DANGER!!! Problems with 10.04 installer (RAID devices *will* get corrupted)

[Alvin Thompson]

On 04/23/2010 05:45 AM, Dave Howorth wrote:
> Actually grub can write into partitions if it's asked to. Multiple
> independent copies can be installed in the MBR area and into every
> partition on a disk if you so wish.

There's no evidence that this was a GRUB problem, but plenty of evidence 
that suggests that it's not.  Please read the bug comments for details. 
  Additionally, until there's evidence that suggests that they are 
separate problems, it's reasonable to assume that they are the same 
problem, since they have the same initial conditions, followed the same 
procedure, and showed the same results.

> Equally, you don't need to create any partitions on a disk to create an
> LVM and I'm speculating that the same may be true for fakeRAID. They
> just write to the /dev/sda device.

So?  Also, see above.

> So there are possibilities for grub to write to the same area as
> filesystem data in certain cases.

Really?  What cases are those?  Also, see above.

> One other difference between #191119 and your report is that #191119 was
> reported two years ago and so using grub 1, whereas your report
> presumably uses grub 2. So that's at least one potentially significant
> difference between the two report environments.

You're proving *my* point with that.  That means that it's very unlikely 
that GRUB is the problem.

> I understand that your problem has to do with a left over filesystem and
> I also agree that it probably isn't a grub problem. But I also believe
> that a grub problem *is* a possible explanation of #191119. Which would
> mean that the two are different problems, not duplicates.
>
> Which is why I still think it is premature to mark them as duplicates.

While it's certainly possible that there are two separate bugs that 
produce the exact same results under the exact same circumstances, the 
simplest and most likely explanation is that they're the same bug. 
That's why I think it's premature to assume they're *not* duplicates.

> Reading the thread, it seems like your reason for marking them as
> duplicates is not for technical accuracy but a political opinion that
> they will get more attention if marked as duplicates. If so, I
> understand your motivation and I simply don't know what rule to use to
> form an opinion as to whether it is 'right'.

It's not a political opinion, it's a logical, provable fact that if bug 
A is marked as a duplicate of bug B, bug B will get at least as much 
visibility as bug A.  However, the reason I marked the bug as a 
duplicate is not that, the reason is because it's more likely than not 
that the issues are the same bug.  Is it outside of the realm of 
possibility that the issues are caused by a different bug?  No, but if 
you use that criteria to argue against marking a bug as a duplicate, no 
bug could ever be marked as a duplicate as another.

Additionally, despite the historical usage of the term "bug" everywhere, 
Launchpad is an ISSUE tracker, not a bug tracker.  If there are multiple 
bugs that cause the same problems, they belong under the same issue. 
For example, Bug #1 is that Windows has majority market share.  There 
are many causes for that.  By your logic all of the the causes would 
have to be separated into different bug reports.