Mountfs and runlevels

Nathan Coulson conathan at gmail.com
Sun Sep 5 08:19:15 PDT 2004


On Sat, 4 Sep 2004 23:02:30 -0700 (PDT), Doug Reich <dreich at hmc.edu> wrote:
> I recently had a great deal of trouble with my system and a corrupt
> partition. When it came to mounting filesystems, my system would hang
> and I received dma_expiry errors. The solution, for anyone who is
> interested, was to run "e2fsck -c {partition}" which checks for, and
> corrects, bad blocks on the filesystem.
> 
> The LFS runscripts attempt to mount all filesystems in
> rcsysinit.d. This is a very bad idea, as I discovered. My problem was
> with the partition on /usr, and my system would hang whenever I would
> attempt to mount it. The entire purpose of runlevel 1 is to have a
> bare-bones system on which the least can go wrong, and thus
> the system can be diagnosed and fixed. Toward this end, only the root
> filesystem should be mounted, and all other filesystems should be
> mounted in higher runlevels. In addition, I also realize two other
> things:
> 
> 1. there should be a statically linked editor on the root
> filesystem so the user can edit files that need changing, and
> 
> 2. the creation of a boot disk should be stressed more, should one
> find their system completely unusable.
> 
> Clearly, the solution is to have "mountfs" only mount the root
> filesystem and /proc and other things. Another file "mountother" or
> something along those lines should run "mount -a" or whatever else
> needs to be done.
> 
> I'd really like this changed so other people can have an easier time
> solving problems when they can't make it to runlevel 3 or higher.
> 
> -Doug
> 
> --

Please send this email to lfs-dev.  Book Development does not occur on
the lfs-book list.

-- 
Nathan Coulson (conathan)
------
nathan at linuxfromscratch org
conathan at gmail com



More information about the lfs-book mailing list