bind9.7.1 Instance seems to not talk to systems on its own network.

[Martin McCormick]

	This sounds like a firewall issue, but read carefully.

	Our master and slave are on 2 different networks that
are each larger than a single octet. Both are producing messages
like:

client 139.78.100.57#33486: error sending response: host unreachable

	That particular VLAN or subnet is a /22 and runs from
139.78.100.0 through 139.78.103.255.

	For the moment, we have the firewall on the bind system
off as there is a firewall for all the hosts on this network but
there is no firewall between the name server and the complaining
systems.

	Our slave is on another subnet that is 2 octets wide and
the complaints on that host are all about other hosts in the
same octet where the DNS lives.

	I should know what this is, but it doesn't quite make
sense. Both DNS's appear to actually be working right except for
these messages.

	Does this sound familiar to anyone?

	We were running bind9.6.3 before upgrading and never saw
similar messages there. The first firewall rule on both systems
is:


#open the firewall for testing.
	${fwcmd} add pass all from any to any

	This should be like not having any firewall at all.

Martin McCormick