We have our NMS polling the ePMP 1000 / 2000 radios regularly for monitoring of the network and this is one of the items we poll for. I do realize that the value returned needs to be divided by 10 to give us 0.0 - 100.0 for the values. However, I question the accuracy of this value. We see many times on completely new AP's that it will spike into the 90%+ range for an extended period of time! We have even set the NMS so that CPU must exceed 90% for 15+ minutes before it is considered and issue and multiple times / day random AP's will report this back. We have done traffic capturing of these AP's during this time, looked at connections,... There is nothing going on, just seems that this SNMP value is pretty much random. As mentioned it can even happen on AP's with NO customers linked!
This has been happening for us for MANY releases and continues in the current 3.5.1 release. Any chance this can be fixed or do we simply need to ignore CPU usage.