Question 1

What is the safest approach for SIP to PJSIP migration?

Accepted Answer

The "Dual-Stack" method is the industry standard for high availability. By running chan_sip (Legacy) and res_pjsip (Modern) on different ports (e.g., 5060 and 5160) simultaneously, you can migrate tenants incrementally. This provides an instant rollback path. If an issue occurs, you simply revert the endpoint configuration to the old port.

Question 2

What are the most common migration mistakes to avoid?

Accepted Answer

The most critical mistakes include Port 5060 binding conflicts (which crash the PBX), failing to explicitly define identify_by parameters (causing routing failures from provider trunks), and incomplete NAT configurations in the transport layer (resulting in one-way audio).

Question 3

Why should service providers use PJSIP Realtime instead of config files?

Accepted Answer

For large-scale deployments, loading static text files causes significant delay during core reloads, potentially dropping active calls. PJSIP Realtime (Sorcery) allows Asterisk to query a database dynamically for endpoint data, enabling instant provisioning without restarting the telephony engine.

Question 4

What real-time monitoring should be active during migration?

Accepted Answer

Beyond standard uptime, monitor the Task Processor Queue Depth to ensure the PJSIP thread pool is keeping up with SIP traffic. Additionally, watch for 401 Unauthorized auth mismatches and 488 Not Acceptable codec mismatches using tools like sngrep or Homer.

Question 5

How do I fix one-way audio after migrating to PJSIP?

Accepted Answer

In PJSIP, NAT handling is defined at the Transport layer, not just on the endpoint. Ensure your transport section correctly defines local_net (your internal subnet) and external_media_address (your public IP). Without this, PJSIP may advertise its private IP in the SDP payload, causing audio failures.

SIP to PJSIP Migration

SIP to PJSIP Migration Zero Downtime Strategy