[SOLVED] Proxmox crash right after displaying login screen

[Ryancooper98]

Hi,

My Proxmox server crashed yesterday morning for an unknown reason. Since I've tried many thing to repair it and I can't find the problem.

The problem is that the server seems to crash right after (something like 1 to 5 secondes) showing the login screen. I know it because I connected a monitor to the server (and disabled some config to enable display output in the boot menu).

I can access the console by using the proxmox recovery mode or I can edit config files when mounting the proxmox drive on my computer.
That way I looked at syslog but couldn't find any problem.

Also, I can't access the VM config files, when I 'ls /etc/pve', there is nothing under there, like no 'qemu-server' folder or anything

I also made a short video to maybe better explain the problem : https://youtu.be/5RJb022VjXw

Feel free to ask for more informations.

Thank you in advance for your help

Edit :
- Also attached journalctl logs (using journalctl -b)
- Added precision with '/etc/pve' empty folder

 

[leesteken]

Looks like your system automatically starts a VM with a passthrough NVidia GPU and that that same GPU is also used to boot the system. Maybe the passthrough no longer works because of a kernel update. Try changing amd_iommu=on (which does nothing) to amd_iommu=off to prevent the VM from automatically starting. Maybe then you can find out what is going wrong?
Did you really need textonly video=astdrmfb pcie_acs_override=downstream,multifunction (does your system really have a AST2000?!)? Maybe you do need video=efifb:off because you have only one GPU (or use the same GPU for boot and passthrough), then maybe you also need video=simplefb:off because of recent kernel updates.

 

[Ryancooper98]

Ok, first thank you very much, by using amd_iommu=off I can get back the access to the webui. I also removed what you said that i maybe don't really need textonly video=astdrmfb pcie_acs_override=downstream,multifunction and added what you suggested : video=simplefb:off.
Now the current boot options are : "quiet amd_iommu=on iommu=pt video=simplefb:off video=simplefb:off".

I also disabled the automatic boot on startup for all of my VMs.

But when I try to boot a Windows VM (with gpu passtrough), Proxmox crashes like before but I don't know the reason why.
Other VMs without passtrough seem to work just fine.

Edit :
- Ok looks like Proxmox was using for an unknown reason the wrong IDs for the PCI Devices
- All PCI Devices works fine except the primary GPU (GTX 1070) passed trough a Windows VM (code 43, etc), don't know if it's related to the grub parameters that I removed earlier or not, but will continue to try to find it out
- Ho and btw, what's an AST2000 ?? The Internet talks about an "SS Industrial Pressure Transducer" but I don't really think that's what you walk about
- Ok, first GPU issue was related to a wrong parameter in the boot options : i wrote 2 times : video=simplefb:off instead of video=simplefb:off video=efifb:off

Thank you very much for your help you made my day <3