1 Gigabyte SD card "full" after 467 MB [SOLVED]

[Derek Broughton]

Ewan Mac Mahon wrote:

> On Fri, Feb 10, 2006 at 07:45:37AM -0500, Eric Dunbar wrote:
>> On 2/10/06, Dave M G <martin at autotelic.com> wrote:
>> > Fortunately, since only the root has this problem, as mentioned
>> > above, I've moved everything into subdirectory, and that seems to
>> > have made the entire 1GB available.
>> >
>>
> It's worth noting that there is a similar limit on most unix-like
> filesystems, including ext3, where it's possible to run out of inodes
> and get disk full errors even when there's free space. Unlike the FAT
> case inodes are used over the whole fs, so if you run out, you've run
> out. In practice most filesystems have huge numbers of inodes, but you
> can see how you're doing with 'df -i'. It's easy to play with using
> tmpfs which allows you to set the number of inodes arbitrarily.

But the difference is that ext reports something to the effect that there
are no inodes...  
>  
>> Perhaps it would be worth filing a bug report with (?) Nautilus? It
>> would be helpful if Nautilus were to report the problem, and, since
>> it's an easy problem to solve, suggest a fix.
>> 
> It's not that easy; the writing app will simply get ENOSPC in response
> to either problem, with no way to tell the difference. The only way to
> do it would be to build understanding of each filesystem in each app and
> have them check for what kind of out-of-space error they were hitting.

I was afraid that was the problem, specifically because I was certain that
the original DOS/Windows implementations never differentiated between a
lack of directory entries and lack of space.

Still, it wouldn't be horrendously expensive for a program to compare
free-space on a filesystem to the size of the data it was writing - after
receiving such an error.  Then if it knows there is _space_ it follows that
the real problem must have been inodes, directory entries, etc.  Or even
just have nautilus (or konqueror or any other application you want to
mention) have its error message for "ENOSPC" suggest that it is either out
of space _or_ has reached a maximum number of files - after all, I would
suggest that simply saying the device is "out of space" _is_ a bug when
that's not really true.
-- 
derek