Newbiw question on write barriers and ordered write barriers

Newbiw question on write barriers and ordered write barriers

[Roger James]

[-- Attachment #1: Type: text/plain, Size: 1208 bytes --]

Hi,

 

Ever since I updated my SuSE Linux 8.1 set up with the latest patches
from SuSE I have been seeing the following messages at boot up time
(once only for each drive).

 

<4>hda: drive_cmd: status=0x51 { DriveReady SeekComplete Error }

<4>hda: drive_cmd: error=0x04 { DriveStatusError }

<4>hda:saw err in ide_end_drive_cmd, disabling write barriers

<4>hda: drive_cmd: status=0x51 { DriveReady SeekComplete Error }

<4>hda: drive_cmd: error=0x04 { DriveStatusError }

<4>hda:saw err in ide_end_drive_cmd, disabling write barriers

<4>reiserfs: using ordered data mode

<4>Using r5 hash to sort names

 

I have struggled to find out whether I should be worried about this. I
guess this is something to do with journaling and ordered writes, but
the messages are somewhat cryptic. Is this the kernel detecting a
problem with the drive based controllers (they are both Maxtor 6Y120L0s)
or the motherboard chipset (Nvidia NForce) and disabling the
functionality? I don't see the message again, but one of the reasons I
did upgrade the kernel was that I was having occasional disk
corruptions. Are there any switches/parameters I should be setting?

 

Can anyone give me a heads up on this.

 

Roger

Re: Newbiw question on write barriers and ordered write barriers

[Alexander Lyamin]

Mon, Sep 01, 2003 at 09:06:37PM +0100, Roger James wrote:
> Hi,
> 
>  
> 
> Ever since I updated my SuSE Linux 8.1 set up with the latest patches
> from SuSE I have been seeing the following messages at boot up time
> (once only for each drive).
> 
>  
> 
> <4>hda: drive_cmd: status=0x51 { DriveReady SeekComplete Error }
> 
> <4>hda: drive_cmd: error=0x04 { DriveStatusError }
> 
> <4>hda:saw err in ide_end_drive_cmd, disabling write barriers
> 
> <4>hda: drive_cmd: status=0x51 { DriveReady SeekComplete Error }
> 
> <4>hda: drive_cmd: error=0x04 { DriveStatusError }
> 
> <4>hda:saw err in ide_end_drive_cmd, disabling write barriers
> 
> <4>reiserfs: using ordered data mode
> 
> <4>Using r5 hash to sort names
> 

To me and my expirience, it seeems like your hard drive about to fail.
And it seems like its about to fail in area where jdata stored.


do backup (dd_rescue) , then try to run vendor utilities which relocate
bad blocks ( i dont know if they have something like that for MAXTOR,
but for IBM/HITACHI it would be DFT utility, and for most of hard-drives
ONTRACK Tech. does something that matches DFT functionality.

> I have struggled to find out whether I should be worried about this. I
> guess this is something to do with journaling and ordered writes, but
> the messages are somewhat cryptic. Is this the kernel detecting a
> problem with the drive based controllers (they are both Maxtor 6Y120L0s)
> or the motherboard chipset (Nvidia NForce) and disabling the
> functionality? I don't see the message again, but one of the reasons I
> did upgrade the kernel was that I was having occasional disk
> corruptions. Are there any switches/parameters I should be setting?
> 
>  
> 
> Can anyone give me a heads up on this.
> 
>  
> 
> Roger
> 

-- 
"the liberation loophole will make it clear.."
lex lyamin

Re: Newbiw question on write barriers and ordered write barriers

[Chris Mason]

On Tue, 2003-09-02 at 07:48, Alexander Lyamin wrote:
> Mon, Sep 01, 2003 at 09:06:37PM +0100, Roger James wrote:
> > Hi,
> > 
> >  
> > 
> > Ever since I updated my SuSE Linux 8.1 set up with the latest patches
> > from SuSE I have been seeing the following messages at boot up time
> > (once only for each drive).
> > 

Which kernel are you running now?

> > <4>hda: drive_cmd: status=0x51 { DriveReady SeekComplete Error }
> > 
> > <4>hda: drive_cmd: error=0x04 { DriveStatusError }
> > 
> > <4>hda:saw err in ide_end_drive_cmd, disabling write barriers
> > 
> > <4>hda: drive_cmd: status=0x51 { DriveReady SeekComplete Error }
> > 
> > <4>hda: drive_cmd: error=0x04 { DriveStatusError }
> > 
> > <4>hda:saw err in ide_end_drive_cmd, disabling write barriers
> > 
> > <4>reiserfs: using ordered data mode
> > 
> > <4>Using r5 hash to sort names
> > 
> 
> To me and my expirience, it seeems like your hard drive about to fail.
> And it seems like its about to fail in area where jdata stored.
> 

The barrier patches try a write cache flush command on the drive, and if
it fails they turn barrier mode off.  These messages mean your drives
don't support the barrier mode, you can ignore them.  Some drives print
the failures if you've already got write cache disabled.  Try hdparm -W
1 /dev/xxx before mounting an FS on the drive.

-chris

Re: Newbiw question on write barriers and ordered write barriers

[Alexander Lyamin]

Tue, Sep 02, 2003 at 08:11:15AM -0400, Chris Mason wrote:
> On Tue, 2003-09-02 at 07:48, Alexander Lyamin wrote:
> > Mon, Sep 01, 2003 at 09:06:37PM +0100, Roger James wrote:
> > > Hi,
> > > 
> > >  
> > > 
> > > Ever since I updated my SuSE Linux 8.1 set up with the latest patches
> > > from SuSE I have been seeing the following messages at boot up time
> > > (once only for each drive).
> > > 
> 
> Which kernel are you running now?
> 
> > > <4>hda: drive_cmd: status=0x51 { DriveReady SeekComplete Error }
> > > 
> > > <4>hda: drive_cmd: error=0x04 { DriveStatusError }
> > > 
> > > <4>hda:saw err in ide_end_drive_cmd, disabling write barriers
> > > 
> > > <4>hda: drive_cmd: status=0x51 { DriveReady SeekComplete Error }
> > > 
> > > <4>hda: drive_cmd: error=0x04 { DriveStatusError }
> > > 
> > > <4>hda:saw err in ide_end_drive_cmd, disabling write barriers
> > > 
> > > <4>reiserfs: using ordered data mode
> > > 
> > > <4>Using r5 hash to sort names
> > > 
> > 
> > To me and my expirience, it seeems like your hard drive about to fail.
> > And it seems like its about to fail in area where jdata stored.
> > 
> 
> The barrier patches try a write cache flush command on the drive, and if
> it fails they turn barrier mode off.  These messages mean your drives
> don't support the barrier mode, you can ignore them.  Some drives print
> the failures if you've already got write cache disabled.  Try hdparm -W
> 1 /dev/xxx before mounting an FS on the drive.

<4>hda: drive_cmd: status=0x51 { DriveReady SeekComplete Error }
<4>hda: drive_cmd: error=0x04 { DriveStatusError }

also appear when ide code access bad blocks on drive...
it would be nice if you could suppress thouse lines when probing for 
cache flush functionality on disks, otherwise some people out there
would be upset, and thats why its sprung my WRONG comment. sorry ;)


-- 
"the liberation loophole will make it clear.."
lex lyamin