On Tue, Jun 10, 2008 at 11:02:02 -0500,
Eric Sandeen <sandeen(a)redhat.com> wrote:
md for example passes barriers on raid1 but not 0 or 5 AFAIK...
I have heard that, but not verified it myself. I am not sure why this is the
case as I wouldn't expect doing raid 1 to be that much easier than other
types of raid.
Got a url?
This is a LWN article pointing to the start of a lmkl thread with the
title: "Proposal for "proper" durable fsync() and fdatasync()"
http://lwn.net/Articles/270891/