misleading key agent.ping

[tekknokrat]

Why is the item with key agent.ping called "Ping to the server (TCP)" ?

In handbook it is classified as checking the agent availability so imo this no ping but heartbeat message from agent -> server
How does this behave behind zabbix-proxy?
Is the ping routed through proxy or is proxy in the context the server ?

Also the handbook should state about the result when agent is not available.
I am assuming 0 ...

[nelsonab]

Ya this one is quite confusing unfortunately. :-( Ping will always return "1" no matter what. Thus you'll need to use the nodata function to determine if a host is not reachable. The nice thing about it is that it is a very simple exercise of the Zabbix agent. Thus if you get the reply of 1 you know a) the remote host is up and b) the zabbix agent is running. However the nodata() function is somewhat buggy in my experience. For instance we recently had a problem where system would show offline (nodata(150)) even though there was lots of data over that period of time. quite frustrating.

[Alexei]

However the nodata() function is somewhat buggy in my experience. For instance we recently had a problem where system would show offline (nodata(150)) even though there was lots of data over that period of time. quite frustrating.

I would be interested to hear more details about this.

[nelsonab]

Unfortunately I don't have access to that system anymore, but I will email you if I see it again. I did see an issue where I had two triggers centered around nodata with different time periods on the same host and they would both trigger at exactly the same time even though they should not have.

I'll be upgrading that system to 1.6, I'll let you know.