Panic threshold prevents health check enforcement in load balancing

When cluster endpoints fail active health checks, Envoy continues routing traffic to them due to default panic threshold behavior. Endpoints show 'failed_active_health_check: true' but remain in load balancing rotation with 'eds_health_status: HEALTHY', causing connection failures to unreachable backends.

Set healthy_panic_threshold to 0 in the cluster's common_lb_config to disable panic mode and enforce strict health check-based endpoint exclusion. Verify with cluster dump that failing endpoints are properly marked unhealthy and excluded from routing. Reference: https://www.envoyproxy.io/docs/envoy/latest/faq/load_balancing/lb_panic_threshold