Hi everybody!
I've installed zabbix recently (zabbix 1.4.5, SLES 10 SP2) and liked it very much. But i've some questions.
Most of my servers have multiply CPU's with multiply cores, but some applications (squid for ex.) are using only one core, but load it by 100%.
I used to monitor cpu load with cat /proc/stat | grep "cpu ", this is how it looks like (preview, click to see full-size):
Zabbix shows something like this (preview, click to see full-size):
I tried to create my own graph CPU utilization by cores and i got this (preview, click to see full-size):
I used system.cpu.util[1,,avg1], etc to get this.
If i run top on this server i see this (preview, click to see full-size):
So, how can i get real utilization per core? Which method is correct? Maybe none of these? How do you measure CPU load of your linux servers?
Another question is how to measure servers http (dns, sql, etc) requests per second using zabbix_agentd?
Thanks in advance.
P.S.: just in case: sorry for my bad English, i'm Russian.
I've installed zabbix recently (zabbix 1.4.5, SLES 10 SP2) and liked it very much. But i've some questions.
Most of my servers have multiply CPU's with multiply cores, but some applications (squid for ex.) are using only one core, but load it by 100%.
I used to monitor cpu load with cat /proc/stat | grep "cpu ", this is how it looks like (preview, click to see full-size):
Zabbix shows something like this (preview, click to see full-size):
I tried to create my own graph CPU utilization by cores and i got this (preview, click to see full-size):
I used system.cpu.util[1,,avg1], etc to get this.
If i run top on this server i see this (preview, click to see full-size):
So, how can i get real utilization per core? Which method is correct? Maybe none of these? How do you measure CPU load of your linux servers?
Another question is how to measure servers http (dns, sql, etc) requests per second using zabbix_agentd?
Thanks in advance.
P.S.: just in case: sorry for my bad English, i'm Russian.