Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 24 Feb 2000 16:04:50 +0100
From:      Brad Knowles <blk@skynet.be>
To:        freebsd-stable@FreeBSD.ORG
Subject:   Re: Multiple Adaptec 2940U2W controllers?
Message-ID:  <v0422080cb4daf715f6bf@[195.238.1.121]>
In-Reply-To: <v04220834b4da0a913e9e@[195.238.1.121]>
References:  <v0422081fb4d9d7f862b6@[195.238.1.121]> <v04220834b4da0a913e9e@[195.238.1.121]>

next in thread | previous in thread | raw e-mail | index | archive | help
At 11:11 PM +0100 2000/2/23, Brad Knowles wrote:

>  	I have switched to a non-SMP kernel (otherwise identical to what
>  I was previously running), I have tried moving both of the 2940U2W
>  controllers to either the primary or secondary PCI busses (on the
>  primary, one of them shared IRQs with the AIC-7890 controller, but
>  on the secondary although they share the PCI bus with the AIC-7890
>  controller they all at least have different IRQs and I/O Base
>  Addresses).  So far, nothing has worked.

	After working with the experts from Germany and the local field 
engineers, it appears that the problem was within the drive array 
configuration.  It has two controllers (with one UltraSCSI 2 LVD 
interface each), and the two logical units on the array are 
accessible via both of them and visible to the host as two different 
devices.

	Although I was mounting only one each of the two logical units, 
it appears that control over these LUs was ping-ponging back and 
forth between the two controllers, thus trashing my throughput.


	Now that the problem has been fixed, initial preliminary test 
results indicate that I do actually see almost 2x speedup using two 
host SCSI controllers to the two different drive array controllers. 
I'm now moving one of the host SCSI adaptors back to the other PCI 
bus, to see if that helps, hurts, or makes no difference to the 
performance.

	If it doesn't hurt, I plan on leaving them on separate PCI busses.


	Afterwards, we're going to delete and recreate the logical units 
on the controller, so that we have a separate device for each column 
of disks (half of which will be made available through each 
controller), and then we're going to try software striping across the 
columns.

	This ought to be fun!  ;-)

-- 
   These are my opinions and should not be taken as official Skynet policy
  _________________________________________________________________________
|o| Brad Knowles, <blk@skynet.be>                 Belgacom Skynet NV/SA |o|
|o| Systems Architect, Mail/News/FTP/Proxy Admin  Rue Col. Bourg, 124   |o|
|o| Phone/Fax: +32-2-706.13.11/726.93.11          B-1140 Brussels       |o|
|o| http://www.skynet.be                          Belgium               |o|
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
     Unix is like a wigwam -- no Gates, no Windows, and an Apache inside.
      Unix is very user-friendly.  It's just picky who its friends are.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?v0422080cb4daf715f6bf>