This issue was generated by our automated monitoring service. The cause was purely in the monitoring service, not in the TURN service, so it had no impact on service availability. We confirmed this through manual tests of the TURN service during the incident. Since the issue was isolated to the monitoring layer with no customer impact, and it occurred during the summer vacation period, we scheduled the fix for after the vacation rather than handling it as an emergency change. This is the reason for the long incident duration and for the multiple related incidents during this period.
