Hi Phaneendra,
I think there may actually be two different scenarios hidden behind the same symptom, and distinguishing between them would probably determine the troubleshooting path.
The first thing I would try to confirm is:
During those 1–4 minutes of silence, has the customer actually disconnected from their phone/network already, or are both endpoints still physically connected and simply not speaking?
I ask because conversational silence and idle RTP are not the same thing.
The Genesys Disconnect on Idle RTP setting disconnects the call when RTP is no longer being received for an extended period. For a normal bidirectional call, Genesys documents that period as 5 minutes.
So, if both parties remain connected and RTP packets are still flowing - even if those packets contain silence - I would not expect Disconnect on Idle RTP to trigger at all.
That would lead me to investigate the situation in two different ways:
Scenario 1 – Both customer and agent are still genuinely connected
If neither side actually disconnected and they simply stopped talking, the SIP dialog and RTP session may be perfectly healthy.
In that case, this seems less like a trunk/carrier issue and more like a requirement for real-time conversational silence detection after the call has already been connected to an ACD agent.
Architect's Detect Silence action can help while the call is still being controlled by a flow, but after the ACD transfer and agent connection, I am not aware of a native Architect mechanism that continues monitoring the live conversation and disconnects it after X seconds of conversational silence.
Scenario 2 – The customer actually hung up earlier, but Genesys kept the call connected
This is where I would start looking at the telephony signaling.
For one of the affected conversations, I would try to correlate the Genesys conversation timeline with the carrier/SBC SIP trace and determine:
-
At what exact timestamp did the customer actually disconnect?
-
Was a SIP BYE generated by the carrier?
-
When did that BYE reach Genesys Cloud?
-
Did Genesys respond with the expected 200 OK?
-
Did RTP continue to arrive after the customer supposedly disconnected?
-
Was the SIP dialog being kept alive somewhere between the carrier/SBC and Genesys?
For example:
Customer hangs up → Carrier/SBC sends BYE → Genesys receives BYE → call disconnects
If the customer hangs up but no BYE reaches Genesys for another three minutes, I would investigate the carrier/SBC side.
If the BYE reaches Genesys immediately but the Genesys conversation remains connected, then I would open a case with Genesys Support using that SIP timeline as evidence.
There is also a third useful case to validate:
RTP actually stops, but no SIP BYE is received.
In that situation, Disconnect on Idle RTP should eventually provide protection, but according to the documented behavior, that would normally be after 5 minutes, which may explain why it does not address the 1–4 minute examples you are seeing.
Your participant disconnect reasons:
External = Endpoint
Internal = Peer
also seem consistent with the external participant ultimately being the side that terminates the interaction. But by themselves, those values probably do not tell us whether the customer attempted to disconnect several minutes earlier and the signaling was delayed somewhere upstream.
So I think the most important question is:
Can you confirm whether the customer is actually still connected during those silent minutes, or whether they have already hung up and the call remains established in Genesys?
If you have one affected conversation where you can compare the Genesys timestamps with the carrier/SBC SIP trace, that would probably tell us very quickly which scenario you are dealing with.
------------------------------
Fernando Sotto dos Santos
Consultor de Atendimento Senior Grupo Casas Bahia
------------------------------