Question
You run a carrier-grade SIP routing platform; SIP proxies discover the media-relay (RTP) servers via DNS SRV records that are updated when relays are added/removed. At 02:10 a media-relay cluster in one POP is decommissioned and its records are removed from DNS. Call-setup success rate immediately drops 8% and stays there: a fraction of new calls fail with no media path (one-way or dead audio), and the failures map to proxies still trying to reach the DECOMMISSIONED relay IPs. The authoritative DNS shows the SRV records were correctly removed at 02:10. The proxies' upstream resolver and the proxies themselves appear to be returning the old relay addresses. SRV records were published with a 3600s TTL. Triage this and give the immediate fix plus the durable service-discovery posture.
Stop the bleeding first (mitigate), then form hypotheses from real signals. Separate root cause from symptom, communicate status as you go, and close with what prevents a repeat.