Community
- Command Central
- MVP Program
- Monthly Mission
- Blogs
- Groups
- Events
- Media Vault
Products
- Observability
- Network Management
- Application Management
- IT Security
- IT Service Management
- System Management
- Database Management
Content Exchange
- SolarWinds Platform
- Server & Application Monitor
- Database Performance Analyzer
- Server Configuration Monitor
- Network Performance Monitor
- Network Configuration Manager
- SQL Sentry
- Web Help Desk
Free Tools & Trials

CPU mis-reporting on clusters

klc2009

We have two clusters:

NS1 (hardware is InfoBlox 550 HA)

which is comprised of nodes ma1 and ma2

and NS2 (also InfoBlox 550 HA)

which is comprised of nodes cf1 and cf2

The nodes on NS1 constantly report 100% CPU usage, but when the admin logs into the machine he can see that they are not truly using that much CPU. The nodes on NS2 report correctly.

I have built a custom poller using a formula that we obtained from InfoBlox support:

cpu_usage = 100 * (ssCpuRawUser.0 + ssCpuRawNice.0 + ssCpuRawSystem.0) / (ssCpuRawUser.0 + ssCpuRawNice.0 + ssCpuRawSystem.0 + ssCpuRawIdle.0 + ssCpuRawWait.0)

that I have assigned and added as a graph and a gauge to the node details page for ma1, ma2, cf1 and cf2. Our admin compared the poller results on ma1 and ma2 to actual usage and says that they are reporting accurately.

My first question is of course why is Orion seeing the CPU at 100% on the first set of nodes and how can we fix this?

And secondly, the poller graph and gauge on the second set (cf1 and cf2) show no data - the gauge piece shows this message: No pollers assigned or no statistic results. Click edit button to change universal device poller assignment!

When I click edit, the correct poller is assigned.

Thanks in advance for any help!

klc2009
.

Find more posts tagged with

100%_cpu

cpu_utilization

Accepted answers

All comments

FormerMember

If these are Windows, then the answer is that we are summing up the utilization across all the processes and thus including virtual memory and getting over 100% memory utilization.

neilmborilla

Hi Denny,

I think the question pertains to CPU Utilization and not for Memory. Or does the formula for CPU is the same with memory?

Question:

If the total memory = virtual memory + physical memory, The threshold for Nodes with HIGH Memory utilization should also be Virtual Memory + Physical memory also, not just physical memory?

Thank you

klc2009

They are not Windows machines, and I was told by SolarWinds support that the MIB on the device is causing the problem. I have resorted to just using the UnDP to set alerts.

klc2009

As mentioned above, I was using my transform poller to trigger alerts for these devices because Orion isn't picking up the correct value. After upgrading to 9.5, my alert suddenly fired and gave me a 100% usage stat, which again wasn't true.

I believe that what's happening is ${CustomPoller.Rate} is not the value I need to set my conditions against. In fact, I can't figure out which variable DOES contain the actual CPU usage rate any more. I have tried so many variable combinations that my head is spinning, and I still can't pinpoint where the actual usage value is being stored.

Please help me figure out which variable I should be using in my alert. The poller is made up of a formula that uses the value returned from 5 other pollers (ssCpuRawUser, ssCpuRawNice, ssCpuRawSystem, ssCpuRawIdle and ssCpuRawWait). Each of these gives the same value whether I use Rate, Counter, or Raw Value so I have them all set to Counter. Then the transform poller does a calculation: 100 * (({ssCpuRawUser + ssCpuRawNice + ssCpuRawSystem) / (ssCpuRawUser + ssCpuRawNice + ssCpuRawSystem + ssCpuRawIdle + ssCpuRawWait)).

When I test the poller against the appropriate nodes, I believe it is pulling the correct value (around 59 right now). However, when I set my alert conditions which checks for InfoBloxCPU (my transform poller) and then checks for Rate > 90, the alert fires.

It might help me if I knew which database table and column to look in to find the value as well. The custom poller tables are quite confusing.

PSI

You need to transform those custom MIBs

Have a look here for the custom mib polles here: