Re: ext4 and reiser4 availability

["Jarod Wilson" <user=74@nabble.blu.org>]

 On Thu, 2008-05-01 at 10:43 -0400, David Rosenstrauch wrote: 
> Mark J. Dulcey wrote: 
> > reiserfs does have some advantages. Even the existing version (reiser3) 
> > outperforms ext2,3,4 on directories with a lot of files, so it's a good 
> > choice for (say) a mail server using maildirs or an NNTP server. reiser4 
> > extends that advantage and adds space efficiency for small files by 
> > packing multiple small files into a single disk block. That's not as big 
> > a deal as it used to be now that disk space costs 20 cents per gigabyte, 
> > but it could matter if you were trying to implement a WinFS-like vision 
> > of file system as the ultimate database. Finally, reiserfs doesn't have 
> > a fixed inode limit; you don't have to worry about configuring your file 
> > system correctly for the mix of files you expect to have, it's all 
> > automatic. The defaults for extN are reasonable for many systems, but on 
> > a mail server you run out of inodes before you run out of space, and on 
> > a media server you waste a bunch of space unnecessarily on inodes you 
> > won't use. 
> 
> I also like reiserfs for the exact reason you mention above. 
> 
> One very important - and often overlooked - negative about reiserfs, 
> though:  the support tools suck!  I've probably had 3 or 4 separate 
> occasions where my reiser fs got damaged, and when I attempted to fix 
> the problems using reiserfsck (which, IIRC, explicitly warns you that 
> it's experimental) it trashed the file system even more, resulting in 
> losing a lot of data. 
> 
> Since then I've been trying to restrict myself to only using reiserfs 
> for things that benefit from its strengths - i.e., apps that require 
> large numbers of files, lots of small files, lots of fast i/o, etc. 
> 
> Word to the wise ...