NetBSD dump seems to default to 10k (which maps very nicely to
	the 20 blocks) per write.

	What happens if you use -b to switch the blocksize to say 63
	(which might be the SunOS default) or even higher?

	Manuel Bouyer recently commited changes to dump to use a read
	cache, which speeded things up quite a bit for normal use. It
	might help your situation. Presumably the next write is not ready
	by the time the drive has finished with the previous :/

		David/absolute

	Abandoning a syncing disk... or syncing an abandoned disk?

On Thu, 25 Mar 1999 rdkeys@unity.ncsu.edu wrote:

> I finally got a sun3 machine set up and running (a bit spartan though
> with only 4M ram and a 240mb scsi drive).  But, when doing the initial
> system dumps of root and usr, the dump seemed to be going inordinately
> slow, like about one pass of 20 blocks every 10 seconds or so.  Should
> it be that slow?  Is there a fix?  I had both NBSD 1.3 and 1.3.2 up
> on the machine, and both acted the same way.  On sunos, dumps stream
> at machine hardware speed, or essentially continuously.  Any insights
> are appreciated.
> 
> Thanks
> 
> Bob Keys
> 
>