Re: extracting base.tgz hangs

[Emil Skoeldberg <esg%SDF.ORG@localhost>]

On Sat, May 07, 2011 at 08:55:49PM +0200, Manuel Bouyer wrote:
> On Fri, May 06, 2011 at 09:37:03PM +0000, Emil Skoeldberg wrote:
> > In the process of trying to upgrading from 5.0 to 5.1 (i386) I ran
> > into a problem when extracting base.tgz The first attempt
> > tar zpxvf base.tgz
> > ran for a while, but then hanged. I could get a response
> > when pinging the machine, and switch between virtual consoles, but
> > nothing more - no response from any shell. A second attempt
> > more or less immediately hanged.
> > 
> > All this is a bit annoying, but a more pressing problem arose during
> > the second rebooting after powering off the machine:
> > 
> > panic: init died (signal 6, exit 5)
> > 
> > and then I'm in the debugger. I suppose the file system didn't really
> > like the power cut. This leaves me with a few questions:
> > 
> > 1) What is the reason for the tar/pax failure? [I don't know how
> > related it is, but I've previously noticed that disk i/o does not
> > complete without a bit of nudging - e.g. pressing return a few times
> > during fsck etc.]
> > 
> > 2) What should I have done instead of cutting the power when the
> > system became unresponsive?
> > 
> > 3) Is there anything I can do now? (except for trying to find a
> > floppy/cdrom drive from which I can boot and do an upgrade)
> 
> You can try to boot using rescue/ binaries:
> at boot prompt, type 'boot -a'
> when the kernel asks for init path, enter /rescue/init instead
> of /sbin/init
> then you may be able to recover/complete the upgrade using the tools
> in /rescue

Having had no luck with booting using rescue/, I have wiped the disk
clean and tried several times installing 5.1 from both CD and NFS
without any luck whatsoever, same error every time - when extracting
base.tgz the computer hangs. After giving up on this I tried to
install 4.0 instead, this worked without any problems, twice, on two
different disks (I had tried both for my failed attempts with 5.1).

Might it be the case that 5.x makes assumptions of slightly more
modern hardware?

/Emil


-- 
esg%sdf.lonestar.org@localhost
SDF Public Access UNIX System - http://sdf.lonestar.org