Hi,
In this serial forking-based failover scenario
UAC --> Proxy --> UAS 1 UAS 2 ... UAS 3
I've got a requirement to terminate the call and return an error to the UAC if there is no final response within 60 seconds.
The usual fr_inv_timer / restart_fr_on_each_reply combination won't do, because even when restart_fr_on_each_reply == 0, the timer is still restarted on provisional replies of increasing response code value (e.g. 183 after 180). As the documentation states, fr_inv_timer is a branch-level construct anyway, so this approach doesn't make sense here.
That's why I looked into max_inv_lifetime. However, when I set it to 60000, I still get the same result, but only when there are new branches being formed on the proxy -> UAS side. It works fine if there is no forking and a single gateway does not provide a final reply within 60 seconds. So, my sense is that the max_inv_lifetime timeout is reset when a new branch is created to UAS 2...UAS N.
Is this accurate?
Thanks,