Over a window of several hours the primary address ran longer anti-DDoS queue times than usual under what looked like a coordinated flood. The board recorded queue waits stretching past a minute on some circuits where they normally clear in seconds.
At no point was the market itself down. The secondary and tertiary addresses stayed responsive throughout, which is precisely the reason the pool holds three addresses. Readers who switched from the primary to a fallback continued without interruption. Readers who waited on the primary saw longer queues but still got in.
The board's guidance during a flood is unchanged. Do not sit on a slow address hoping it recovers. Move to the next address in the pool, verify it, and continue. A flood raises queue times, it does not compromise the addresses, and the verification rule protects you either way.
The window closed on its own as the flood subsided. No rotation was needed and no address was retired.