Active agent over unreliable connection

[wilson.waters]

Hi all,
We have a situation where a number of hosts are behind a unreliable NATed connection. We don't need to access these hosts from the Internet normally though I would still like to monitor them using Zabbix. These hosts also get shut down for 8+ hours per day (they are on a remote site powered by gensets).

I have them set up as active agents in Zabbix but every time the host goes down the Zabbix queue jumps by 100 or so items and triggers an alert.
"More than 100 items having missing data for more than 10 minutes"
I gather this is correct behavior (as these items are indeed queued while the host is offline) but is there a way to change this?

Ideally I would like to disable all item checks for an active agent if it hasn't provided data for the last x seconds (and obviously re-enable them all as soon as the host is back).

Thanks,
Wilson.

[richlv]

so in your environment you probably don't care about that item/trigger - just disable it

also no need to disable any items for active agents as they will simply stop passing data... disabling themselves pretty much

[wilson.waters]

Fair call!

My only concern is it would make monitoring performance of the zabbix server confusing when we eventually do hit real performance bottlenecks? Perhaps I just don't like seeing red boxes anywhere

[richlv]

you shouldn't see red Zs with active items - they should only appear because of failed passive items

[wilson.waters]

Correct, no red Zs. I only see red items under administration->queue->Zabbix agent (active)->More than 10 minutes.

I have done as you suggest and disabled the trigger. I'll deal with Zabbix performance issues if/when we hit them (I presume this is the purpose of the queue admin page? Performance monitoring?).