Johnny Hughes wrote:
> On Mon, 2006-03-13 at 08:44 +0100, Peter Kjellstr?m wrote:
>
>> Hello,
>>
>> I had a very similar problem with a bunch of supermicro boards (X6DVL-EG2).
>> The problem was that the bios somehow confused libata so badly that it fell
>> back to using the old ide legacy whatever thingy (resulting in hda instead of
>> sda and of course no dma).
>>
>> The solution was to fiddle with the sata settings in bios. Strangely enough,
>> enabling mixed mode (pata and sata) in bios resulted in something the kernel
>> could understand and libata has been happy ever since (reporting combined
>> mode operation in dmesg).
>>
>>
>
> cap makes a very good point. All the sata drives I use are registered
> as scsi drives and not ide drives (sda and sdb NOT hda and hdb).
>
> try our new 2.6.9-34.EL kernel.
>
Hmmm...I'd love to try the new kernel, but this machine is running 3.6.
8-( Is there a recommended procedure for doing this with a 3.6 box?
> also make sure the motherboard has the latest bios from the
> manufacturer.
>
The machine is located about 100 miles away so I'm planning to visit it
sometime in the next few days to do that and to fondle the SATA settings
per Peter's suggestion. I only knew something was amiss when my
remaining P3-733mhz machines were doing disk to disk backups at 4-5
times the speed of these relatively new P4 systems. 8-) Nobody ever
noticed before because these machines don't do anything very disk
intensive (except backups).
Cheers,
