On Fri, 30 Aug 2002, Richard Earnshaw wrote:

# ... a lot of pointless trivial examples

Those "pointless trivial examples" are exemplified if just ONE block goes
bad under /rescue (the new situation) while it's much less likely that
you're going to lose blocks under Every Single Binary - or even a good
portion of them - in /bin or /sbin

# /sbin/init: bad sector
#
# Hmm, don't seem to have a replacement for that one....

[we should have been supporting an alternate init LONG before now]

# fsck, ifconfig, mknod, badsect, ...
#
# I could go on.
#
# If you've got a hosed system then there are MANY critical tools which are
# likely unusable.

Yes, but it's MUCH LESS FRAGILE with distributed files than it is with
A Single Binary!

# > That is also an option. But /rescue is rather vulnerable, plus I probably
# > don't even know if it works until the day I'm going to use it.
#
# So run some tests on it.  You are capable of doing that I suppose, you do
# know the root password after all.

I've heard it stated that we're trying to avoid having to use a CD or
floppy-based rescue.

I don't think that's, in general, a sane idea.  ALWAYS have backup media,
even with the current scheme.  That's just common sense, no matter which
direction we're going, here!  It applies to, e.g. Luke as much as it
does to me.

The object of this move did NOT include letting systems administrators
get lazy.  If you don't have a recovery medium, whose fault is that?

				--*greywolf;
--
NetBSD is PAR -- Powerful, Advanced, Reliable.  Is your OS up to PAR?