On Wed, 05 Jan 2005 18:48:32 -0600, Mike Parson wrote:

> On Thu, Jan 06, 2005 at 12:33:56AM +0100, Pavel Cahyna wrote:
>> On Wed, 05 Jan 2005 14:16:31 -0600, Mike Parson wrote:
>>
>>> About 6 months ago, I moved my homedirs on this box to a new hard-drive.
>>> After the newfs, I used a dump/restore pipeline to copy over the files.
>>>
>>> Before the copy, it took nearly 20 seconds to open my
>>> +lists/netbsd/users nmh folder in mutt with over 1000 msgs, afterwards,
>>> just a couple of seconds.  Now, its slowed down again, lots of activity,
>>> lots of little files getting created and deleted, over time, things
>>> scatter, get less efficient.
>>>
>>> This would be the similar under FFS, ext3, xfs, lfs, reiser, whateverfs.
>>
>> Are you sure about LFS? I thought that it does not scatter files, but
>> write everything in continuous fixed-size segments instead.
> 
> Okay, I admit I don't know hardly anything about LFS, but can it really
> prevent any kind of filesystem fragmentation?

The truth is AFAIK that LFS never tries to prevent fragmentation, even if
the disk is completely empty :-) More accurately, it provides time
locality of files (data writen at the same time are close to each other)
while FFS and EXT2 try to achieve space locality (related files, such as
files in the same directory are close to each other).

Bye	Pavel