Anders Dinsen writes:
> On Mon, 24 Jul 2000, Greg Oster wrote:
> 
> > Anders Dinsen writes:
> > > On Mon, 24 Jul 2000, Anders Dinsen wrote:
> > > :-) it's actually faster...
> > 
> > What does your raid config file look like?  It could be that there is some 
> > room for tweaking the performance a bit :)
> 
> Here it is:
> 
> START array
> 1 3 0
> 
> START disks
> /dev/sd1e
> /dev/sd2e
> /dev/sd3e
> 
> START layout
> 32 1 1 0
> 
> START queue
> fifo 100
> 
> I suppose the only parameter I can experiment with is the stripe unit
> of 32?

Yup... If you don't have any real data on that set yet, bump that to 
64 or maybe even 128.  You'll likely need to redo your disklabel and
re-newfs things though...

> Here's what dmesg has to say about my disks; 
> 
> sd1 at scsibus0 targ 1 lun 0: <DEC, RZ25     (C) DEC, 0700> SCSI2 0/direct fi
> xed
> sd1: 406MB, 1476 cyl, 9 head, 62 sec, 512 bytes/sect x 832527 sectors
> 
> If I used that information, a cylinder is 34 blocks, so I'm pretty close
> at that. No, there's probably not much to do than replace the old disks
> :-)

Try changing the 32 to 64... it should get faster..

> While I'm pasting from dmesg: why does it say:
> 
> /dev/sd1e is not clean!
> 
> for each of the components during boot?

If this wasn't a RAID 0, it'd be cause for concern.. what it's saying is 
that the component didn't get shutdown correctly, and didn't get marked 
'clean'.  I.e. it thinks that the parity may be incorrect... 
You can run 'raidctl -i' to fix that... (although since RAID 0 has no parity, 
it's not a big deal...  (and I should probably just make the kernel shutup
about the parity if it's a RAID 0 set.))

Later...

Greg Oster