CPU Load Monitoring

[clayton713]

I've been handed zabbix monitoring, and I'm finding out that the items added for CPU Monitoring are not functional. We have been using a custom script that does not work properly. I'm a intern in the IT department with no one to ask

Basically I have been trying to use the key system.cpu.util[avg1]
but this returns no data. All of the other keys work fine.

About the server I'm trying to monitor Its on a 64 bit VM and the OS Is Win Server 2003 R2 SP 1.

The zabbix version is 1.8

I've looked all over the forums for the solution, and indeed found the system.cpu.util[avg1] key, but it doesn't seem to work. Nor does system.cpu.util[,,avg1]

I'm probably missing something or double posting, and if so I apologize.
Thanks in advance.

Sean Clayton

[MrKen]

You missed a comma.

system.cpu.load[,avg1]

works for me. Without the comma doesn't work.

MrKen

Edit: and reading your post again, I see you want system.cpu.util - I messed up!

Check here, it may not be supported by windows http://www.zabbix.com/documentation/1.8/manual/config/items#supported_by_platform

[clayton713]

thank you very much

It looks like the system.cpu.load is supported by windows with no parameters. Such is life. It looks like I might be able to use the perf_counter method.
I just wanted to say I appreciate your reply. Thanks.

[Kerrygeek]

I'm monitoring CPU utilization on Windows servers with SNMP, no agent needed. Just enable SNMP on the server and add the IP address of your Zabbix server to the settings on the Windows box. You can use snmpwalk to see what parameters are available on Windows. Be sure to set the right community string in Windows to give Zabbix read-only access.

[obeiro]

Windows Monitoring SNMP Template

Hi,

Would you please share your SNMP Windows Monitoring template?
It would be great.

Thank you.

[Kerrygeek]

I set up a template but it turns out that different brands/models of servers use different MIBs for the cpu load. Besides CPU, I monitor bandwidth in/out and a ping to each server. Just do snmpwalk to each windows server and you'll see what parameters are available to you. Bare minimum, you'll be able to get CPU utilization, bandwidth for each network interface and you can ping it for an up/down test.

For example, CPU is hrProcessorLoad.1 (or .2, etc) for each CPU.
Bandwidth is ifInOctets.1 and ifOutOctets.1 for the first interface.

Good luck!
Kerry