Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 2 Mar 2009 13:52:16 -0600
From:      Josh Paetzel <josh@tcbug.org>
To:        Scott Long <scottl@samsco.org>
Cc:        freebsd-scsi@freebsd.org
Subject:   Re: twa driver and 3ware 9690SA issues
Message-ID:  <49E31E02-E9B0-48EB-8B7A-0517C6501FC2@tcbug.org>
In-Reply-To: <49AC370A.1040700@samsco.org>
References:  <09CE3D22-431A-433F-9CAF-6896FF77DDB1@tcbug.org> <49AC370A.1040700@samsco.org>

next in thread | previous in thread | raw e-mail | index | archive | help

On Mar 2, 2009, at 1:44 PM, Scott Long wrote:

> Josh Paetzel wrote:
>> This is somewhat of a repost from questions@, and I'm currently  
>> engaged with 3ware about this issue.
>> I have several 3ware 9690SA SAS RAID controllers.   This item is  
>> very similar to their 9650SE controller.  It uses the same firmware  
>> and driver, the difference being it can handle SAS drives.  Up  
>> until this point all of the arrays we've used with these drives  
>> have been RAID 1 arrays with 7200 RPM SATA drives, but recently  
>> we've started using 15,000 MBA series Fujitsu SAS drives in places.
>> The arrays have always been detected as:
>> da0: <AMCC 9690SA-4I  DISK 4.06> Fixed Direct Access SCSI-5 device
>> da0: 100.000MB/s transfers
>
> This 100.000MB/s number is fictitious.  The driver is just filling  
> in a
> dummy value.  Filling in the real number would require the driver  
> being
> able to query the card for the negotiated bus speed.  I have no idea
> if that's possible for TWA given the programming information that is
> public for the card.
>
>> Which wasn't that much of an issue, as SATA drives aren't capable  
>> of sustained sequential 100 MB/sec transfers anyways, but the SAS  
>> drives we are getting are supposedly capable of 180 Megs/sec and  
>> I'm not seeing it.  I'm unsure of how to eliminate caching from the  
>> equation, simple tests like dd if=/dev/zero of=testfile bs=8m  
>> count=1000 seem to support what dmesg reports by returning 96 Meg/ 
>> sec transfer rates
>
> FreeBSD limits individual I/O segments through the CAM subsystem to
> 64K.  It can safely be raised to 128K via some patches that I have,  
> and
> you will see a corresponding improvement in performance.  I don't
> believe that it can be safely raised above that for the TWA  
> controllers.
>
> Scott



I'd like to try the patches.  Where can I get them from?

Thanks,

Josh Paetzel







Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?49E31E02-E9B0-48EB-8B7A-0517C6501FC2>