On Wed, Aug 01, 2001 at 06:53:10PM -0700, Jeffrey Siegal wrote:
> Jay R. Ashworth wrote:
> > But you see, if the sales pitch is "journalling is worth the speed hit
> > because it protects your *data*", then they're lying. It only protects
> > your file system structures.
>
> It *can* protect your data, when used in connection with application
> programs which do so.
>
> For example, some application programs are written to rely on the
> rename() system call being atomic. They'll write out a new, consistent
> version of a file, and then rename the new version over the new version.
> As long as the file system remains consistent, then the application
> data will also be consistent as long as the file system guarantees
> consistency.
Unless the machine dies in the middle of a write to one of the files.
Oh yeah, the filesystem will be consistent, but you know what? I don't
know many suits who give a crap. They want data *not to be lost*, and
it is my perecption that they're being led to believe that journalling
filesystems do this.
> Other applications may make careful use of O_SYNC or fsync to ensure
> consistent data, again assuming that the file system itself remains
> consistent.
This is closer; at least the app knows if the data got written... if
it's still running.
> A journalling file system takes you part of the way to having consistent
> data; the applications you use can take you the other part.
Note that I had no questions about "consistent"...
Cheers,
-- jra
--
Jay R. Ashworth [EMAIL PROTECTED]
Member of the Technical Staff Baylink RFC 2100
The Suncoast Freenet The Things I Think
Tampa Bay, Florida http://baylink.pitas.com +1 727 804 5015
"So easy to use, no wonder the Internet is going to hell!"
-- me
--
To unsubscribe from this list, send a message to [EMAIL PROTECTED]
with the command "unsubscribe linux-embedded" in the message body.
For more information, see <http://waste.org/mail/linux-embedded>.
