On Wed, Jan 17 2001, Russell Cattelan wrote:
> > On Tue, Jan 16 2001, srn@xxxxxxxxxxxx wrote:
> > > As Jack noted, the temp patch mentioned in this pv has
> > > been applied to the snia tree for now as a short term
> > > fix for QA SNIA testing. This bug should probably not
> > > have been closed since there's another more complete
> > > fix in the works according to anath and chait. --steve
> >
> > Which bug was this specifically?
>
> This is the infamous elevator starvation problem
> For XFS this means hammer on one part of the disk
> and log requests to a completely different part of the disk
> will get starved out.
Exactly, it's been known for some time that the sequencing
in the stock kernel was much too high. dbench tuned :-)
> BTW what is the status of this patch in regards to the
> actual 2.4.x tree?
> Somebody mentioned one of the 2.4.1-acX had it included?
It was integrated in -acX yes, and Linus also took it now.
2.4.1-pre5 has blk-13C integration, pre6 adds a couple of
further optimizations.
--
* Jens Axboe <axboe@xxxxxxx>
* SuSE Labs
|