Downgrade ECONNRESET Error: socket hang up in logs to debug

[inqueue]

Kibana version: 6.3.1
Elasticsearch version: 6.3.1

Kibana logs this error when connections are made that do not establish an actual TLS session; e.g., connection checks either with netcat/telnet or load balancers:

{"type":"error","@timestamp":"2018-08-15T22:09:52Z","tags":
["connection","client","error"],"pid":14699,"level":"error","error":{"message":"socket hang 
up","name":"Error","stack":"Error: socket hang up\n    at TLSSocket.<anonymous> 
(_tls_wrap.js:876:25)\n    at emitOne (events.js:121:20)\n    at TLSSocket.emit (events.js:211:7)\n    at 
_handle.close (net.js:557:12)\n    at Socket.done (_tls_wrap.js:356:7)\n    at Object.onceWrapper 
(events.js:315:30)\n    at emitOne (events.js:121:20)\n    at Socket.emit (events.js:211:7)\n    at 
TCP._handle.close [as _onclose] (net.js:557:12)","code":"ECONNRESET"},"message":"socket hang up"}

Reproduce on a TLS enabled Kibana with nc kibana.host 5601, then CTRL+C. If a load balancer causes it, the log message populates the log at an uncomfortable rate.

Could this logging be suppressed to debug level only?

[bhansenmis]

I am also having this issue. Happening on kibana nodes behind an AWS loadbalancer. It would be great if these errors could be suppressed

[tylersmalley]

We're looking into this.

@bhansenmis you can terminate SSL at the LB to resolve this. With the upside of having centralized SSL management.

[bhansenmis]

@tylersmalley Thanks for the reply. I am currently terminating SSL at the load balancer with a company cert, and then we have self-signed certs between the load balancer and the kibana nodes. We didn't want to have unencrypted traffic between the lb and the kibana nodes. So with that setup, this issue still occurs.