> Maybe rspfd is deleting the static route when it shouldn't?
TD> You added the static route when you shouldn't have (after
TD> rspfd had started).
I don't think it's reasonable to prohibit manual modification of the routing
table while rspfd is running. As far as I can think of, rspfd should be
stateless except when actually testing a route. Perhaps we need to get a lock
added to the IP routing table so that manual changes are prohibited during
those brief intervals when a route is being tested.
I have not really thought this through in detail and I would interested to hear
Craig's comments, but I can't see any reason why rspfd should be getting its
routing information from anywhere other than the real table (or /proc/net,
which amounts to the same thing).