Subject: Re: hard driver CRC error with RAID 5 under 2.0 release
To: Rui-Xiang Guo <rxg@myrealbox.com>
From: TYMYT <mig144ru@yahoo.com>
List: port-i386
Date: 12/16/2004 05:40:13
--- Rui-Xiang Guo <rxg@myrealbox.com> wrote:

> > In the dmesg you posted, wd2 gets errrors too.
> > 
> > ATA electrical specifications assumes 18 inches
> cables. With longer cables
> > you're out of specs, and all sort of things can
> happens.
> 
> I re-used the standard cables for all hard
> drivers.(I fixed the case ;))
> It looks fine with ATA 100 during boot. I'm tring to
> rebuild the RAID frame.
> I will report the result later...
> 
> > Note that another thing to look at is the power
> supply. With 8 hard drives,
> > it's possible that your power supply is too short.
> 
> There is a double power supply on the server. I
> think there is no problem
> about this.
> 
> Thanks for your reply!

Such errors intend to float and appear from time to
time. 
Personally I do not trust cable change because I have
seen cases when error comes back again.
If you look for driver code you may find:
WDCE_CRC 0x80 - CRC error (Ultra-DMA only)
WDCE-BBK 0x80 - bad block detected
From this point I can assume that turning UltraDMA
mode off
would downgrade driver to avoid possible errors and
save appropriate bad block checks.

If any one thinks different - welcome, this is
important issue. 

Vyacheslav Karpenko


		
__________________________________ 
Do you Yahoo!? 
Yahoo! Mail - 250MB free storage. Do more. Manage less. 
http://info.mail.yahoo.com/mail_250