Running out of space upgrading a 12MB package?

[supremedalek]

Let's say I have an 11.10 install with a disk layout that kinda looks like this:


root@ubuntu2:~# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda2 1.9G 1.1G 734M 60% /
udev 2.0G 12K 2.0G 1% /dev
tmpfs 512M 8.0K 512M 1% /tmp
tmpfs 793M 288K 793M 1% /run
none 5.0M 0 5.0M 0% /run/lock
none 2.0G 0 2.0G 0% /run/shm
/dev/mapper/ubuntu_vg0-home
2.0G 69M 1.9G 4% /home
/dev/sda1 472M 156M 292M 35% /boot
/dev/mapper/ubuntu_vg0-usr
2.0G 864M 1.1G 46% /usr
/dev/mapper/ubuntu_vg0-var
2.0G 1016M 898M 54% /var
/dev/mapper/export 1007M 85M 872M 9% /export
root@ubuntu2:~#

in which I need to upgrade a package (linux-headers-server in this case). During the package upgrade process I get the following messages:


Fetched 12.3 MB in 1s (11.2 MB/s)
Selecting previously deselected package linux-headers-3.0.0-28.
(Reading database ... 152680 files and directories currently installed.)
Unpacking linux-headers-3.0.0-28 (from .../linux-headers-3.0.0-28_3.0.0-28.45_all.deb) ...
dpkg: error processing /var/cache/apt/archives/linux-headers-3.0.0-28_3.0.0-28.45_all.deb (--unpack):
unable to create `/usr/src/linux-headers-3.0.0-28/arch/sh/include/asm/unistd_64.h.dpkg-new' (while processing `./usr/src/linux-headers-3.0.0-28/arch/sh/include/asm/unistd_64.h'): No space left on device
No apport report written because the error message indicates a disk full error
dpkg-deb: error: subprocess paste was killed by signal (Broken pipe)
Selecting previously deselected package linux-headers-3.0.0-28-server.
Unpacking linux-headers-3.0.0-28-server (from .../linux-headers-3.0.0-28-server_3.0.0-28.45_amd64.deb) ...
dpkg: error processing /var/cache/apt/archives/linux-headers-3.0.0-28-server_3.0.0-28.45_amd64.deb (--unpack):
error creating symbolic link `./usr/src/linux-headers-3.0.0-28-server/include/linux/timb_gpio.h': No space left on device
No apport report written because the error message indicates a disk full error
Preparing to replace linux-headers-server 3.0.0.27.31 (using .../linux-headers-server_3.0.0.28.32_amd64.deb) ...
Unpacking replacement linux-headers-server ...
Errors were encountered while processing:
/var/cache/apt/archives/linux-headers-3.0.0-28_3.0.0-28.45_all.deb
/var/cache/apt/archives/linux-headers-3.0.0-28-server_3.0.0-28.45_amd64.deb
E: Sub-process /usr/bin/dpkg returned an error code (1)

Why is it claiming I have no space? I am not going to say the space I allocated for this install is massive, but I think that 12MB should not break the bank.

[snowpine]

What do you mean by "kinda looks like this" exactly? Are those the real accurate numbers or not??

[oldfred]

You only have 472MB for /boot which is where the kernel & grub files go. Actually that is fairly large, but have you housecleaned any kernels or accumulated many?

Determine your current kernel:
uname -a
uname -r
In synaptic search for linux-image to choose to delete old ones
Also command line in post #8
http://ubuntuforums.org/showthread.php?t=1283521
cd /boot
ls vmlinuz*
sudo apt-get remove linux-image-[version]-generic linux-image-[version]-generic

For all the versions you may want to erase, keep at least two or three kernel images, the one you are using and one or two more.

Other HouseKeeping:
sudo apt-get update #resync package index
sudo apt-get upgrade #newest versions of all packages, update must be run first
sudo apt-get autoremove #removes dependencies no longer needed
# removes .deb
sudo apt-get autoclean # only removes files that cannot be downloaded anymore (obsolete)
Sometimes this does as upgrade only does existing packages, this adds new
sudo apt-get dist-upgrade #updates dependencies
#dist-upgrade is also able to remove existing packages if required
apt-get dist-upgrade
# fix Broken packages -f
sudo apt-get -f install
dpkg --configure -a

# empty trash
# remove log files if no issues
cd /var/log
sudo rm -f messages.*
sudo rm -v /var/log/*.gz

[supremedalek]

You only have 472MB for /boot which is where the kernel & grub files go. Actually that is fairly large, but have you housecleaned any kernels or accumulated many?

I found that ubuntu and centos would create IMHO rather small boot partitions (256 and 200MB respectively), which would cause it to be filled depending on what extra stuff I wanted during boot or if I was being too lazy. So I doubled it (ok I was bitten by the 1000 vs 1024 reporting issue) and called it good; when nagios tells me it is more than half full I then have plenty of time to do some housecleaning.


