Quoting Gilad Ben-Yossef, from the post of Wed, 12 Jan: > > Look at this: > > http://seclists.org/lists/linux-kernel/2005/Jan/0641.html > > Am I crazy or does it look that the journal code that ext3 uses never > took into account a situation where a journal commit can fail?
haven't looked into the code yet, but a quick glance at the patch shows some overlooks in buffer handling not only in ext3. I'm left to wonder who else is in danger? what about reiser3 and reiser4 for instance? > > It sounds completely ridiculous to think that a system like Linux that > the likes of Oracle recommends to use with it's multi tera bytes and > dollars wont stand a simple shutdown of the SCSI disk... ?! when you run a tera dolar installation (you'd wish!) you have lots of backup mechanisms :-) this does not seem to effect OCFS or Raw devices which are Oracle's latest recommended modes of data access, plus if you have reliable external RAID arreys, it also should never happen. either way, this is an annoying truth to learn indeed! > I didn't see any reaction to this on LKML - anyone can figure out why > it's not as bad as it looks? no idea. I'm surprised as well. maybe write the poster and ask him if anyone answered off list? > > (sorry for the sensational tone and title, I'm simply shocked...) copy that. -- Flaming heterosexual Ira Abramov http://ira.abramov.org/email/ ================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
