is then shared for data/VM network traffic?
Running a Minisforum UM780 XTX (Ryzen 7840HS - Zen 4) here. Came across your thread because someone in the unofficial Proxmox Discord mentioned using CPU Type = host had been known to cause problems with random server reboots with absolutely no trace of any problems in journalctl logs.
After reading this thread I updated my Windows 11 VM to use the following CPU settings and have had zero crashes since making the change.
This is the full cpu line in my vm config:
After reading this thread I updated my Windows 11 VM to use the following CPU settings and have had zero crashes since making the change.
This is the full cpu line in my vm config:
cpu: x86-64-v4,hidden=1,flags=+virt-ssbd;+amd-ssbd;+aesThis probably works because you are not using the type "Host" in this config.
I am still using `x86-64-v4` without any crashes since my last post.
But I still can't use Host type CPU. I mean I don't dare. Didn't retry since then.
Because of this
- I can't do any nested virtualization, like installing docker or hyper-v
- I can't passthrough a GPU