[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: I am confused



Sam writes:
> The way that you're told to do it can create a fragmented journal file
> which would, probably, suffer under load.  From past experience I knew how
> to create a largely unfragment journal file:
> 
> dd if=/dev/zero of=journal.dat bs=20M count=1
> 
> That's going to create a 20 meg journal file in one fell swoop, in a
> single write().

Actually, there is nothing in the current ext2/ext3 code or the kernel
that would make this any less fragmented than doing 2000 1k writes, unless
you have a lot of other filesystem activity going on.  There has been some
talk about adding delayed file allocation to the VFS in 2.5, but for now
the ext2/ext3 file allocation will behave the same way - pre-allocating
a fixed number of blocks for each file.  The only thing that will affect
this is how your filesystem is layed out.  Obviously, it is best to create
the journal on an empty fs, but this isn't possible when converting your
existing filesystems to ext3.

Cheers, Andreas
-- 
Andreas Dilger  \ "If a man ate a pound of pasta and a pound of antipasto,
                 \  would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/               -- Dogbert





[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]