On Thu, May 15, 2003 at 04:39:04PM +0100, David Brownlee wrote:
> On Thu, 17 Apr 2003, Manuel Bouyer wrote:
> 
> > On Thu, Apr 17, 2003 at 02:29:56PM +0100, David wrote:
> > > 	I'm getting various 'pciide0:1:0: lost interrupt' errors on two
> > > 	boxes under load. They are running raidframe on 1.6.1.
> > > 	Running bonnie++ is a great way to trigger a massive load of them.
> > > 	I'm not seeing any problems on a variety of other i386 boxes
> > > 	running raidframe including a virtually identical box (PII 233 rather
> > > 	than 266). These two are external DNS and gateway boxes hence the
> > > 	small disks :)
> >
> > Did you do the usual checks, especially, did you check that the
> > cables aren't too long ?
> > Also, did you try disconnecting the ATAPI CD ?
> 
> 	The cables are quite short and the drives are quite old.
> 	Disconnecting the CD seems to have helped, but I'm confused
> 	why there would be 'lost interrupt' on both wd0 and wd1 when
> 	they were on different channels, and the CD (on channel 0) was
> 	never accessed?

The CD could create problems on the IDE bus it's connected to.
I've already have troubles with incompatible ATA hardware.

I don't know why it would cause problems on the second channel, but it's
still the same chip. Who knowns what happen inside it.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 24 ans d'experience feront toujours la difference
--