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

Re: ext3/kjournald overhead



Hi,

On Tue, Nov 06, 2001 at 03:10:59PM -0800, Mike Fedyk wrote:

> You can try a larger journal file.  You'll have to remove the current
> journal flag, fsck, and recreate a larger journal...
> 
> There are three modes you can choose, ordered (default), data (full data
> journal), or writeback (possibly fastest, but no guarantees that ext2 didn't
> also).

No, writeback offers enormously greater guarantees than ext2: it
guarantees full metadata consistency at all times without any guessing
after a crash.  What it doesn't guarantee is when the data blocks
get written to disk --- ordered and journaled modes ensure that data
is written ahead of metadata when necessary so that you never see
stale disk blocks in a newly-created file after a crash.

Cheers,
 Stephen





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