If you're talking about servers, we already return an "average" fake-cpu for this. You should NOT alert on anything other than the "average" fake-cpu on such devices.

adam.

Sent from Mailbird

On 01/09/2015 09:58:54, Cameron Murray <cameron@techpath.com.au> wrote:

Adam

Yes we have a delay of 30 minutes however when a single core is busy for a period on a task we get alerts however overall the servers usage may be less than 10%.

Maybe aggregate cpu and then alert of this?


-------- Original message --------
From: Adam Armstrong
Date:01/09/2015 18:56 (GMT+10:00)
To: observium@observium.org
Subject: Re: [Observium] Multiple CPU

We return an "average" fake-cpu on unix systems. This might be useful to extend to others, but there is the issue that on some device types it might be difficult to distinguish between cores used for control plane and cores used for data plane.

Also, you should definitely have a delay set for processor usage.

adam.

Sent from Mailbird

On 01/09/2015 09:52:10, Cameron Murray <cameron@techpath.com.au> wrote:

Adam,

The silence is painful.  At the moment we've had to disable cpu alerts.


-------- Original message --------
From: Ben Hohnke
Date:01/09/2015 11:03 (GMT+10:00)
To: Observium Network Observation System
Subject: Re: [Observium] Multiple CPU

I'm interested in this also, as we have the same issue with our multi-core routers from Mikrotik



On Tue, Sep 1, 2015 at 10:58 AM Cameron Murray <cameron@techpath.com.au> wrote:

Guys,

 

I have checked the guides however Not seeing how to configure CPU alerting based on multiple CORE CPU’s.

 

We have quite a number of windows virtual machines with 2 – 16+ cores and currently use “processor_usage greater 65” however this is alerting when a single core reaches >65%

 

What is the best way to manage this?

 

Regards,

Cameron Murray

_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium