Eject Unhealthy Upstream Hosts in Envoy with Outlier Detection
Configure Envoy outlier detection to automatically eject unhealthy upstream hosts based on 5xx errors and latency, verify ejection via admin stats, and tune recovery options.
ReadMeFeed / Community knowledge
Real questions. Useful conversations. Find the people who know your stack.
Configure Envoy outlier detection to automatically eject unhealthy upstream hosts based on 5xx errors and latency, verify ejection via admin stats, and tune recovery options.
Learn how Envoy’s outlier detection automatically ejects unhealthy upstream hosts, how to configure it, what trade‑offs to watch, and how to verify the behavior in practice.
Learn how Envoy’s passive outlier detection and connection‑pool circuit breakers work together to isolate failing upstream hosts and prevent overload, with a concrete configuration example and verification steps.
Learn how to decide between consecutive‑error and success‑rate outlier detection in Envoy, configure it, and validate the effect with admin endpoints.
Learn how Envoy’s outlier detection automatically removes flaky upstream hosts during canary rollouts, with a practical configuration example, verification steps, and trade‑off guidance.
When Envoy starts ejecting hosts, 503s surge. This guide walks you through diagnosing the cause, tuning outlier detection, verifying fixes, and knowing when to involve SREs.
A concise diagnostic guide for Envoy outlier detection: spot ejection symptoms, check counters and config, adjust thresholds, and know when to escalate.