grub Error 15: File not found

Ken Moffat ken at kenmoffat.uklinux.net
Thu Feb 12 10:25:02 PST 2004


On Thu, 12 Feb 2004, Barry Sharpe wrote:

>
> I will have to start the LFS build over again, but i don't mind, i seem to be learning everytime


Barry, any chance you can get your mailer to wrap at around 72
characters per line ?  It will make it much easier to quote you in
future replies.

I know how much it hurts to lose the build.

> Just two questions
>
> why has the sources and tools directory survived when all others seem to have gone?
>

We don't know what went wrong.  The two likely causes are "Act of Root"
and filesystem corruption.  If you had filesystem corruption, you're
heading for recurring problems until the cause is addressed.

 It's a bit late to look in the logs from the old (RH) system, so we'll
probably never know.  I had disk corruption a couple of months ago on a
beta platform - the fsck ended by removing /bin, but I'd been booting the
new system for a while before that.  If I were you, I'd take a good look at
/tools to make sure the obvious programs and libraries are still there.

> and sfdisk now reports the following
>
> /dev/hda1     *     0+   649     650-   5221093+     83    linux
> /dev/hda2            0        -           0                  0        0    empty
> /dev/hda3            650  714        65     522112+     82    linux swap
> /dev/hda4            715  2433   1719 13807867+    83   linux
>
> is the above configuration a good setup? Before reinstalling LFS is thier
 anything i should take into account here? specfically hda2 and the fact
 that hda1 and hda4 seemed to share the same id.
>

 Unusual to have hda1,3,4 only.  Probably won't hurt (you will likely
get warnings from all of the fdisk family, but that doesn't matter).
The ID shows which type of partition it is, so it's perfectly normal to
have several partitions with the same ID.

 What I don't think you've considered is your upgrade path.  You've got
around 5GB for the RH host, and 13GB for the new LFS.  What happens in
2 or 3 years time when you want to upgrade LFS itself ?  You'll be back
to building the next LFS on hda1, but you won't be able to fit all of
the goodies from /home in to that.

 For my boxes used for general stuff, I like to have two partitions set
aside for the basic systems (say, 3GB each) with a common /home
partition.  At one time I separated /usr/local and /opt, but since
almost everything in them was linked against /lib/libc.so they didn't
look totally happy when I tried to use them with a system where I'd
upgraded libc.  For boxes set aside for testing, I might have up to 8
partitions which could be used for '/', but that's not a common
requirement :)  For something providing production services, you might
want to separate /usr and /var.  Some people put music or videos onto a
separate partition.  And I mostly prefer to have a common partition
for the kernels and boot manager (although I mounted it at somewhere
other than /boot when I installed RedHat because I was a bit wary of
how RH would install grub).

  Clearly, if you want more than four partitions you'll need to use
a logical parttition.  Theoretically you could reduce the size of the
RedHat partition, reinstall, then copy all of the tools and source
directories into it, turn off swap, remove the LFS and swap partitions,
and make extended partitions in the space.  Depending on your skills
and experience, this might put your source and tools files at risk.

 In short, everybody develops their own view on how best to partition
and the fact that hda3 has zero cylinders is almost certainly not a
problem.  If you can review what you did last time, it might help you to
understand what went wrong.  Remember, just because the build host is
temporary doesn't mean you have to be in a rush to dispose of it or
overwrite its bootloader.

Ken
-- 
Brighton tops UK Jedi league
http://www.theregister.co.uk/content/28/35186.html




More information about the lfs-support mailing list