> | it was decided that 10% was the best minfree figure ?
> At work we only have 2.35, 3.55 and 5.1 gig drives; we always change minfree
> to 5% or even 1%. I have not seen any performance or other problems.

I think I saw an article from Chris Torek (?) sometime back on USENET
mentioning that they had found that the current UFS code optimizes fine
with 5% free.  Of course, with modern systems and modern disks, it's
kind of an interesting question how much work one needs to go to to
get decent performance -- as anyone using Net/Free/386/BSD with a SCSI
disk and telling newfs that it has 64 heads per track and 1Mb per
cylinder can attest.

> It now takes and returns off_t offsets which are 64 bits. A *lot* of stuff
> broke when that changed.  This is a very good change though. Right now
> we cannot copy disks at work with a single dd of /dev/sdXc, or create
> files greater than 2 gig.

Heh, I never have any trouble with that at work.  (I wonder how long it
would take me to port NetBSD to the KSR/1...?  :-)

------------------------------------------------------------------------------