cancel
Showing results for 
Search instead for 
Did you mean: 
Create Post
Level 12

Multiple Server CPU Monitoring

I tried to search for this already but was unable to find anything. I wanted to know how Orion handles a windows or unix server with multiple cpus? It only displays 1 cpu guage/chart. Does it average out all cpus into that 1 graph?

It would be nice if it can display an individual cpu guage/graph for each physical cpu in a server.

BB
0 Kudos
11 Replies
Level 12

quote:Originally posted by BryanBecker

Don...any update to this topic?

Thanks.

BB


Bump....SW support????
0 Kudos
Level 7

Along the same lines of this thread,    I am trying to figure out exactly WHAT Orion is monitoring with regards to the CPU?   Is it looking at the 1,5,15 minute load average?   Or is it taking all the IOWait, CPUUser,Kernel, NICE procs and averaging those?

 

The graphs of the CPU utilization do not seem to match what things such as TOP or SAR says with regards to a multiple CPU Linux box.   I am trying to clarify exactly what is being monitored and recorded for CPU utilization.

 

Thanks for any insight into this. 

0 Kudos
Level 12

Bumping this again.  We are not getting to the point where we will be monitoring alot of servers, most with multiple CPU and I have already been asked about how Orion handles multiple CPUs and how they are displayed in the graphs.

BB
0 Kudos
Level 12

I am also being asked to monitor more and more servers with Orion and I need the ability to handle multiple CPUs and Cores as well. 

0 Kudos
Level 19

It's the load average.

For Cisco devices, we use the 1 minute average (CISCO-PROCESS-MIB::cpmCPUTotal5min - 1.3.6.1.4.1.9.9.109.1.1.1.1.5) if your polling interval is less than 5 minutes, otherwise the 5 minute load average.

For servers (Windows, Unix, Linux, etc.), we use HOST-RESOURCES-MIB::hrProcessorLoad (1.3.6.1.2.1.25.3.3.1.2).

0 Kudos
Level 12

Outside of creating custom pollers why not split these out?

BB
0 Kudos
Level 19

Simplicity, mainly.

0 Kudos
Level 10

Quick Update on this Post.
 
You can Monitor each CPU using Custom Pollers.
 
Note this is more easier to do now with Orion 9 Universal device Poller which replaces v8 Custom Poller.
 
Upgrade to Orion 9.
 
You can download a template for such Poller for UnDP here:
 
Import this Template and assign it to Nodes that you need that passes  the Test Option.
 

Poller to monitor multiple CPU cores on single system. 

The default CPU monitor averages utilization of all CPUs\CPU cores. 

This poller will display a chart for each individual CPU\CPU core.

0 Kudos
Level 7

What good does this poller do?  All that it tells me is that each CPU/Core is either busy or not.  Isn't there some way to get CPU utilization from each CPU/Core?

 

 

Here is the OID I found for HP servers running Windows.  It seems to grab the current CPU utilization from each processor/core.  The next few OIDs in the tree are for 5 min, 30 min, and hour

 

MIB Name: CPQHOST-MIB:cpqhoCpuUtilMin

OID: 1.3.6.1.4.1.232.11.2.3.1.1.2

0 Kudos
Level 12

Don...any update to this topic?

Thanks.

BB
0 Kudos
Level 9

Bryan,
The current release averages the CPUs. A future release will support multiple CPUs either individually or combined as an average or seperate lines on the same chart.

The issue is CPU load is a property of the Node, not its own object as a child of the Node at this time.
0 Kudos