On Sun, Mar 14, 2004 at 02:09:57PM -0600, Donald Lee wrote:
> >On Sat, Mar 13, 2004 at 01:49:38PM -0600, Bill Studenmund replied:
> >> 
> >> I *did* put both drives on the same ribbon cable, with one master and
> >> the other slave.  I've been abusing all three drives for about 40 minutes
> >> now, and have yet to see a single "lost interrupt" message in the
> >> log.
> >
> >Could you try the other cable, just for completeness?
> 
> The cable I have been using up to this
> time was the ribbon cable that came with the Sonnet card.  It's a "normal"
> grey ribbon cable with 3 connectors on it.
> 
> When I bought the 200G drive, I also bought a cable It's an 18", Belkin
> "round" cable.  They put the wires in a cute rubber tube with little
> "flares" on the connectors.)  I'll call this the yellow cable.
> 3 connectors.
> 
> I tried the yellow cable with both drives on it.
> 
> I tried it on the "near" channel.  (the one that I had been using with
> the single drive.  It's the one nearest to the front of the machine.)
> I also tried it on the "far" channel.  (nearer the back of the machine)
> 
> Both attempts had both drives connected master/slave on the yellow cable.
> For each, I heaped some abuse on it, and it performed flawlessly,
> with one exception.
> 
> When I put the drives on the "back" channel, it worked fine under heavy load
> for about 30 minutes.  When I went to shut down, I got a single "lost interrupt
> message.  The machine seemed to hang there, and I had to power it down.
> The "lost interrupt" I/O was apparently very late in the shutdown
> process, possibly when unmounting the drives, because the message is
> not in the syslog, and when I rebooted (power cycled) it did a full
> fsck of the disks.
> 
> I'm back on the "front" channel, on the grey ribbon cable.  Life is good,
> and my new disk seems to be working, and getting good performance.

Is your adapter manufacture by promise, or is it some other brand using
the same chipset ?

I've got troubles with a bad implementation of a chipset once (a HPT on
a motherboard , FWIW). It had lots of CRC errors and other problems on the
motherbord. The same drives, same cables connected to an add-on HPT
adapter using the exact same chipset (identical in PCI registers dumps and
on-chip label) worked without a problem.

The Ultra/133 I have here is from promise.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--