Agent monitoring

[marc55]

Hi,
In zabbix (2.2), I can see two ways to monitor agent availability:
using agent.ping or using zabbix[host,agent,available] (internal check)
Whereas agent.ping returns no data in case of unavailability, zabbix[host,agent,available] returns a 0 (if unavailable).

What is the difference between these two checks, and which one has to be used to properly monitor agent availability?
Thank's a lot
Regards,
Marc

[jamesNJ]

Did you mean net.tcp.service[] simple check?

It probably doesn't matter which you use so long as you get the desired effect.

[marc55]

Thank's James,
But I was really talking of internal check zabbix[host,agent,available] and agent.ping.
Seems that internal check one's just read the agent status that is displayed by zabbix on UI, but of course in background, zabbix must do something to check the status.
So in fact, I really don't know what is the difference between the two checks.

[LEMORALES]

HI @marc55,

Well, I can't tell what is the difference between the two checks, but I want to share the solution for a issue that I had, and could be solved thanks to your post.

let me say that I used to check the agent status using agent.ping. Suddenly, one day, all my dashboard was full of issues telling me that more than 40 host had the agent unreachable. I doubled checked in the hosts, the agent was up in each host, the network was fine, I was able to telnet from zabbix server to the hosts using port 10050, and also when I ran zabbix_get -s hostname -k agent.ping I got back the number 1. So, everything seemed to be ok.

I'm suspicious about a bug (I'm using zabbix server 2.2.5 and the clients are AIX servers using agent 2.2.1 version), but I'm not sure how to diagnose the possible bug, either the server or the agent.

So, I used zabbix[host,agent,available] (internal check), and the checking of the agent become reliable again. And as you pointed, now when the agent doesn't respond I receive the number 0 instead of blanks when I look at the graph, which is more refined.

I'm sure that there is a technical explanation about the difference between agent.ping and the internal check, and what is the proper use of each one, but I don't know what it could be. I think that I know someone who could give us some enlightment about this, so I will try to ask him about it, and if he give me an answer, I'll let you know.