If the default route for the network is on, for example, eth1, after
booting up and bringing up network interfaces, ifconfig, route -n, etc
seem to show that the network is healthy but indeed it is not - ping, etc,
show that the network is unusable.
However, if the administrator manually enables the interface responsibile
for communication with the default route before other interfaces are
brought online, everything works.
This was my mistake. Somehow, I'm still trying to figure out how, out of several
interfaces on eth0 one of them was in the same subnet as the default gateway,
which should have been out eth1. I'm scratching my head over this one because i
copied the ifcfg-eth* files from a working rh6.0 system that i was working on
rebuilding from the ground up. maybe someone else was screwing around with it
and thought they were on another system, you know how it is when you have eight
similar systems on the same KVM and multiple admins.
Might be nice if there were a sanity check in the initscripts that would check
for this kinda thing, but it's really my fault for not fully proofreading my
config files. Sorry.