Kamailio Transactions: Why Stateful Forwarding Changed My SIP Debugging
Once I separated stateless forwarding from transaction-aware forwarding, retransmissions, replies and failure handling in Kamailio became much easier to reason about.
Once I separated stateless forwarding from transaction-aware forwarding, retransmissions, replies and failure handling in Kamailio became much easier to reason about.
Database traffic volume looked normal even when applications were rolling back more transactions than usual.
SIP became easier to debug once I stopped treating an entire call as one exchange and separated individual transactions from the dialog that ties them together.
A PostgreSQL deadlock can resolve automatically by aborting one transaction, leaving the service apparently healthy after the incident.