Strange poller utilization stats

[guzzijason]

I noticed something strange today.

My zabbix (3.0.4) server polls all hosts via proxies - it doesn't do any direct polling. Today, I noticed that one of the proxies was reporting high poller utilization. I've been running with the default 5 pollers, so changed this to 10 and the alert went away as expected. However, the strange thing is that the server was reporting (and alerting on) the same exact poller usage, even though if you look at the processes on the server directly, the pollers are fairly idle (as expected).

So, it seems to me that this might be a reporting bug? Here is a link to a screenshot of charts for both the server and proxy in question. You can see the poller utilization matches up perfectly between the two. Also, when I doubled the number of pollers on the proxy, the chart dropped equally on both server and proxy. Note that I *only* changed the number of pollers on the proxy - I never touched the server config. This doesn't make much sense to me.

Graphs: https://snag.gy/PVnjMD.jpg

Can anyone think of why this may be happening? That server graph seems completely inaccurate.

Thanks,

__Jason

[Semiadmin]

Hi, guzzijason.
Probably, this proxy monitored directly from zabbix-server. In this case "Internal checks" items display data from zabbix-server instead of zabbix-proxy. Just monitor this proxy via zabbix-proxy, i.e. via itself.

[guzzijason]

Thanks. Actually, it already is monitored via proxy, although that wasn't always the case. Originally, I did monitor it directly, but changed that several weeks ago.

__Jason