Unable to mount the root fs on stable/8 r264339, GENERIC kernel, with MBR, FreeBSD slice, and UFS volume labels

[Trond Endrestøl]

On Sat, 12 Apr 2014 10:53-0600, Warren Block wrote:

> On Sat, 12 Apr 2014, Trond Endrestøl wrote:
> > On Fri, 11 Apr 2014 23:09-0600, Warren Block wrote:
> > > 
> > > My first guess would be missing bootcode in the BSDlabel slice.  But I may
> > > have missed that when I missed earlier parts of this thread.
> > 
> > How? Why? The system has booted FreeBSD stable/8 for the last five
> > years using MBR, slice, and UFS labels. Why should it stop all of a
> > sudden?
> 
> Accidental overwrite or (long shot) disk sector failure?  I don't know,
> really.  Check the smartctl output for that disk.  The BSDlabel bootcode can
> be reinstalled without harming the filesystems.  I'd still make a backup
> first, something questionable is going on.
> 
> Inclusion of the geom_raid kernel module for the first time, and old RAID
> metadata on the disk?  That would show in /var/log/messages or dmesg.  If
> that's the problem, it can be disabled in loader.conf or at a boot prompt with
> kern.geom.raid.enable=0.

I wouldn't rule out anything yet, but the spare system has a fairly 
recent and hardly used harddrive, dated 2010 on the sticker, an WDC 
WD5000AAKB-00H8A0. Incidently, this is the same kind of drive as I 
have in the amanda box. :-/

On Saturday, I installed 8.0-RELEASE on the spare system using the 
standard menu option, which results in MBR and slice. I overrode the 
newfs options by adding -L volumename to each fs. Before rebooting I 
edited /etc/fstab to use the UFS labels found in /dev/ufs instead of 
the usual device names.

It turns out this arrangement is readable by 8.4-RELEASE dvd1 and also 
by the stable/9 snapshot marked 20140405 r264148. I wiped the drive 
clean and installed 8.4-RELEASE the same way as I did with 8.0-R. This 
arrangement is readable by 8.0-RELEASE dvd1 and by the stable/9 
snapshot.

Something must have happened in stable/8 between r258344 and r264339. 

Today's plan is to rewind the working copy of stable/8 on the spare 
back to r258344 and confirm that the UFS labels are accepted by the 
kernel of that revision. If all goes well, I'll move forward one 
revision at a time until I reach the revision where it all breaks.

I haven't checked the disklabels with any recent stable/10 snapshot, 
so I might do that as well.

Thanks for all your input.
I'll keep you posted.

-- 
+-------------------------------+------------------------------------+
| Vennlig hilsen,               | Best regards,                      |
| Trond Endrestøl,              | Trond Endrestøl,                   |
| IT-ansvarlig,                 | System administrator,              |
| Fagskolen Innlandet,          | Gjøvik Technical College, Norway,  |
| tlf. mob.   952 62 567,       | Cellular...: +47 952 62 567,       |
| sentralbord 61 14 54 00.      | Switchboard: +47 61 14 54 00.      |
+-------------------------------+------------------------------------+