Subject: Understanding SIP 408 Request Timeout Behavior vs. PCAP Capture (180 Ringing)

Telephony
1

Hi LiveKit Team,

I’m investigating a SIP trunking call failure where a call was logged by LiveKit Cloud as a 408 Request Timeout, but the packet capture (PCAP) for the session only contains 4 frames and stops abruptly while the call was still ringing.

Call id: SCL_pG3AYKUgoYTm

Here is the exact sequence captured in the PCAP:

  • Frame 1 (0.00s): INVITE sent to our SIP trunk provider (TTSL) for destination 9720393150

  • Frame 2 (0.74s): 100 Giving it a try (from the carrier)

  • Frame 3 (5.20s): 180 Ringing (from the carrier)

  • Frame 4 (7.62s): 180 Ringing (repeated/refreshed)

My Questions:

Since the PCAP cuts off during the 180 Ringing phase, does LiveKit Cloud internally time out and map unanswered ringing sessions to a 408 Request Timeout after a specific duration threshold?

Is there a configurable ring timeout or transaction timer on the LiveKit SIP integration side that we can adjust if a destination takes longer to pick up?

2

This looks like it was a timeout from the carrier side. Do you have a PCAP from that side?

LiveKit did not time out this call and did not generate the 408. The carrier returned a real SIP/2.0 408 Request Timeout as the final response to our INVITE, roughly 34.8 seconds after it was sent, following ~7.6 s of ringing and then 27 s of silence.