I've just been ignoring htree indexing for now, but the regression
failure in Ted's rebased 64bit tree looks suspiciously htree-related.
What's the interaction between htree and 64 bit file systems? What
needs to change, if anything?
-VAL
On Tue, Feb 03, 2009 at 05:22:02PM -0500, Valerie Aurora Henson wrote:
> I've just been ignoring htree indexing for now, but the regression
> failure in Ted's rebased 64bit tree looks suspiciously htree-related.
>
> What's the interaction between htree and 64 bit file systems? What
> needs to change, if anything?
There shouldn't be any. The block numbers used in htree are logical
block numbers (i.e., block 0 is the first block in the directory,
block 1 is the 2nd block i the directory, etc.)
- Ted
Theodore Tso <[email protected]> writes:
> On Tue, Feb 03, 2009 at 05:22:02PM -0500, Valerie Aurora Henson wrote:
>> I've just been ignoring htree indexing for now, but the regression
>> failure in Ted's rebased 64bit tree looks suspiciously htree-related.
>>
>> What's the interaction between htree and 64 bit file systems? What
>> needs to change, if anything?
>
> There shouldn't be any. The block numbers used in htree are logical
> block numbers (i.e., block 0 is the first block in the directory,
> block 1 is the 2nd block i the directory, etc.)
>
> - Ted
What if the directory is REALLY HUGE? :) Say >68.719.476.736 files.
Can ext4 64bit have directories >8TiB?
MfG
Goswin
On Sat, Feb 07, 2009 at 02:06:56PM +0100, Goswin von Brederlow wrote:
> What if the directory is REALLY HUGE? :) Say >68.719.476.736 files.
>
> Can ext4 64bit have directories >8TiB?
No, ext4 directories are limited to 2TB. Htree directories are
currently limited to a depth of two, which for smaller block sizes
will be more of limitation than the maximum size limitation.
- Ted