I can’t seem to figure out why our sip registrations are constantly flapping on three of our fifteen byoc trunks in the apac region. this started happening right after the carrier pushed a firmware update on their side last tuesday. we are seeing a pattern where the registration succeeds, holds for about four minutes, then drops with a 408 request timeout error, only to immediately re-register and repeat the cycle.
the environment is genesys cloud platform version 2023-12, using the standard byoc configuration with tls 1.2 and sdes srtcp for encryption. the sip credentials have not changed, and the ip addresses are whitelisted correctly on both ends. i have checked the call detail records and the sip trace logs provided by the carrier. the logs show that the initial registration request is accepted with a 200 ok response, but the subsequent keep-alive packets seem to be dropped or delayed beyond the expiration timer set in our trunk configuration.
we have tried adjusting the registration expiration time from the default 3600 seconds down to 1800 seconds, but the flapping persists, just at a faster rate. we also attempted to disable srtcp temporarily to rule out any encryption negotiation issues, but the problem remains. the carrier support team is claiming that the issue is on our side because they are seeing the registration requests come in, but they are not seeing any abnormal behavior in their sip proxy logs.
is there a known issue with how genesys cloud handles sip registration keep-alives when the carrier’s sip proxy introduces latency? we are seeing this specifically on the trunks connected to the singapore and sydney regions, while the trunks in tokyo and bangkok are unaffected. this is causing intermittent call setup failures, especially for outbound calls where the system tries to use a trunk that is currently in a failed registration state. any insights or workarounds would be greatly appreciated. we need to stabilize this before our peak call volume period starts next week.