Re: [RFC PATCH] ext3 writepage() journal avoidance

[Dave Jones <davej@redhat.com>]

On Fri, Mar 10, 2006 at 08:40:47AM -0800, Badari Pulavarty wrote:

 > >  > > I don't know how much usage it's had, sorry.  It's only allowed in
 > >  > > data=writeback mode and not many people seem to use even that.
 > >  > 
 > >  > would you be prepared to turn it on by default in -mm for a bit to see
 > >  > how it holds up? The concept seems valuable in itself, so much so that I
 > >  > feel this should be 1) on always by default when possible and 2) isn't
 > >  > really the kind of thing that should be a long term option; not having
 > >  > it almost is a -o pleaseAddThisBug option for each bug fixed.
 > > 
 > > It'd be good to get that hammered on, as it doesn't see hardly any testing
 > > based upon the experiments I did sometime last year.  It left me with
 > > an unmountable root filesystem :-/
 > 
 > Yuck. You are talking about "nobh" option for writeback mode, correct ?
 > Have any idea on what you were doing ?

Actually, I think I may have neglected to make those mounts writeback.
In retrospect, it was silly, I basically forced nobh on for all mounts.
A few boots later my / reached its maximum mount count, and got a fsck,
which moved a bunch of useful things like /lib/ld-linux.so.2 to lost+found.
There was so much mess that it was easier to reinstall the box than
to pick through it. (Thankfully I tested it on a scratch box ;-)

		Dave
-- 
http://www.codemonkey.org.uk