Strange HDD order

Erik Trulsson ertr1013 at
Sun Feb 3 06:59:42 PST 2008

On Sun, Feb 03, 2008 at 02:34:48PM +0100, Christian Baer wrote:
> On Sun, 3 Feb 2008 14:03:33 +0100 Erik Trulsson wrote:
> >> D 0 WDC WD3200SD-01KNB0 08.05J08 (ad4)
> >> D 1 WDC WD3200SD-01KNB0 08.05J08 (ad6)
> >> D 2 SAMSUNG HD501LJ CR100-11     (ad8)
> >> D 3 SAMSUNG HD501LJ CR100-11     (ad10)
> >> D 4 Seagate ST3500320AS SD04     (ad12)
> >> D 5 Seagate ST3500320AS SD04     (ad14)
> >> 
> >> The device in brackets is the one I'd expect to get. Instead, I get this:
> >> 
> >> ad4: 476940MB <SAMSUNG HD501LJ CR100-11> at ata2-master SATA300
> >> ad6: 305245MB <WDC WD3200SD-01KNB0 08.05J08> at ata3-master SATA150
> >> ad8: 476940MB <SAMSUNG HD501LJ CR100-11> at ata4-master SATA300
> >> ad10: 305245MB <WDC WD3200SD-01KNB0 08.05J08> at ata5-master SATA150
> >> ad14: 476940MB <Seagate ST3500320AS SD04> at ata7-master SATA150
> >> ad18: 476940MB <Seagate ST3500320AS SD04> at ata9-master SATA150
> >
> > How disks ars numbered do depend on which ports on which controller they
> > are attached to.  You want to use the lowest-numbered ports on the Promise
> > controllers.  Which ports that is when both controllers identify as a single
> > controller is another question.  From what you show it looks like all the
> > odd-numbered ports are on one controller and all the even-numbered ports are
> > on the other.
> Do you mean physically or logically? Physically the first four drives in
> the top list (the two WDCs and the two Samsungs) are connected to the
> first controller. The two Seagates are connected to the second.

In that case my guess on how they were numbered seems wrong.
FreeBSD seems to think that the two Samsung drives are connected to ports
1 and 3 on the first controller, while the two WDC drives are connected to
ports 2 and 4 on the same controller.
The two Seagate drives appear to be connected to ports 2 and 4 on the second
controller, with ports 1 and 3 on that controller being unused.

This may or may not be the same numbering of the ports as shown in the
manual for the controller (and possibly printed on the card itself.) If it
is the same, then all is fine, and working just as it should.
If it is different then there is a bug somewhere.  If this bug
is in the controller, the manual for the controller, or FreeBSD's ATA
driver is not something I can answer. 

Using Google to search for 'Promise SATA300-TX4 "port numbers"' indicate
that this is a problem that several other people have encountered with this
controller with both FreeBSD and LInux.
The ports on the card are labeled 1-2-3-4 but are detected as 3-2-4-1

> > Try experimenting by changing which disk is connected to which port and see
> > if you can get the behaviour you want.
> I don't really want or nead any behaviour in particular. This is just a
> matter of what's in the fstab. I was more wondering how this could be and
> if this order (and thus the device names) will stay constant or if the OS
> could get funny ideas after a csup and change this again, which would
> certainly mess up my directory tree.

It should stay constant unless there would be some change specifically in
how the ATA driver enumerates the ports on this controller.

> I was also wondering about the "gap" there is in the device names. The
> first controller should get ad4,6,8 and 10. But then, the second should
> get 12 and not 14. What happened to 12?

12 is probably one of the two ports you are not using. (16 being the other.)

<Insert your favourite quote here.>
Erik Trulsson
ertr1013 at

More information about the freebsd-questions mailing list