Determine your current kernel:
uname -a
uname -r
In synaptic search for linux-image to choose to delete old ones
Also command line in post #8
http://ubuntuforums.org/showthread.php?t=1283521
cd /boot
ls vmlinuz*
sudo apt-get remove linux-image-[version]-generic linux-image-[version]-generic

For all the versions you may want to erase, keep at least two or three kernel images, the one you are using and one or two more.

I am afraid it is a bit more complicated than what we originally thought:


root@ubuntu2:~# uname -a
Linux ubuntu2 3.0.0-26-server #42-Ubuntu SMP Wed Sep 5 08:55:42 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux
root@ubuntu2:~# uname -r
3.0.0-26-server
root@ubuntu2:~# ls /boot/vmlinuz-3.0.0-*
/boot/vmlinuz-3.0.0-12-server /boot/vmlinuz-3.0.0-24-server
/boot/vmlinuz-3.0.0-14-server /boot/vmlinuz-3.0.0-26-server
/boot/vmlinuz-3.0.0-19-server /boot/vmlinuz-3.0.0-27-server
/boot/vmlinuz-3.0.0-23-server
root@ubuntu2:~# apt-get remove linux-image-3.0.0-14-generic linux-image-3.0.0-14-generic
Reading package lists... Done
Building dependency tree
Reading state information... Done
Package linux-image-3.0.0-14-generic is not installed, so not removed
You might want to run 'apt-get -f install' to correct these:
The following packages have unmet dependencies:
linux-headers-server : Depends: linux-headers-3.0.0-28-server but it is not going to be installed
E: Unmet dependencies. Try 'apt-get -f install' with no packages (or specify a solution).
root@ubuntu2:~#


Other HouseKeeping:
sudo apt-get update #resync package index
sudo apt-get upgrade #newest versions of all packages, update must be run first
sudo apt-get autoremove #removes dependencies no longer needed
# removes .deb
sudo apt-get autoclean # only removes files that cannot be downloaded anymore (obsolete)
Sometimes this does as upgrade only does existing packages, this adds new
sudo apt-get dist-upgrade #updates dependencies
#dist-upgrade is also able to remove existing packages if required
apt-get dist-upgrade
# fix Broken packages -f
sudo apt-get -f install
dpkg --configure -a

# empty trash
# remove log files if no issues
cd /var/log
sudo rm -f messages.*
sudo rm -v /var/log/*.gz

I think other issues need to be solved first:

root@ubuntu2:~# apt-get upgrade
Reading package lists... Done
Building dependency tree
Reading state information... Done
You might want to run 'apt-get -f install' to correct these.
The following packages have unmet dependencies:
linux-headers-server : Depends: linux-headers-3.0.0-28-server but it is not installed
E: Unmet dependencies. Try using -f.
root@ubuntu2:~#

BTW, I did try to record the disk usae (just a script around df) as "apt-get -f install" was run, but either the sampling speed (1s) was not fast enough or the amount of disk used did not increase that much. Which leads to: why should it increase a lot? I mean, where is it unpacking the .deb files in the step

dpkg: error processing /var/cache/apt/archives/linux-headers-3.0.0-28_3.0.0-28.45_all.deb (--unpack):
unable to create `/usr/src/linux-headers-3.0.0-28/arch/sh/include/asm/unistd_64.h.dpkg-new' (while processing `./usr/src/linux-headers-3.0.0-28/arch/sh/include/asm/unistd_64.h'): No space left on device
No apport report written because the error message indicates a disk full error
dpkg-deb: error: subprocess paste was killed by signal (Broken pipe)

And how much space should it actually need to do so?

[oldfred]

It looks like you are running -26, but have -27 installed and it wants to upgrade to -28?
Then I might only uninstall -24.
With only 3 kernels you should have room, is some other partition then have issues? None looked full, but you do need some working room.


Did you have a typo in your remove command as you tried to remove -14?

[supremedalek]

You are right; I probably should have run

apt-get remove linux-image-3.0.0-14-server linux-image-3.0.0-14-server
but it is still barking about 3.0.0-28. Since I am running right now -26, is there a way I can tell it to get rid of -28 so I can deal with the rest?

About the working room, how much do I actually need? Yes, I can throw another 20GB at the problem but that does not solve it. After all, why would I need more disk space than I have? Why do you think it is not enough? What is the criteria?

I am really trying to understand what is going on here.

[oldfred]

Is this a server install?

If this is what you have installed, then it is what you should remove. I think the commands I posted were more for Desktops.

sudo apt-get purge vmlinuz-3.0.0-12-server
I used remove before but purge is probaly more correct, see man apt-get

I am only familiar with Desktop installs where the only extra partition may be /home. I do not know best or any reasonable allocation of space to the various system partitions. Last time I dealt with all those partitions was an install of Redhat 5 about 12 years ago.

More for Desktops:
Herman on advantages/disadvantages of separate system partitions post#3
http://ubuntuforums.org/showthread.php?t=1410392