Re: default hdparms? readahead=256? Re: WD 400GB xATA Drives

[Justin Piszcz]

Turned out I had too many devices in the box, I have removed them now, 
still getting those pesky drive errors though, I will try the hdparm soon, 
thanks.


On Mon, 13 Feb 2006, Mark Lord wrote:

> Linda Walsh wrote:
> > ..
> > set to known values may not contain "great values".  My "readahead"
> > value seems to boot with a value of "256" for all of my drives.
> >
> > Is this not a bit "excessive"?
>
> Wow.. that does look a bit HUGE.  Thanks for pointing it out,
> I'm resetting mine back to 128 for now.
>
> > Justin Piszcz wrote:
> > > When I write to this disk for a while, I see this in dmesg (only once so 
> > > far):
> > >
> > > [31230.223504] ata6: translated ATA stat/err 0x51/04 to SCSI SK/ASC/ASCQ 
> > > 0xb/00/00
> > > [31230.223511] ata6: status=0x51 { DriveReady SeekComplete Error }
> > > [31230.223515] ata6: error=0x04 { DriveStatusError }
> > >
> > > Is there some sort of smart testing going on constantly?  I only get 
> > > 26-27MB/s on this 400GB/SATA/16MB/7200RPM drive.  I use smartmontools to 
> > > do a daily test.  However, even with smart disabled, I get:
> > >
> > > # hdparm -t /dev/sde
> > > /dev/sde:
> > >  Timing buffered disk reads:   78 MB in  3.06 seconds =  25.46 MB/sec
> > >
> > > Does anyone know if this drive has problems in Linux or something?
> > >
> > > [    6.895914]   Vendor: ATA       Model: WDC WD4000KD-00N  Rev: 01.0
> ...
>
> Some drives suck at sequential reads, in favour of doing random seeks
> very very well.  Basically, the on-drive seek algorithm may not favour
> doing much read-ahead.  Try "hdparm -A1" and see if that helps.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/