How to combine several metrics in one?

[Aer0]

Hi!

I am a newbie here, have just started to use Zabbix.

I'm trying to create an Item which will monitor every core of the server CPU.

I found that i can create an Item to monitor every core separately with the key:
system.cpu.util[1]
system.cpu.util[2]
system.cpu.util[3]
system.cpu.util[4]
...

But in this case every item will work by their own schedule, so it will be hard to find the common peaks.

Can i combine all cores into single Item to get more precise data?

How?

Or it necessary to create an unique Item for every core anyway?

[cyber]

Hello.

Thank you.

I want to understand how evenly loads are distributed between all CPU cores.

Yes, I am using key system.cpu.util without additional parameters, but all that i can see at the Graph just only one line.

Same time we have lot of servers with more than 64 Cores per CPU, it will be hard to create such Items manually for a lot of CPU this kind.

No need to make them manually, discovery is for that.. system.cpu.discovery item and from there you can already create as many items as you have cores...

https://www.zabbix.com/documentation/6.0/en/manual/discovery/low_level_discovery/examples/cpu

And you can always have one item for "all" and separate ones for each core..

[ISiroshtan]

Why not just use system.cpu.util[]? By default it takes aggregated value of all CPUs. Any specific reason you want to split it into per-cpu view?

From documentation:

system.cpu.util[<cpu>,<type>,<mode>,<logical or physical>]

The CPU utilization percentage.
Return value: Float.
Supported platforms: Linux, FreeBSD, Solaris, HP-UX, AIX, OpenBSD, NetBSD.

Parameters:

cpu - <CPU number> or all (default);

type - possible values: user (default), idle, nice, system, iowait, interrupt, softirq, steal, guest (on Linux kernels 2.6.24 and above), or guest_nice (on Linux kernels 2.6.33 and above);

mode - possible values: avg1 (one-minute average, default), avg5, or avg15;

logical or physical - possible values: logical (default) or physical. This parameter is supported on AIX only.

[Aer0]

Hello.

Thank you.

I want to understand how evenly loads are distributed between all CPU cores.

Yes, I am using key system.cpu.util without additional parameters, but all that i can see at the Graph just only one line.

Same time we have lot of servers with more than 64 Cores per CPU, it will be hard to create such Items manually for a lot of CPU this kind.

[cyber]

Hello.

Thank you.

I want to understand how evenly loads are distributed between all CPU cores.

Yes, I am using key system.cpu.util without additional parameters, but all that i can see at the Graph just only one line.

Same time we have lot of servers with more than 64 Cores per CPU, it will be hard to create such Items manually for a lot of CPU this kind.

No need to make them manually, discovery is for that.. system.cpu.discovery item and from there you can already create as many items as you have cores...

https://www.zabbix.com/documentation/6.0/en/manual/discovery/low_level_discovery/examples/cpu

And you can always have one item for "all" and separate ones for each core..