The answer to the configuration in the original question is, as the previous answers indicated, Auto Port Aggregation software which has to be licensed for a nominal (read as not-so-cheap) software.
Coming to the configutrations to the 3 questions in your later post
1) possible but having two default routes is not advisable in my opinion. Even though you will have two routes, there is no guarantee (that I know of) that they will be used in a balanced fashion.
2) is not very different that my answer to the first question
3) if you want a seamless (i.e. no down time) solution, it goes through APA, which costs money. If you can tolerate a down time between the failure and detecting it, you can very easily switch interfaces with two simple commands in a script and call this script from the function which tests and finds a network failure:
lets assume you have lan1 and lan2 to switch between
you will have a file, lets say, /var/adm/activelan, which will have the active instance number at any given time, either 1 or 2
failed=[ some test condition here that you will develop ]
# failed=0 means no failure
# failed=1 means network failed
if [ ${failed} -eq 1 ]
then
activelan=`cat /var/adm/activelan`
ifconfig ${activelan} down
let activelan=${activelan}+1
if [ $activelan -gt 2 ]
then
let activelan=${activelan}-2
fi
ifconfig ${activelan} inet a.b.c.d netmask 0xffffff00 up
echo ${activelan} > /var/adm/activelan
fi
it is upto you to determine that there is a network failure how ever you wish to handle it.
then you need to run the script at the intervals that you can tolerate a downtime of.
APA saves you a lot of headaches even though it is a bit on the expensive side.
________________________________
UNIX because I majored in cryptology...
