🔒 Paid
Geo-DNS Failover Flapping
Global clients are seeing a growing rate of connection resets and errors. Neither US-East nor US-West ever reports as unhealthy — each region's own p99 latency only blips briefly before recovering on its own. But GeoDNS failover activity keeps climbing: the controller is flipping traffic back and forth between regions on every transient blip, and every flip tears down in-flight client connections. Diagnose why the failover itself has become the outage, and restore stable routing without simply disabling automated failover.
- Difficulty
- Senior
- Expected Time
- ~15 min
- Category
- API & Traffic
Skills
- Distinguishing DNS-level routing instability from app-layer load imbalance
- Recognizing when the failover mechanism itself is the outage, not region capacity
- Evaluating a brittle pin-one-region shortcut versus a genuine threshold-tuning fix
- Reading flap/oscillation metrics distinct from steady-state health signals
System Architecture
- Global Clients
- GeoDNS Failover Controller
- Health Checker
- Region: US-East
- Region: US-West
This is a Pro scenario. Upgrade to Pro to unlock it.
Upgrade to ProBack to Incidents