Re: Lots of con-current I/O = resets SATA link? (2.6.25.10)

Gerhard Wiesinger <lists@xxxxxxxxxxxxx> · Tue, 8 Jul 2008 08:59:54 +0200 (CEST)

Hello!

BTW:
The error in syslog also accoured when Reallocated_Sector_Ct didn't go up.

Ciao,
Gerhard

--
http://www.wiesinger.com/

On Tue, 8 Jul 2008, Gerhard Wiesinger wrote:

On Mon, 7 Jul 2008, Justin Piszcz wrote:

Hi Gerhard,

It /could/ be the port itself if you have changed the cable and disk..

Yes, but it is very unlikely. I have written TB of data there without any 

problems. Anyway this is my 3rd exchanged SAMSUNG disk ...

Have you tried loading the disk with dd and seeing if you can reproduce the 

problem? You are getting the same error I get generally, I can recommend 

turning OFF NCQ first and see if the problem goes away.

# Define DISKS.
cd /sys/block
DISKS=$(/bin/ls -1d sd[a-z])

# Disable NCQ on all disks.
echo "Disabling NCQ on all disks..."
for i in $DISKS
do
 echo "Disabling NCQ on $i"
 echo 1 > /sys/block/"$i"/device/queue_depth
done

I tried to disable NCQ on all disks and tried to rebuild the raid, but it 

still failed to rebuild with the same error message.

I also tried the nolapic kernel parameter without success.

/dev/sda:  5 Reallocated_Sector_Ct   0x0033   100   100   010    Pre-fail 

Always       -       0

/dev/sdb:  5 Reallocated_Sector_Ct   0x0033   100   100   010    Pre-fail 

Always       -       0

/dev/sdc:  5 Reallocated_Sector_Ct   0x0033   091   091   010    Pre-fail 

Always       -       413

/dev/sdd:  5 Reallocated_Sector_Ct   0x0033   100   100   010    Pre-fail 

Always       -       0

/dev/sde:  5 Reallocated_Sector_Ct   0x0033   100   100   010    Pre-fail 

Always       -       0

/dev/sdf:  5 Reallocated_Sector_Ct   0x0033   100   100   010    Pre-fail 

Always       -       0

The only thing is that the Reallocated_Sector_Ct is still >0 on /dev/sdc 

(keep in mind this is my 3rd new Samsung disk on /dev/sdc and I had up to 

3000 Reallocated_Sector_Ct on previous disks in < 1 day !!!).

Should I replace the disk a fourth time?

When you search in google you find a lot of threads with the timeout problem. 

Might this be a software issue?

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html