More about the 9500S-4 issues with the MoBo and FreeBSD...

Ken Gunderson kgunders at teamcool.net
Tue Nov 29 14:08:11 GMT 2005


On Tue, 29 Nov 2005 05:18:12 -0800
ray at redshift.com wrote:

> At 01:25 PM 11/29/2005 +0100, Olaf Greve wrote:
> | Hi,
> | 
> | A little bit of further information (which I just checked on the machine 
> | when I went home for lunch):
> | 
> | I identified the Texas Instruments device. It's not the memory (which of 
> | course makes sense), but rather, it seems to be a FireWire controller.
> | (TSB43AB22/A IEEE-1394a-2000 Controller (PHY/Link) (prog-if 10 [OHCI]) ).
> | 
> | When considering the following FreeBSD output:
> | twa0: <3ware 9000 series storage controller> port 0xa000-0xa0ff mem 
> | 0xd8005000-0xd80050ff, 0xd2000000-0xd27fffff irq 16 at device 6.0 on pci5
> | 
> | And:
> | fwohci0: <Texas Instruments TSB43AB22/A> irq 16 at device 11.0 on pci5
> | 
> | Could this perhaps be the culprit???
> | 
> | I don't recall having seen any way to disable FireWire on the MoBo, but 
> | perhaps there is...
> | 
> | Also, I checked the chipset: it's nForce 4.
> | 
> | Hopefully this additional information contributes to identifying the 
> | clashing devices...
> | 
> | Cheers,
> | Olafo
> 
> Olafo,
> 
>  Did you remove all the cards and clear the CMOS on the motherboard yet?  There
> should be a jumper you short.  That will reset all the IRQ stuff.  Then disable
> everything in the BIOS you can.  Then plug in the 3ware card.  That might shift
> around the IRQ's.  You almost might change the setting about "Plug and Play OS"
> to see if that has any impact.

Try setting "Plug and Pray OS" to "No".
 
>   You might also set an IRQ in the BIOS as reserved and see if that makes any
> difference.  I don't know the exact impact of that as it relates to the
> motherboard, but it's possible that will keep motherboard devices (such as the
> fire wire stuff) off those channels and allow them to be used by cards?  I don't
> know, but it's worth trying.
> 
>   I know you aren't going to like to hear this, but perhaps a motherboard brand
> chance is in order?  In the long run, it may save time/money.  In my experience,
> a big part of building a good, stable server is to buy the right hardware out of
> the gate.  For AMD CPU's, I only use Tyan here.  I run the Tyan boards on my
> servers and Workstation and they are some of the most stable boards I have ever
> used.

IMHO the Asus was a sub optimal choice for a server board.  I like some
of their boards for workstation use but for AMD servers, w/the exception
of 1 machine, have had pretty good results w/Tyan barbone units in the
Transport line.  Super Micro would also make my short list.  The Sun
x2100 that's been discussed recently may be an attractive candidate
pending on how well other's experiences w/it prove out. The mobo may
still be new enough to RMA in wh/case I'd go w/a non nVidia based Tyan
server board.  Yes, I know the nVidia stuff is supported much better
under 6.0 than 5.4 but it's still not as mature as the AMD chipsets so
I'm still staying away form them for a while yet for applications like
webservers where rock solid stability is a concern 

Olafo-- refresh my meory on this- what are the disk config requirements?

-- 
Best regards,

Ken Gunderson

Q: Because it reverses the logical flow of conversation.
A: Why is putting a reply at the top of the message frowned upon?



More information about the freebsd-amd64 mailing list