On Wed, Mar 01, 2006 at 10:49:10AM +1100, Malcolm Herbert wrote:
> On Wed, Mar 01, 2006 at 12:44:25AM +0100, Zbigniew Baniewski wrote:
> |I'm not really sure: is that well-known from Linux-world, "laborious"
> |killing all the daemons and shutting down all the services, not needed, in
> |fact, at all?
> 
> presumably the machine can only do that if there aren't processes holding
> the disks open at the time ... if so, then the unmount will stall
> 
> afaik that would hold for both NetBSD and Linux ...

Certainly 'uadmin 2 0' will shutdown a solaris (or other SVR4) system
within a fraction of a second, it sends 'kill -TERM' to all processes,
send 'kill -9' to those that haven't died, unmounts all the filesystems
that are idle, then halts.
Unfortunately this doesn't work if you have any filesystems that require
process level code running to unmount (and ISTR that includes NFS).

I suspect that the NetBSD 'halt' has much the same effect.


	David

-- 
David Laight: david@l8s.co.uk