Empty istio-proxy logs "- - -" 0 UH

I’m seeing a whole lot of logs like this one:

[2019-04-09T20:25:58.189Z] "- - -" 0 UH "-" 0 0 0 - "-" "-" "-" "-" "-" - - 172.x.y.z:6380 10.a.b.c:33023 -

These easily outnumber real logs by 30 to 1. Not sure why this is empty. I’ve looked at Envoy access logging documentation and notice there is the UH flag (No healthy upstream). The liveness checks are working, but I don’t see any logs for the readiness checks. The pod is at a 2/2 and shows ready in get pods.

Are these logs inbound or outbound? Why no status code? Why no method, path, protocol?

How do I figure out what is causing this?

Looks like this log exists in all the pods in the namespace, not just this one. And liveness/readiness probes are returning 200.

The redis box at 172.x.y.z:6380 is definitely up and running and available through the network. I’ve manually connected to it from a pod on the cluster (without a proxy) and received a successful result from the redis-cli command info keyspace. However, running the same commands from a pod with a sidecar I get the error Error: Connection reset by peer. I’ve verified that a ServiceEntry exists for that IP address and port.

I have ran into the exact same issue trying to connect to Postgres. Can someone give us some help onto how to debug this please

1 Like