If you have a /24 subnet (netmask 255.255.255.0), then
you have a 24-bit net$
Sigh.
Compensating manually,
[...], though a proper, old-school class-C subnet
can't have a
network number smaller than 192.0.0 or larger than 223.255.255.
True but somewhat incomplete. In CIDR terms, class A is 0.0.0.0/1,
class B is 128.0.0.0/2, and class C is 192.0.0.0/3. The netmask width
(for contiguous subnet masks) is orthogonal to this - the current
pernicious abuse of "class C" to mean "/24" is a mistake, and a very
annoying one....
There are also class D, 224.0.0.0/4, and class E, 240.0.0.0/4, but
practically nobody remembers them any longer, probably at least in part
because they saw fairly little use (they were not for normal unicast
host addresses - you can find the whole list in RFC1812 2.2.5.1).
Come to think of it, if you have class D or E space, it's possible
pre-CIDR software will get upset at your trying to use it at all, so if
you're trying to use addresses somewhere in the
224.0.0.0-255.255.255.255 range, that could be the problem and it might
be worth trying something in an RFC1918 block just to see if it'll be
willing to configure that way.
If [...], you have to select the class your network
number falls
into, then manually override the number of bits for the host number
to match your subnet mask. [...]
With any luck you're not on a network that breaks
down into the old
class-based system with a part that doesn't fit into an 8-bit chunk,
Or one with a noncontiguous subnet mask; based on the description of
picking what today we'd call the CIDR width, there's may not even be
any way to configure such a thing. Fortunately, they were never
anything but "very rare"; I know of only one person ever using such a
netmask in production.
/~\ The ASCII Mouse
\ / Ribbon Campaign
X Against HTML mouse at
rodents-montreal.org
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B