Fix:
server device1 device1.local:8080 check resolvers mydns A stale device record in a redirector is a state inconsistency problem caused by missing deregistration, long TTLs, or broken health checks. It degrades availability and misdirects traffic. The fix involves immediate manual cleanup, then implementing TTL reduction, health checks, and a device heartbeat/deregistration protocol. Long-term, shift to service discovery with real-time consistency. a stale record of the device was found in the redirector.
backend devices server device1 10.2.3.4:8080 check inter 5s rise 2 fall 3 Stale if device1 moves to 10.2.3.5 . Fix: server device1 device1
# Remove stale from runtime echo "disable server devices/device1" | socat stdio /var/run/haproxy.sock echo "set server devices/device1 addr 10.2.3.5 port 8080" | socat stdio /var/run/haproxy.sock echo "enable server devices/device1" | socat stdio /var/run/haproxy.sock Better: Use DNS resolution: then implementing TTL reduction