Having problems freeing space in /boot

[TheFu]

I am fairly sure that it is essential to remove an older, non-working nvidia driver before attempting to install another so follow TheFu's idea and you may get things going again.

I haven't been removing older nvidia driver versions, but I did remove 3 older versions today. I'm on a GT 1030, so high-end GPUs might be completely different. My other physical machine with a GPU has an AMD and it "just works", once the kernel with support for that iGPU version was installed (which wasn't the normal kernel for my hardware and OS release).

[elpidiovaldez5]

I managed to get the machine to boot with nvidia-driver-470

NVIDIA-SMI 470.129.06   Driver Version: 470.129.06   CUDA Version: 11.4

paul@clio:~$ ubuntu-drivers devices
== /sys/devices/pci0000:00/0000:00:1c.7/0000:09:00.0/0000:0a:07.0/0000:0e:00.0 ==
modalias : pci:v000014E4d000043A0sv00001043sd00008659bc02sc80i00
vendor   : Broadcom Inc. and subsidiaries
model    : BCM4360 802.11ac Wireless Network Adapter
driver   : bcmwl-kernel-source - distro non-free

== /sys/devices/pci0000:00/0000:00:03.0/0000:01:00.0 ==
modalias : pci:v000010DEd00001B06sv00003842sd00006696bc03sc00i00
vendor   : NVIDIA Corporation
model    : GP102 [GeForce GTX 1080 Ti]
manual_install: True
driver   : nvidia-driver-515 - distro non-free recommended
driver   : nvidia-driver-510 - distro non-free
driver   : nvidia-driver-510-server - distro non-free                                                                                                                   
driver   : nvidia-driver-515-server - distro non-free                                                                                                                   
driver   : xserver-xorg-video-nouveau - distro free builtin                                                                                                             

== /sys/devices/pci0000:00/0000:00:02.0/0000:02:00.0 ==
modalias : pci:v000010DEd0000128Bsv00001462sd00008C93bc03sc00i00
vendor   : NVIDIA Corporation
model    : GK208B [GeForce GT 710]
driver   : nvidia-driver-418-server - distro non-free
driver   : nvidia-driver-470 - distro non-free
driver   : nvidia-driver-450-server - distro non-free
driver   : nvidia-driver-390 - distro non-free
driver   : nvidia-driver-470-server - distro non-free
driver   : xserver-xorg-video-nouveau - distro free builtin

It is showing as manually installed, which I imagine means that it won't be automatically updated. That is not a problem for now.

I really want to make sure /boot is now cleaned and consistent. The following commands show the situation:

paul@clio:~$ uname -r
5.4.0-121-generic

paul@clio:~$ dpkg -l | tail -n +6 | grep -E 'linux-image-[0-9]+'
rc  linux-image-5.4.0-107-generic                 5.4.0-107.121                        amd64        Signed kernel image generic
ii  linux-image-5.4.0-110-generic                 5.4.0-110.124                        amd64        Signed kernel image generic
ii  linux-image-5.4.0-121-generic                 5.4.0-121.137                        amd64        Signed kernel image generic

paul@clio:~$ ls -lh /boot
total 368M
-rw-r--r-- 1 root root 233K Apr 14 13:19 config-5.4.0-110-generic
-rw-r--r-- 1 root root 233K Jun 15 14:13 config-5.4.0-121-generic
drwx------ 3 root root 4.0K Jan  1  1970 efi
drwxr-xr-x 4 root root 4.0K Jul  9 03:51 grub
lrwxrwxrwx 1 root root   28 Jul  9 03:51 initrd.img -> initrd.img-5.4.0-121-generic
-rw------- 1 root root  55M Jul  9 01:02 initrd.img-5.4.0-110-generic
-rw------- 1 root root  56M Jul 10 02:54 initrd.img-5.4.0-121-generic
-rw------- 1 root root 111M Aug 25  2021 initrd.img-5.4.0-77-generic
-rw------- 1 root root 111M Sep  8  2021 initrd.img-5.4.0-81-generic
lrwxrwxrwx 1 root root   28 Jul  9 01:05 initrd.img.old -> initrd.img-5.4.0-110-generic
drwx------ 2 root root  16K Jan 22  2021 lost+found
-rw-r--r-- 1 root root 179K Aug 18  2020 memtest86+.bin
-rw-r--r-- 1 root root 181K Aug 18  2020 memtest86+.elf
-rw-r--r-- 1 root root 181K Aug 18  2020 memtest86+_multiboot.bin
-rw------- 1 root root 4.6M Apr 14 13:19 System.map-5.4.0-110-generic
-rw------- 1 root root 4.6M Jun 15 14:13 System.map-5.4.0-121-generic
lrwxrwxrwx 1 root root   25 Jul  9 01:01 vmlinuz -> vmlinuz-5.4.0-121-generic
-rw------- 1 root root  14M Apr 14 13:56 vmlinuz-5.4.0-110-generic
-rw------- 1 root root  14M Jun 15 14:18 vmlinuz-5.4.0-121-generic
lrwxrwxrwx 1 root root   25 Jul  9 01:05 vmlinuz.old -> vmlinuz-5.4.0-110-generic

There seem to be some left over images taking a lot of space:
initrd.img-5.4.0-77-generic
initrd.img-5.4.0-81-generic

Can I just delete these files ? Are there any other places where bits of these old kernels might be lieing around ? I don't know how the system got into such a state, but it has persisted for a long time as all attempts to fix it failed. Finally, thanks to help on this forum, things seem to be getting better.

