Bug report out of context: Harcy handling of /dev/sd? devices is unacceptable

Derek Broughton news at pointerstop.ca
Thu Aug 7 15:06:03 UTC 2008


Rashkae wrote:

> Alan E. Davis wrote:
> 
> 
>> 
>> Assignation of blame is a minor issue.  I am concerned that (contrary to
>> your opinion) Ubuntu has slipped a fast one.  There is no UUID in gentoo
>> /etc/fstab at this point, so in what way was that *mandated by kernel
>> changes?*
>> 
> 
> UUID is not mandated by kernel, but change for /dev/hd? to /dev/sd? is.

Precisely.  UUID was a very smart way to deal with the issue.
 
> You will learn the mistake of your assumptions in time.  I know you
> don't believe me, and I won't convince you, but the change was made by
> people who know much more for a *very* good reason.  UUID is the only
> way to stop the system from not working when hardware is changed, or
> kernel changes device assignment, or or or....  Otherwise, we're back to
> the days of instructing people with broken system to boot from rescue cd
> and edit fstab... yeah, that's so much less hassle for average users.

There _is_ a post from someone here saying he's having issues with plugging
new USB devices changing the location of _existing_ mounts, and if so, that
_is_ a serious bug, but nothing reported here is a bug.

> This is a direct result of mixed SATA/PATA, assuming your CD-rom is IDE.
>  When you boot from the CD-rom, the BIOS gives ide precedence, and your
> IDE hard drive is considered drive 0.  Then when you switch to booting
> on SATA, the BIOS gives SATA precedence, and identifies the SATA drive
> as drive 0, and of course, re-numbers everything else.  I'd be surprised
> if any distro can install from an IDE device to a SATA device, with
> other IDE hard drives, and not need a little post-install tweaking.

Thanks for making it more clear than I could.

>> I've not seen this elsewhere.  If it's not a bug, what is it?
> 
> PEBKAC

LOL :-)
-- 
derek





More information about the ubuntu-users mailing list