[CentOS] Software RAID1 with CentOS-6.2

Tue Mar 6 03:16:41 UTC 2012
Nataraj <incoming-centos at rjl.com>

On 02/28/2012 07:26 PM, Kahlil Hodgson wrote:
> Hi Emmett,
>
> On Tue, 2012-02-28 at 18:18 -0800, Emmett Culley wrote:
>>   I just had a very similar problem with a raid 10 array with four new
>> 1TB drives.  It turned out to be the SATA cable.
> ...
>
>> All has been well for a week now.
>>
>> I should have tired replacing the cable first :-)
> Ah yes. Good point. That was one of my first thoughts. I forgot to
> mention that I swapped both cables between the first and the second
> install:-)
>
> Cheers,
>
> Kal
>
>
> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> http://lists.centos.org/mailman/listinfo/centos

You can also try swapping the drives.  I bet the problem won't move. 
I've seen problems like this and I don't believe that your IO errors are
real errors on the disk.  I think something is going offline.  It kicks
the second drive out of the array because it is trying to read the first
drive and sync by copying it to the second drive.  Since it can't resync
it kicks the drive which it thinks is not 'current' out of the array.  I
don't think it is a raid problem, your just seeing raids reaction to a
problem somewhere between the disk controller and the drive.

Is your kernel using the AHCI driver and if not which driver is being
used?  Do you know if you have an Intel controller onboard?  If your not
getting the AHCI driver and your using an onboard controller, your BIOS
may not be setup properly for native AHCI.

Do you actually have a two channel controller or is there any kind of
port multiplier on there?  I have seen problems like you describe with
some port multipliers.  In fact I'm having a problem like that right
now.  You could test for problems like this by installing the system on
just one drive with the other disconnected and then try the second drive
with the first disconnected.  Obviously you won't be able to configure
raid, but you should be able to do substantial IO testing.

You can tell which driver is being used by looking at the log output,
also lsmod to see which driver is loaded.

Mar  5 16:06:33 myhost kernel: ahci 0000:00:1f.2: PCI INT A -> GSI 20
(level, low) -> IRQ 20
Mar  5 16:06:33 myhost kernel: ahci 0000:00:1f.2: AHCI 0001.0300 32
slots 6 ports 3 Gbps 0x3f impl SATA mode
Mar  5 16:06:33 myhost kernel: ahci 0000:00:1f.2: flags: 64bit ncq sntf
pm led clo pmp pio slum part ems sxs apst
Mar  5 16:06:33 myhost kernel: scsi0 : ahci
Mar  5 16:06:33 myhost kernel: scsi1 : ahci
Mar  5 16:06:33 myhost kernel: scsi2 : ahci
Mar  5 16:06:33 myhost kernel: scsi3 : ahci
Mar  5 16:06:33 myhost kernel: scsi4 : ahci
Mar  5 16:06:33 myhost kernel: scsi5 : ahci

Here's what you might see in log output (/var/log/messages) if you have
a port multiplier:

Mar  5 16:06:33 myhost kernel: ata6.15: Port Multiplier 1.1,
0x197b:0x2352 r0, 2 ports, feat 0x0/0x0
Mar  5 16:06:33 myhost kernel: ata6.15: Asynchronous notification not
supported, hotplug won't
Mar  5 16:06:33 myhost kernel:         work on fan-out ports. Use
warm-plug instead.
Mar  5 16:06:33 myhost kernel: ata6.00: hard resetting link

I am actually looking for any recommendations for port multipliers that
work well with the Centos 6 driver, because the one that I have here
doesn't.

Nataraj