Sounds like Rik has a good handle on this. I'm certain he'll have us a
fix out fairly quickly.
Richard Gooch wrote:
> Jeff V. Merkey writes:
> > Dirty pages should be flushed as fast as possible (they should become
> > eligible after 2 seconds to being flushed out to disk). Pre-6 does have
> > some problems.
> > >
> > > Hi, all. I know there's been grumbling about VM/IO/LRU's/page aging
> > > recently. To give an idea of how bad it is:
> > >
> > > I have a drive that is capable of delivering >12 MB/s. By using dd on
> > > the block device (reading in more data than I have RAM), I quickly get
> > > to a state where I'm getting << 1 MB/s. My guess is that I'm getting
> > > 100 kB/s or less. Once this happens, it seems easiest to reboot to
> > > bring things back to a sane state.
> > >
> > > It seems the current LRU algorithm is "Least Recently Used pages will
> > > remain in RAM the longest". I'd call that a BFULRU.
> These pages aren't dirty. I was doing
> # time dd if=$device of=/dev/null bs=1024k
> Permanent: email@example.com
> Current: firstname.lastname@example.org
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to email@example.com
> Please read the FAQ at http://www.tux.org/lkml/
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to firstname.lastname@example.org
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun May 07 2000 - 21:00:12 EST