Can't start the Debian installer from a GUI generated VM

[nickt]

Hi,

As a long time fan of iohyve hosting Debian VMs on my FreeNAS machine, I thought I should move with the times and build my next VM using the new GUI. It hasn't gone well, and I'm hoping someone can tell me what stupid thing I'm doing wrong.

I've followed the wizard, and prepared a VM in the GUI as follows:

• 3 vCPU, 2GB RAM, UEFI bootloader
• 2 newly created Zvols connected as DISK devices with VirtIO. When I install Debian, I will use one for / and one for /home
• 1 Debian image connected as a RAW device with VirtIO (https://cdimage.debian.org/debian-cd/current/amd64/iso-cd/debian-10.3.0-amd64-netinst.iso)
• 1 GE adapter connected as NIC device with VirtIO

I start the VM, I immediately get the Debian install menu, choose "Install" and without much delay get:

Code:

error: no suitable video mode found.
Booting in blind mode

At which point absolutely nothing happens. All I can do is power off the VM and start again.

I'm rather stumped and surprised. I haven't done anything fancy. I know the Debian install image would have worked if I built it with iohyve, because I've done it many times before. I'm keen to get the benefits of VirtIO (the iohyve connected disks always perform rather poorly), which is why I decided to try the GUI way.

Any help greatly appreciated!

My FreeNAS setup per signature.

Many thanks!

 

[dweimer]

Change your VNC resolution to 800x600. I struggled with this for a while myself before stumbling across the fix a few weeks back.

 

[nickt]

Guru - thank you. There were a few other things I had to do:

• Actually configure a VNC device - I had been hoping to build without VNC using a serial console as I usually do for iohyve
• Use the legacy interface. The new interface still didn't work

 

[Patrick M. Hausen]

Use VNC for installation, then configure a serial console for GRUB and the running system afterwards. The remove VNC again. That's what I do for all my Ubuntu installs.

 

[nickt]

So I got my Debian build up and running. Everything worked really well - could operate over VNC (web client) and UEFI boots worked fine. Once I had a functioning SSH config, I decided to get rid of the VNC device, which promptly killed it.

Now when the VM starts, I get:

Code:

Boot Failed. EFI Misc Device
Boot Failed. EFI Misc Device 1

After that, it drops out to the UEFI shell and I have no idea how to get it back.

I tried adding a VNC device back in, but no joy.

Help!!

 

[KrisBee]

@nickt So you removed the VNC device AND configured your debian VM to use a serial console as Patrick advised?

 

[nickt]

Hi @KrisBee. I did, although I found the serial console unreliable. From an SSH session to my FreeNAS host:

Code:

$ cu -l /dev/nmdm5B -s 9600

worked from power up to see the failure output per my previous post.

Since then, I've figured out I can exit from the EFI shell and use the menu system to find the boot manager. From there, I can boot from file and locate a .EFI file that boots my VM again.

But the serial console stops working (just goes blank / silent) once Debian starts booting for real.

I've also discovered that the NIC device has come up with a different identifier since the VNC removal. In the original build, it had enp0s5, but after removing the VNC device, it changed to enp0s4. Multiple restarts and it has stayed as enp0s4. All very odd.

 

[KrisBee]

Serial console can be a bit iffy - i just disconnet and then re-connect - but with ssh is in use, the console is there if you need it. I don't know how you changed your default grub entries but I stuck with typical 115200n8 setting in linux and don't use a speed setting with cu.

The bhyve slot allocated to the virtual NIC device changes when you remove VNC device as can be seen if you tail the middlewared.log. So systemd's persistant interface naming menas your NIC name changes in the VM.