First, thank you for such a quick and clear response.

I have to check abou the DMA, but if it is on and causing the problems, what
are my alternatives or options? I need fast disk access as my application is
extrememly heavy on disk access, lots of video recording and playback. Is
there a way to check if it is a disk problem, or perhaps a chipset or other
thing? Patches?

The disk isn't getting that hot, this can happen even during a regular
maintanence boot without running the heavy application.

Do lilo parameters perhaps have something to do with the problem? I am using
new etx's and new disks, but I am still setting in LBA32, is that right? I
am not specifying any other special parameters.

I have considered asking my company to switch model disks, would this be a
good idea? Does anyone know anything about this model drive? (Maxtor
Diamondmax Plus 8 40 GB)?

Thanx again,
nachum


-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
On Behalf Of Ez-Aton
Sent: Tuesday, April 20, 2004 4:31 PM
To: Nachum Kanovsky
Cc: [EMAIL PROTECTED]
Subject: Re: hard drive errors - bad disk or?


Hi Nachum.
Drive seek errors are caused due to either bad-sectors, or DMA usage, where 
DMA is either not supported, buggy, or doesn't do the job (generally, when
it 
fails, but it is not one the previous options).
I know RH do not activate DMA usage by default, due to problems similar to
the 
one you have described. 

However, if you happen to get "I/O error... ... sector xxxxx", it means that

this sector might be faulted. Usually, bad sectors are consistent, aka, they

remain bad forever, but there are two issues which might lead to a passing 
disk-test, or lead to a "usually working system with BS reports, once a 
while". They are:
1) Auto repair function of the disks. They don't actually repair the BS, but

they try to checksum it, and make sure the data remains valid. Working some 
of the time, if you happen to have a real bad sector.
2) Over-heat of the disk. This will (I have experianced such a problem in
the 
past) cause bad-sectors "look-alike" errors, I/O problems, etc. You should 
make sure the disk(s) are not too hot to touch. Might get critical on
certain 
days/tasks, and might lead to a real disk failure.

So, to sum things up, what do we have?
It's either bad-sectors, or over-heat of the disk(s), which, in turn,
_would_ 
lead to bad sectors and dead disks.

Good luck.

Ez.

On Tuesday 20 April 2004 10:33 am, Nachum Kanovsky wrote:
> Please help...
>
> I have a project with a number of mirrored disks. I mirror them by 
> running fdisk, mkswap, mke2fs, and cp -ax to the new disk. The disks 
> are Maxtor DiamondMax Plus 8 40GB disks. I let the BIOS autodetect 
> them, and then I let linux do the same, so i am not giving any special 
> parameters for the fdisk to create the partitions. I am running Debian 
> Unstable, and using Lilo to boot. The board that we are running on is 
> a custom made board, running with an Advantech ETX, and our own 
> motherboard (ie none of the hard drive logic or controlling chips were 
> done by us, but the hd cable does connect through our board to the PMC 
> connector on the ETX).
>
> On some of the disks I get errors that make me think there is a 
> physical
> problem:
>
> end_request: I/O error, dev, 03:02 (hda), sector xxxxxxx..
>
> I have also gotten 'DriveReady SeekComplete Error' errors, but I don't 
> have the exact error to give at the moment.
>
> I have on some of the disks ran e2fsck with a non-destructive physical 
> check, and I have found no errors. What else can this be? Is there a 
> more intensive way to check the disk, can this error be due to a 
> cable? Might this be due to bad parameters when creating the 
> partitions? I have been trying to deal with this error for almost half 
> a year now, and I have searched the internet quite a bit, but I have 
> not found anything which has explained this for me.
>
> Nachum Kanovsky
> Software Developer
> Mango DSP Ltd.
> E-mail: [EMAIL PROTECTED]
> Work: +972 2 588 5039
> Cell: +972 67 508 121
>
>
> To unsubscribe, send mail to [EMAIL PROTECTED] with the 
> word "unsubscribe" in the message body, e.g., run the command echo 
> unsubscribe | mail [EMAIL PROTECTED]

=================================================================
To unsubscribe, send mail to [EMAIL PROTECTED] with the word
"unsubscribe" in the message body, e.g., run the command echo unsubscribe |
mail [EMAIL PROTECTED]


================================================================To unsubscribe, send 
mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]

Reply via email to