On Fri, Jun 03, 2005 at 12:40:52PM -0400, Perry E. Metzger wrote:
> 
> Sean Davis <dive-nb@endersgame.net> writes:
> > On Fri, Jun 03, 2005 at 05:53:17PM +0200, Frank van der Linden wrote:
> >> On Fri, Jun 03, 2005 at 11:42:12AM -0400, Perry E. Metzger wrote:
> >> > I didn't think Juergen was the author of NEW_BUFQ_STRATEGY, and I
> >> > don't remember anyone complaining about it in any way at all for
> >> > literally years now. Are you sure about this?
> >> 
> >> Hmm.. maybe it wasn't Juergen, but I remember a discussion on the internal
> >> developers mailing list.. I'll let others speak up who remember that.
> >
> > I don't recall exactly who said what, but I do know that every time I've
> > tried NEW_BUFQ_STRATEGY, once I put the system under some heavy load, any
> > interactive apps I have running come to a screeching halt (we're talking 30
> > second lag between a keystroke and the letter showing up in an xterm here).
> >
> > Please do not make it the default.
> 
> Er, that's the behavior I get when I *don't* use NEW_BUFQ_STRATEGY.
> Many others have the same complaint. That's why I use it, and that's
> why a lot of us would like it as the default.

Are you running with softdeps enabled, too? I've had softdeps and
new_bufq_strategy bring this system to it's knees when doing a lot of stuff
at once - but running without softdeps, without new_bufq_strategy,
everything works just fine. (I don't trust softdep in NetBSD; perhaps it's
just a -current thing, but it seems that every time I try it, and put the
system under serious load, it locks the machine and causes tons of data
loss.)

-Sean