Re: Broken loop device
Hello Andreas,
sorry for the late reply, I was too busy to read all my emails last week :-(
How many memory is installed in your SLC ?
The diskless installation is requiring more memory space than other types of
installation due to the fact that it has to download in a ramdisk the rescue
disk before installing kernel on the target (same for drivers disk) [*].
In effect, it creates a 1.44MB ramdisk.
All installations I tried were on 16MB or more sparcs.
Is your SLC only 4 or 8 MB equipped ?
Moreover, the sparclinux kernel compiled in Debian does not provide swap on
NFS. So you are unlikely to use your SLC if it has too few memory [**]
Regards.
[*] Rescue image have to be loaded in a ramdisk before mounting it because the
loop device is broken over NFS (mount -t loop server:/path/to/image /mnt
seems to work but further reads from /dev/loopN fail).
This is why you see the message "broken loop device detected...".
If installing on a local harddisk, the image could be copied on the harddisk
before mounting it, thus not requiring so many memory filled up by the
ramdisk.
[**] Such a patch could exists somewhere on the net...
Andreas Jaehnigen wrote:
> Hi all,
>
> I recently tried to install Debian "slink" (2.1) on a Sun Sparcstation SLC.
> The machine should boot up from the net - so it has NO disks attached and
> NO floppy drives.
> The tftpboot-image just boots up fine! (But I suggest to include a hint to
> this mechanism in the official installation documents!) :-)
> Both the sources (from Debian's ftp server) and the filesystems to be used
> later are exported via NFS from a Sun Enterprise 450. After "playing
> around with it" a bit I figured out that memory was completely filled up,
> as (after several usages) every command issued on the shell results in
> a "segmentation fault"... So I connected a SCSI disk to the SLC wich
> is to be used as a (temporarily) swap device during installation. This
> seems to be working, too.
> But let me explain my file system structure on the server: As mentioned
> above, everything is exported via NFS from a Sun Enterprise 450 running
> Solaris 2.6. There are only two NFS partitions for the client (Sun
> Sparcstation SLC): a root filesystem and a /usr-filesystem (both empty).
> After I got how it's working, I selected the option "Install Operating System
> Kernel and Modules". The source for installation is NFS. And yes, I'm sure
> I've set the "root" and "rw" options for the NFS share... ;-)))
> But here, after a short while, the SLC hangs with a window on its screen
> saying "Broken loop device detected. Copying resc1440.bin to local before
> processing..."
> I can't start a shell on the other console after this happens, so I'm
> quite helpless.
> Can anyone explain to me why this happens, please?? Thanks...
>
> If you're questioning yourself why the SLC has no drives at all
> (apart from the swap disk, which will be removed after installtion is
> finished. The swappimg then will hopefully go via NFS as qwell):
> I try to set up the SLC as an X-terminal...
>
> Thanks in advance for any hints!!
>
> --
> cya
> ANDI
>
>
> --
> To UNSUBSCRIBE, email to debian-sparc-request@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
>
--
Eric Delaunay | "La guerre justifie l'existence des militaires.
delaunay@lix.polytechnique.fr | En les supprimant." Henri Jeanson (1900-1970)
Reply to: