Zabbix Proxy getpeername() failed: [107] Transport endpoint is not connected

[Megafisshy]

Hi all!

In our architecture we're running a Zabbix proxy in our production environment that phones back to our mothership central Zabbix server.

We're having some trouble periodically with the proxy server which leads to us missing some metrics coming in and the error message below showing up in the proxy logs:

failed to accept an incoming connection: connection rejected, getpeername() failed: [107] Transport endpoint is not connected

We do run many hosts through this proxy, some active agents, but mostly passive ICMP pings.

A Netstat query while experiencing this issue reveals we have around 439 open http connections waiting for responses/timeouts.

Usually restarting the Zabbix proxy service is enough of a kick in the pants to resolve this, but in the most recent episode we had to restart the entire proxy service.

Googling this error message does not reveal much information or troubleshooting steps. Has any one seen this before or can any one help us better understand the message we're seeing?

Running Zabbix_proxy (Zabbix) 5.2.6 on Ubuntu 18.04.5 LTS - 8 CPU + 16GB RAM on Proxy +/- 300 Active Agents & 3500 ICMP monitored hosts

[lptarik]

Hello did you able to solve issue?

[Megafisshy]

We upgraded to the latest available version of zabbix and added an additional proxy server to better manage the load, this seems to have solved the issue for now.

[syco]

I had the same problem, turned out the firewall was doing content filtering on port 10051 and 10050. TCP was allowed, but TLS was not.
An indication of this is that both the proxy and the agent report the connection terminated by the peer.
In this case was terminated by the firewall.

[santosra02]

Hi

I managed to fix this error in my zabbix proxy just enabling the parameter "StartTrappers=17" in zabbix_proxy.conf and restarting the service