F9 Live USB IO Errors

[Douglas McClendon]

Chuck Anderson wrote:
> On Mon, May 26, 2008 at 12:30:54PM -0700, Douglas McClendon wrote:
>> Andrew Overholt wrote:
>>> * Andrew Overholt <overholt at redhat.com> [2008-05-26 09:54]:
>>>> I've made some live USB sticks with persistent overlays for a few
>>>> people.  3 of the people have now told me that they can no longer boot
>>>> as they get IO errors with what looks like corrupted filesystems.
>>> 2 of the 3 have now confirmed for me that they hit hard lockups while
>>> running and had to hard-reboot their machines.  Afterwards, they could
>>> not re-boot from the USB stick.
>>>
>>> Andrew
>> Well known nature of the feature.  See a post of mine on fedora-livecd-list
>>
>> https://www.redhat.com/archives/fedora-livecd-list/2008-May/msg00051.html
>>
>> I've suggested several times over the last 6 months that this issue be
>> well documented to set expectations appropriately.
> 
> I don't understand how it could corrupt the UNDERLYING vfat partition, 
> but that is what happend to me twice.  I was unable to mount the USB 
> stick as a normal vfat data drive from a regular (non-Live) booted 
> system after filling and crashing the persistent overlay when it was 
> booted as a Live USB system.  It seems there is a vfat bug that is 
> corrupting the filesystem.

Agreed, any vfat corruption would be seem to be due to kernel bugs 
outside of liveusb persistence, though perhaps it is better at 
triggering the bugs than typical usage.  I have run across similar vfat 
corruption on usbsticks with linux.  So bad that an fsck.vfat goes 
basically crazy, i.e. asks a million questions, and if you do yes to 
all, results in a filesystem that is mountable but very corrupt, but 
vaguely reminiscent of the original contents.

But note, the original post doesn't mention that kind of problem.  Just 
that they can't boot, with what looks like a corrupted filesystem (which 
sounds like the natural behaviour when booting with a persistent liveusb 
that had filled up its overlay).

-dmc