[oldfred]

You may want to also check /lib/modules & /usr/src.

Sometimes best way to fix it to reinstall, so all the bits are there. Then the uninstall works. When you have some files, but not others, you have to manually house clean everything.

I also like synaptic but sometimes use command line.

ls /var/cache/apt/archives/linux-image*
ls -al /usr/src/
ls -al /lib/modules/
ls -al /boot/
dpkg -l | grep linux-
ls -al /boot/vmlinuz* /boot/initrd.img*

[ActionParsnip]

If the kernel is removed, then you can clean up with:

sudo dpkg -P linux-image-5.4.0-107-generic

[TheFu]

There is an overriding rule about packages and files.

Never manually delete a file from the system areas that was installed with a package and expected to be part of the package removal, except through the package manager. This is because when it is time for the package management system to remove a package, it is going to try to remove all the files it installed and expect to still be there. If 1 of those file doesn't exist, the removal will fail. That's a hassle.

So, those specific files that you believe are extra in /boot/ ... which package are they part? There are apt and dpkg commands that accept the absolute path to a file and will return the package to which they belong. Some examples:

dpkg -l {part-of-package-name}*

#  list package names based on specific file location
dpkg -S /usr/bin/passwd
dpkg-query --search '/path/to/file'
apt-file search vim

#  list of files in a known package:
dpkg-query -L {pkg-name}

If those files you believe are extra aren't part of any installed package on the system, then you can delete them manually. If it turns out you made a mistake, you can create a zero-length (empty) file with the exact name, and use the normal APT methods to remove the package and it will remove those empty files. I suppose there's a -f (force) option, but sometimes it is just easier and safer to avoid that during removals.

BTW,

$ df -Th /boot

is a more helpful command to see about space in /boot.

[elpidiovaldez5]

I checked /usr/src and it looks good. However /lib/modules looks wrong. There are a lot of folders for kernels that are no longer installed.

paul@clio:~$ dpkg -l | tail -n +6 | grep -E 'linux-image-[0-9]+'
rc  linux-image-5.4.0-107-generic                 5.4.0-107.121                        amd64        Signed kernel image generic
ii  linux-image-5.4.0-110-generic                 5.4.0-110.124                        amd64        Signed kernel image generic
ii  linux-image-5.4.0-121-generic                 5.4.0-121.137                        amd64        Signed kernel image generic

paul@clio:~$ ls -l /lib/modules
total 60
drwxr-xr-x 3 root root 4096 Jul  9 03:51 5.4.0-100-generic
drwxr-xr-x 3 root root 4096 Jul  9 03:51 5.4.0-107-generic
drwxr-xr-x 6 root root 4096 Jul  9 01:21 5.4.0-110-generic
drwxr-xr-x 6 root root 4096 Jul 10 02:54 5.4.0-121-generic
drwxr-xr-x 3 root root 4096 Jun  3  2021 5.4.0-65-generic
drwxr-xr-x 3 root root 4096 Jun  3  2021 5.4.0-66-generic
drwxr-xr-x 3 root root 4096 Jun  3  2021 5.4.0-67-generic
drwxr-xr-x 3 root root 4096 Jun  3  2021 5.4.0-70-generic
drwxr-xr-x 3 root root 4096 Jun  3  2021 5.4.0-71-generic
drwxr-xr-x 3 root root 4096 Jun  3  2021 5.4.0-72-generic
drwxr-xr-x 3 root root 4096 Jan 23 15:19 5.4.0-73-generic
drwxr-xr-x 3 root root 4096 Jan 23 15:20 5.4.0-74-generic
drwxr-xr-x 3 root root 4096 Jan 23 15:20 5.4.0-77-generic
drwxr-xr-x 3 root root 4096 Jan 23 15:32 5.4.0-81-generic
drwxr-xr-x 3 root root 4096 Jan 23 15:34 5.4.0-88-generic

These old kernels are not found by the package manager, and I believe are the result of many failed attempts to clean up the kernels. Is it safe to just delete everything except modules for packages 107, 110 and 121 ?

[TheFu]

/lib/modules is for kernel module packages. You should remove those that aren't matched to one of the kernels still installed. Remove the PACKAGES, not the files. Let the package manager deal with the files.

Here's mine:

$ ls -F /lib/modules
5.4.0-120-generic/  5.4.0-121-generic/  5.4.0-89-generic/

I haven't manually removed any module packages, so on a properly working system, apt should be addressing those as needed. Just run autoremove and autoclean every few weeks.

[Impavidus]

Look for the packages providing these modules directories. For example (on my system):

dpkg --search /lib/modules/5.15.0-40-generic/
linux-headers-5.15.0-40-generic, linux-modules-extra-5.15.0-40-generic, linux-modules-5.15.0-40-generic: /lib/modules/5.15.0-40-generic

This tells me that this directory in modules is provided by the linux-headers-..., the linux-modules-extra-... and the linux-modules-... packages, not by the linux-image-... package. Then remove those packages.

[elpidiovaldez5]

The packages appear to have already been removed. Somehow the system got itself into a horribly inconsistent state. However things seem to be better now - I have freed space on /boot and my graphics driver is working.

dpkg --search /lib/modules/5.4.0-{65,66,67,70,71,72,73,74,77,81,88}-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-65-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-66-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-67-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-70-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-71-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-72-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-73-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-74-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-77-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-81-generic/
dpkg-query: no path found matching pattern /lib/modules/5.4.0-88-generic/

I am going to remove the superfluous files manually.