Describe the bug
Our edge routers emit an error every ~15 seconds:
{"error":null,"file":"github.com/openziti/ziti/router/xgress_common/connection.go:222","func":"github.com/openziti/ziti/router/xgress_common.(*XgressConn).ReadPayload","level":"error","msg":"failed to set write deadline, connection is fully closed"}
This correlates with short-lived connections created by ProxySQL health checks. When we stop ProxySQL monitoring, the router errors stop immediately. Business traffic is unaffected, but the router logs are noisy and make operational alerting harder.
Environment
-
Controller: v1.7.1 (Docker)
-
Routers:
-
1× Docker router (
openziti/ziti-router:1.7.1) -
1× Linux binary router (
ziti-router), version:1.7.1
-
What happened
-
ProxySQL monitoring (default or similar intervals) opens short connections to database backends that traverse Ziti.
-
On the router, we see the error above every ~15s (aligned with cleanup/health intervals).
-
Stopping ProxySQL monitoring removes the error entirely. Restarting it brings the error back.