Brian Stark wrote
> 
> Hello,
> 
> Since my problem happened on the i386 platform, I am sending this to
> port-i386 in addition to netbsd-help.
> 
> I have a 1.3.3 system that I tried to upgrade to 1.4 last night and during
> the installation I saw these messages (sysinst was running from the boot
> disk):
> 
>   uid 0 on /: file system full
>   /: write failed, file system is full
>   Foating point exception
> 
> There was a core dump for sysinst and at the command prompt I ran df.
> The output for the root file system is:
> 
>   /dev/md0a  1375  1124  251 81% /
> 
> I then re-ran sysinst and got the following message:
> 
>   Help! No /etc/fstab in target disk wd0.
> 
> Anyone else seen this? Luckily, I did have a backup, and I was able to 
> perform a new installation on the disk without problems. 
> 
> I have one more system that I need to upgrade, and I'm wondering if 
> others have seen this problem, or know how to solve it. In case it
> matters, the system is a Dell Pentium, 166Mhz, with 128MB memory.

G'day,

I observed the same problem here and have since upgraded NetBSD-1.4 manually
as I didn't have time to troubleshout this phenomena.

What happens is that sysinst dumps core onto the root system which in
my case was to small to accommodate the file, consquently running
out of disk space and hence the "disk full" error message.

During the attempted upgrade process, sysinst moved the /etc directory
to /etc/etc.old (by memory) and by that time has gathered filesystem
information form the old fstab file.

When trying to rerun the upgrade using sysinst a second time, it will
be unable to find this file since it has been moved into another
location.

I can't explain the coredump, but perhaps it was due to the 
original filesystem being of the old type "165", as this happens after
sysinst fsck'ed the partitions - not much debugging info here...


cheerio Berndt
-- 
Name	: Berndt Josef Wulf
E-Mail	: wulf@ping.net.au
Sysinfo	: DEC AXPpci33+, NetBSD-1.3.3