TCP/IP Routing Questions

We have two Ethernet cards with these addresses in our production machine:

Internet Subnet Line Line
Address Mask Description Type

127.0.0.1 255.0.0.0 *LOOPBACK *NONE
567.17.0.40 255.255.0.0 ETHLINE *ELAN
567.17.0.42 255.255.0.0 ETHLINE2 *ELAN
567.17.255.42 255.255.0.0 ETHLINE2 *ELAN

The routes look like this:

Route Subnet Next Preferred
Destination Mask Hop Interface

*DFTROUTE *NONE 567.17.0.165 567.17.0.40
567.15.0.41 *HOST 567.17.0.20 567.17.255.42
567.15.0.41 *HOST 567.17.0.25 567.17.0.40
567.15.255.43 *HOST 567.17.0.20 567.17.0.40

I'm not sure why this setup or how it got this way. The 567.15 addresses are our offsite backup machine and the 255.42 and 255.43 are the addresses that Mimix talks on.

We tested a planned Mimix switch. I had a client access session on 567.15.255.43. Part of the switch disables 567.17.0.40 (the preferred interface for 3 of the routes) and when that happened my client access session disconnected and Mimix processing stalled. I suspect disabling 567.17.0.40 caused both the *DFTROUTE and the 567.15.255.43 destination to be inoperable. However, somewhere between 30 minutes and an hour later my client access session reconnected and Mimix took off again.

I have the feeling that the *DFTROUTE should have *NONE as the preferred interface and the other three entries are unneeded, because the 567.17.0.165 default gateway on the network knows all these destinations.

Any thoughts?

Second question; Why did everything start working again 30+ minutes later? I have a feeling there is some retry threshold that was exceeded and is probably configured way too high.

Thirdly, in what order are the routing entries consulted? Is the *DFTROUTE the last one checked if there is no hit on any other entry?

Thanks, Sam

(Sorry for the long post and if these seem like simplistic questions I apologize--I'd rather be programming...)