Using Ubuntu 20.04 on AWS (ec2), I wanted to change my kernel from AWS to a generic one. When I try it the suggested way with changing the Grub Config, it stops the machine.
This is because Jibri (Jitsi Video Recorder) requires the use of ALSA and to modprobe snd-aloop , but anything that would want a lowlatency or different kernel would have the same need.
What I tried:
Grab the image:
sudo apt install linux-image-extra-virtual
You’ll see it reference something like Linux 5.4.0-84-generic in the install packages, it helps later on.
Now, list your boot entries that you have for that new generic kernel
grep -A200 submenu /boot/grub/grub.cfg |grep -P '^(?=.*menuentry)(?=.*generic)'
I see:
menuentry 'Ubuntu, with Linux 5.4.0-84-generic' --class ubuntu --class gnu-linux --class gnu --class os $menuentry_id_option 'gnulinux-5.4.0-84-generic-advanced-e8070c31-bfee-4314-a151-d1332dc23486' {
menuentry 'Ubuntu, with Linux 5.4.0-84-generic (recovery mode)' --class ubuntu --class gnu-linux --class gnu --class os $menuentry_id_option 'gnulinux-5.4.0-84-generic-recovery-e8070c31-bfee-4314-a151-d1332dc23486' {
In what is likely the top line, and not the one that mentions recovery mode, copy it’s menuentry at the start of the line. On mine looks like Ubuntu, with Linux 5.4.0-84-generic , but it will be different for everyone.
Now, I will edit /etc/default/grub ,
Comment out the GRUB_DEFAULT=0 (loads the first entry in your grub bootlist), and change it to the string we found earlier. For me, this section of the file now looks like:
#GRUB_DEFAULT=0
GRUB_DEFAULT="Ubuntu, with Linux 5.4.0-84-generic"
This is not a proper setup yet - we are using it to let grub check and tell us the best way to do it.
At this point, we can sudo update-grub, which will give me a friendly warning like this (this is good!)
Warning: Please don't use old title `Ubuntu, with Linux 5.4.0-84-generic' for GRUB_DEFAULT, use `Advanced options for Ubuntu>Ubuntu, with Linux 5.4.0-84-generic' (for versions before 2.00) or `gnulinux-advanced-e8070c31-bfee-4314-a151-d1332dc23486>gnulinux-5.4.0-84-generic-advanced-e8070c31-bfee-4314-a151-d1332dc23486' (for 2.00 or later)
Again, this is good! We know that we’ve specified something it recognises. Everyone is running grub newer than 2.00 these days, so we now edit /etc/default/grub again and change it to it’s final value:
(my values will be different to yours)
#GRUB_DEFAULT=0
#GRUB_DEFAULT="Ubuntu, with Linux 5.4.0-84-generic"
GRUB_DEFAULT="gnulinux-advanced-e8070c31-bfee-4314-a151-d1332dc23486>gnulinux-5.4.0-84-generic-advanced-e8070c31-bfee-4314-a151-d1332dc23486"
sudo update-grub , and I see the warning is gone now.
Now I do a sudo reboot now
At this point, grub has given me a config that should try and boot the generic kernel, falling back to the normal one.
Instead it gets stuck in a boot loop.
In the serial console, I see this during it's boot loop repeatedly.
[ 0.000000] Linux version 5.4.0-84-generic (buildd@lgw01-amd64-050) (gcc version 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04)) #94-Ubuntu SMP Thu Aug 26 20:27:37 UTC 2021
(Ubuntu 5.4.0-84.94-generic 5.4.133)
[ 0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-5.4.0-84-generic root=PARTUUID=5198cbc0-01 ro console=tty1 console=ttyS0 nvme_core.io_timeout=4294967295 panic=-
1
[ 0.000000] KERNEL supported cpus:
[ 0.000000] Intel GenuineIntel
[ 0.000000] AMD AuthenticAMD
[ 0.000000] Hygon HygonGenuine
[ 0.000000] Centaur CentaurHauls
[ 0.000000] zhaoxin Shanghai
... lots of cpu init that's hard to copy ..
[ 10.366218] rtc_cmos 00:00: setting system clock to 2021-09-21T11:25:30 UTC (1632223530)
[ 10.373185] md: Waiting for all devices to be available before autodetect
[ 10.460173] md: If you don't use raid, use raid=noautodetect
[ 10.464642] md: Autodetecting RAID arrays.
[ 10.468024] md: autorun ...
[ 10.470867] md: ... autorun DONE.
[ 10.473906] VFS: Cannot open root device "PARTUUID=5198cbc0-01" or unknown-block(0,0): error -6
[ 10.562806] Please append a correct "root=" boot option; here are the available partitions:
[ 10.569825] Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0)
[ 10.674637] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.4.0-84-generic #94-Ubuntu
[ 10.682527] Hardware name: Amazon EC2 t3.small/, BIOS 1.0 10/16/2017
[ 10.686735] Call Trace:
[ 10.691100] dump_stack+0x6d/0x8b
[ 10.759061] panic+0x101/0x2e3
[ 10.761974] mount_block_root+0x23f/0x2e8
[ 10.765568] mount_root+0x38/0x3a
[ 10.768584] prepare_namespace+0x13f/0x194
[ 10.771873] kernel_init_freeable+0x23f/0x263
[ 10.775368] ? rest_init+0xb0/0xb0
[ 10.859141] kernel_init+0xe/0x110
[ 10.862101] ret_from_fork+0x35/0x40
[ 10.865843] Kernel Offset: 0x36a00000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff)
It's telling me that it can't find the boot device or partuuid according to it.
At this point, I pretty much have to terminate the instance and try again, as I can't recover it easily.
Is there some better way to change my kernel to the generic one? I'd be happy with getting to use the generic image in anyway possible at this point if it boots.
