question on detecting host availability

[mconigliaro]

right now i have triggers set up to alert me when a host is unpingable or when the 'status' check = 2 (unreachable).

as i understand it (and i hope someone will correct me if im wrong), the value of the 'status' check is calculated based on whether zabbix_server has received new data from a host in a certain period of time. if this is the case, does that make my two 'availability' triggers redundant? it seems that since the ping check returns data just like any other check, zabbix_server would take that into account when calculating the value of the 'status' check. therefore, if i want to know if a host is up or down, i only need the 'status' trigger. is this correct?

i guess im just wondering what the best way of detecting host availability is.

[James Wells]

Greetings,

The answer is a bit more complex and alos a bit easier than you might think.

The status is, as you stated, a calculated value, however, it is calculated only on server to agent communication. If you use the icmpping, then status never gets updated, if, on the other hand, you use the agent ping, then yes, the status does get updated. Okay, so that is the easy part...

Now, where it get's complicated is when this actually happens. Like every other item, status is checked, or in this case calculated, based on the delay you set for the item. When the item fires on the server it goes through and checks to see if it has had any valid communication with the agent for the past X seconds. If it has had valid communication with the agent, and status previously had a value, then the value of status gets set to 1. If it has had valid communication with the agent, but the previous value of status is 0, then it remains at zero. Finally, if there has been no valid communication, the value of status gets set to 2.

Now, for your final question. There really isn't a best practice for this sort of thing. And to make matters even more difficult, ping can succeed, even when the machine is non-reponsive to any thing else. At least with the status check you know if the system is at least responsive to more than just pings.