Why Proxy Pollers above 75% busy?

[jonathan.vargas.cr]

Hi,

We setup a very small Zabbix scenario for testing the architecture we want to monitor, and now I get a "ZabbixPoller processes more than 75% busy" alert and I would like to get an idea it's being triggered.

My architecture diagram is attached. Resuming:

1 Zabbix Server monitoring 4 Cloud Servers
1 Zabbix Proxy monitoring 4 Internal servers

I'm providing:

* My Architecture Diagram: http://i.imgur.com/tnXKauE.png
* Zabbix Performance Graph: http://i.imgur.com/JKSBCVV.png
* Zabbix Queue Graph: http://i.imgur.com/kHkuSUq.png
* Zabbix General Stats: http://i.imgur.com/SezzmTs.png

Question is: Why, if I have only one Proxy server monitoring only 4 hosts, does it keeps 75% of time busy?

I've tried increasing StartProxyPollers from 5 to 10 but that doesn't poduces any change in performance. The CPU usage and general performance of both Zabbix Server and Zabbix proxy operating systems is ok.

Where is the "busy" bottleneck? Between Server and Proxy? If so, what could I do to improve it?

Thanks in advance.

[tchjts1]

When you changed that poller setting from 5 to 10, did you restart the proxy process for it to take effect? If you made the change on the existing line, did you remove the comment symbol # from that line? Changes to commented lines are ignored and it will use the default setting.

If the above were done correctly and you still have this issue, would you mind sharing the contents of your zabbix_proxy.conf file? You can xxxx out any sensitive information.

But I agree with you initially. 75% seems high for the small number of items/triggers/NVPS.

[jonathan.vargas.cr]

When you changed that poller setting from 5 to 10, did you restart the proxy process for it to take effect? If you made the change on the existing line, did you remove the comment symbol # from that line? Changes to commented lines are ignored and it will use the default setting.

If the above were done correctly and you still have this issue, would you mind sharing the contents of your zabbix_proxy.conf file? You can xxxx out any sensitive information.

But I agree with you initially. 75% seems high for the small number of items/triggers/NVPS.

Thanks for the reply tchjts1,

Sure, zabbix server was restarted. I just noticed I changed a different parameter, named "StartPollers", not "StartProxyPollers".

After changing the later parameter from 1 to 2, now the busy indicator decresed to 37%, which seems to be fine.

Thanks!

[fernandomullerjr]

Hi,
Adjust the conf, changing the values of the Zabbix StartPollers and Zabbix StartPollersUnreachable fields.

Values depend on several issues, adjustments, tests, and validations are necessary, and may vary depending on the environment.
Here is what was applied in our case: https://devopsmind.com.br/en/trouble...-processes-75/