Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.
This is the main idea do not execute on_reply route in case 503 and available failover endpoint. The same behavior as with timeout.
There is no possibility to know did we try all candidates. In our case when we use the rtpengine we do delete but after that we can have successful reply from other failover candidate.
As I understand the behavior will change only in case when reply status will be changed in the reply route (to or from 503 in default). So only in case when we want to try or visa-versa don't want to try other candidates, but again we don't know anything about candidates.
This changes were done more than year ago and already works on few thousands installations without any problems (but only with new behavior).
Please let me know if any additional checks can be done from my side.
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.