From: Bjoern F. <li...@fe...> - 2003-11-26 18:51:12
|
> On Wed, 26 Nov 2003, Jeff Pratt wrote: > > > > From my experience, the hardware is not likely to have caused the > > > problems. I suggest you check your I/O by copying some stuff locally. > > > If that gives any reasonable performance, the problem is down to: > > > > > > -netatalk itself > > > -redhat's implementation of netatalk > > > -your configuration error of netatalk > > > > > > :) > > > > I'll take option number 3! It seems that in my situation, the > > --with-flock-locks configure option is a performance killer. I'm now > > experiencing a ~5% hit on some operations (most likely due to > the journalled > > fs), but that's something I can live with. > > It has recently been suggested on this list (or maybe on netatalk-devel) > that --with-flock-locks is broken and should be removed as an option > altogether. netatalk 1.6.3 shouldn't even compile with --with-flock-locks, there's an #error directive somewhere in the code. flock-locks have been removed in CVS and will not be in 1.6.4 anymore. Bjoern |