Interesting Behavior with Solaris 2.8

[John Tabasz (jtabasz)]

Hi,

I have a Sun V100 running Solaris 2.8. I want dmfe0 to be under the
control of the dhcpagent on Solaris and dmfe1 to be statically
controlled. 
Previous to running a DHCP server in this environment I had no problems
configuring the Sun the way I want them to be. 

My secondary interface, dmfe1 I want to be configured as 10.1.1.1 with
the netmask of 255.255.0.0. This is a legitimate combination of values.
I do the following in order to do this.
#cat /etc/netmasks
192.168.25.0	255.255.255.0
10.1.0.0	255.255.0.0

#cat /etc/hostname.dmfe1
10.1.1.1

Before I introduced dhcp to the environment, on booting, the server
would come up with the following interface configuration for dmfe1:

dmfe1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index
3
        inet 10.1.1.1 netmask ffff0000 broadcast 10.1.255.255

This looks completely legitimate to me. 10.1.1.1 is a legitimate address
in the subnet 10.1.0.0. 

However, I introduced dhcpd into the environment and use it to deliver
address and dns and hostname information for the other interface only. 
On the client I run 
#echo primary > /etc/dhcp.dmfe0   to allow dhcpagent to control the
interface. 
When the V100 comes up, even though the dhcpagent should have nothing to
do with the dmfe1 interface, dmfe1 gets configured such with the same
values in the /etc/netmasks and /etc/hostname.dmfe1 files:

dmfe1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index
3
        inet 10.1.1.1 netmask ffffff00 broadcast 10.1.1.255

So for some reason the control of dmfe1 is bleeding over from dmfe0, and
it's behaving in a way that looks like it violates subnetmasking rules.

Can anyone comment on this? Do I need to upgrade my dhcpagent rev or
something?

Thanks in Advance,

John