Wrong CPU utilization values

Products

• DX Unified Infrastructure Management (Nimsoft / UIM)

Issue/Introduction

We are using SNMPCollector probe to monitor devices and we have observed that for one device the CPU utilization is going more than 100%. The device has multiple CPUs and for few CPUs the utilizations was more than 100%. Same device is also being monitored via snmp using the Performance Center tool and that shows the correct value. I would like to know why it showed a wrong value in UIM using the snmpcollector probe.

 

Environment

Release : 9.x, 20.x

Component : UIM - SNMPCOLLECTOR

 

Resolution

We have tested this in our in-lab. Please find our findings as follows:

For the device with issue, the OID responsible for CPU utilization for CPU number 38 is  1.3.6.1.2.1.25.3.3.1.2.38. Here are more details when we run the vendor test on CPU number 38 from snmpcollector probe:

 

INDIVIDUAL SNMP OID RESPONSES (NAME :: OID = Current Value :: Last Value)
hrDeviceDescr :: 1.3.6.1.2.1.25.3.2.1.3.38 = Unknown Processor Type :: Unknown Processor Type
hrProcessorLoad :: 1.3.6.1.2.1.25.3.3.1.2.38 = 7 :: 7

DELTA OIDS (NAME :: OID = Current Value :: Last Value)

UNSUPPORTED DEVICE OIDS (NAME :: OID = Current Value :: Last Value)hrDeviceStatus :: 1.3.6.1.2.1.25.3.2.1.5.38 =  :: 

METRIC NAME (or) VARIABLE NAME :: VALUE :: EXPRESSION
Indexes :: 38 :: hrProcessorIndex

Utilization :: 7 :: hrProcessorLoad

cpuIdleUtilization :: 93 :: 100-hrProcessorLoad

Names :: CPU 38 :: "CPU " + hrProcessorIndex

Descriptions :: CPU Unknown Processor Type 38 :: "CPU " + (isdef(hrDeviceDescr)? (hrDeviceDescr + " ") : "") + hrProcessorIndex



TABLE OID :: VALUE 

 

 

The issue here appears to be with the OID 1.3.6.1.2.1.25.3.3.1.2.38, this needs to be checked by the vendor as to why snmp agent deployed on the end server is providing the erroneous result. The snmp manager (which in this case is snmpcollector probe) just receives the data from the snmp agent.

 

With regards to the question with Performance Center (PC) showing correct value, most likely the CPU Utilization in PC is being calculated by another OID, not 1.3.6.1.2.1.25.3.3.1.2.38. 

 

The following is the vendor test screenshot from the snmpcollector probe: