Mailinglist Archive: opensuse (1420 mails)

< Previous Next >
Re: [opensuse] lack of space on /boot for new kernel
On 2014-01-16 16:18 (GMT-0500) Fr David Ousley composed:

On Thu, 16 Jan 2014 14:16:42 -0500, Felix Miata wrote:

Those installation tools lie about need. They report gross need, not how
much available space is short of the actual need for you to use to
>calculate how much needs to be freed up.

Not sure what to do with this: zypper gags, claiming not enough space. So
even if there is actually enough, I don't know how to get from here to

df would tell you how much is used and available, which you can compare to what zypper said was required.

If your /boot uses a journaling filesystem it is doing so for virtually
no reason and wasting space in the process. Most of /boot's infrequent
>accesses are only reading. Writing rarely happens except at new kernel
installation or old kernel removal times. All my /boot partitions are

Mine is ext4, which I guess is unnecessary, from what you say. ext4 is
journalling, right?

Right. I don't know if there is any way to convert to non-journaled without backup->format->restore->reinstall Grub. Might be something better reserved for some other time now that smaller initrds have solved your problem.

You could unmount your /boot before installing the kernel, and it would
install to the root filesystem instead. To do that, after unmounting
>and before updating the kernel, mount it somewhere else to copy its
current content to the new /boot, update fstab, and update
/etc/>grub.conf. If Grub is on MBR you may be good to go already just by
updating, but I'd ensure booting is possible before trying a reboot
>according to what is currently installed where. Grub probably isn't
already installed to your root filesystem and may need to be.

This suggests the best long-term solution (someone else suggested moving
/boot to the root file system).

That used to not work with RAID if it needs to be home to Grub. IIRC it's supposed to be OK with Grub2.

Let me see if I've got it:

unmount /boot (I'm on a running system now)
mount boot somewhere on the root filesystem
update fstab to point to where /boot now is
edit /etc/grub.conf to do the same

then install the new kernel.

Did I miss anything? Is a reboot needed before installing the new kernel?

Grub2 specifies root= via UUID by default. If you were using Grub2 to reboot before installing the new kernel, Grub2's menu config file would be using the old UUID, which may or may not be appropriate to the relocated arrangement. Installing the new kernel after the moves should fix it up automatically. There ought to be a howto available somewhere from Google for this kind of move. I've only ever done it involving ext2 & ext3 without RAID or LVM. It's actually something I commonly do for the first installation to a new or wiped HD.

I don't know how to "ensure booting is possible" other than by trying to
reboot -- can you help me with that one?

That's more a do you know how to and are you ready to boot rescue media to fix it if it fails comment. First, think about the boot process itself and decide how many bases there are and how many were touched. Maybe my unfamiliarity with with Grub2 and you not telling us whether you are booting from Grub or Grub2 is causing me to miscount.

It should be instructive to have complete fdisk -l output, plus the content from /etc/grub.conf, in order to confirm your count of the bases.

Many thanks! While I've been using linux (and unix and xenix) for 25
years, I'm an end user and not really expert in all this.

I too used Xenix (and before that DOS) long before OS/2, Windows and Linux, but since close to 30 years ago.
"The wise are known for their understanding, and pleasant
words are persuasive." Proverbs 16:21 (New Living Translation)

Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata ***
To unsubscribe, e-mail: opensuse+unsubscribe@xxxxxxxxxxxx
To contact the owner, e-mail: opensuse+owner@xxxxxxxxxxxx

< Previous Next >