I originally ran into this problem trying to redeploy a machine using MAAS, but I can reproduce it just with the already-installed OS and GRUB, with Ubuntu 20.04. The machine is set up for UEFI boot.
To reproduce (but see below):
- Boot the machine and get the GRUB menu.
- Press 'c' to get a command line, run
net_dhcp efinet1 (or just net_dhcp; efinet1 is the primary network interface for that machine).
- Press escape to go back to the book menu, select Ubuntu and press enter.
- Blank screen appears, nothing happens.
By adding set debug=linux I get the message loader/efi/linux.c:96: kernel_addr: 0x3c7d1000 handover_offset: 0x190 params: 0x3ffff000 that suggests it's gotten as far as intending to launch the kernel, but there is no boot output from the kernel.
Here's where it gets weird: we have 8 almost-identical machines, and this is the only one that's exhibiting this behaviour. The only differences I'm aware of with this one is that it has a ConnectX-5 EN NIC instead of ConnectX-6 Dx (but that's not the primary interface, i.e. not involved in DHCP), and a different GPU (they're all RTX 30xx though). The BIOS is the same version (the latest), the BIOS settings are all the same, the BMC settings are all the same. I've eyeballed the DHCP offers and they seem to be the same apart from the IP address. Since this also happens when MAAS does a PXE boot I think it's unlikely to be due to anything on the hard drive.
Any ideas, either to narrow down the cause or to identify what makes this machine different?
EDIT: I also reproduce the behaviour with net_dhcp efinet2, and that interface doesn't even have a cable plugged in. So I think that rules out the DHCP offer as being the culprit, leaving me more confused than ever.
