Yup, it's definitely fedora's fault, not my hardware.

[Jeff Vian]

On Mon, 2004-11-22 at 22:40 -0500, Paul Tomblin wrote:
> On Mon, 22 Nov 2004 20:43:48 -0600, Randy <toucan at tropicalrain.us> wrote:
> >     The fix was to set the drive on the secondary IDE controller to
> > 'master' instead of 'cable select'.  What is interesting is that the
> 
> Interesting.  I never trusted "cable select" because I go back a long
> way to when it was considered very unreliable, so until recently I
> always set up systems with one master and one slave.  When I took the
> hard drive off of IDE1 and put it onto IDE0, I noticed that hda was
> already jumpered CS, (cable select) but hdc was jumpered MA (master). 
> I changed to to CS when I put it on IDE0 (and made it into hdb).
> 
> One interesting thing, which I wouldn't think would matter but it
> might, is that the CD-ROM drive is jumped as SL (slave), but it's at
> the *end* of the cable, where if it were jumped CS it would it would
> be the master.  Tomorrow if I have some time I'll try changing the
> CD-ROM to cable select to see if that helps, but I don't see how it
> could since both the hard drives are now on the other controller.
> 

I have often seen flaky drive performance when a drive jumpered as
either master or slave is used on a cable select cable.

My rule of thumb (and I have always had good results with it) is that if
the cable is a cable select cable then jumper the drives for cable
select.  If the cable is not cable select then jumper the drives as
master/slave.