[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Floppy and tftp - was Woody Freeze

On Mon, Apr 09, 2001 at 01:05:34PM +0200, skaya@enix.org wrote:
> > 3. It retrieves the boot-loader (GRUB) via tftp from the server.
> > 4. GRUB then loads the kernel (and maybe modules too if needed) 
> do you mean there is a way to "pre-load" kernel modules now ?
> I think GRUB allows loading of modules via multiboot specification,
> but does Linux complies to this spec ?

Well I have not tested it (yet) but as far as I have understood
the documentation it should allow this... But I might be mistaken.

> > 7. Some directories are mounted with ramfs.
> in my diskless setup, I use a /mnt/writable ramdisk created
> at boot time in init scripts, and I have a farm of symlinks
> for /tmp /var/log ... pointing to /mnt/writable/* . could
> you explain how your ramfs solution works ?

I just add some lines to /etc/fstab:

/dev/ram        /tmp            ramfs   defaults,rw,auto,dev      0       0
/dev/ram1       /var/run        ramfs   defaults,rw,auto,dev      0       0
/dev/ram2       /var/state      ramfs   defaults,rw,auto,dev      0       0
/dev/ram3       /var/lock       ramfs   defaults,rw,auto,dev      0       0
/dev/ram4       /var/account    ramfs   defaults,rw,auto,dev      0       0
/dev/ram5       /var/log        ramfs   defaults,rw,auto,dev      0       0

And then the init script mounts them for me. :)

> by the way, it looks like the only thing preventing / to be
> mounted read-only is /etc/mtab, which could be symlinked to 
> /proc/mounts. does anybody have any pro/con comments about that?

Ohh I missed that part. Yes it have to be a symlink to /proc/mounts.
But maybe there are some problems with this because none of the
/dev/ram* seems to be listed if you use df. But I have not digged
into it deeper, so I actually do not know.

> > Atually I would like to boot another kernel at the server for
> > updating the image but I have not found any good solution. The
> > user-space-linux project looks nice but it uses a file as
> > the filesystem...
> and it is not practicable in cross-platforms setup, i.e. when
> the NFS server is a PC and the client a Macintosh. looks like
That is true. Have not thought about that. :)

> there are two "solutions" : either netboot a client with write
> access to do apt-thingies, either a "dpkg --root /images/foo",
> with tons of tricks to fool postconf scripts ...

Or make a chroot to the image, and mount /proc inside it. But
then you can have problems with the server if libc (and stuff) are
not of the same version...

But let the client have write access should fix it... What
directories should not be "localized" to make that work?


// Ola

 --------------------- Ola Lundqvist ---------------------------
/  opal@debian.org                     Björnkärrsgatan 5 A.11   \
|  opal@lysator.liu.se                 584 36 LINKÖPING         |
|  +46 (0)13-17 69 83                  +46 (0)70-332 1551       |
|  http://www.opal.dhs.org             UIN/icq: 4912500         |
\  gpg/f.p.: 7090 A92B 18FE 7994 0C36  4FE4 18A1 B1CF 0FE5 3DD9 /

Reply to: