I recently added our Mimosa Backhaul to our Solarwinds server. I noticed the CPUs on both units are running above 90%. Is there a way to check to see what the issue is? I have looked through the gui and can find nothing.
The chip set we use in our products has several CPU’s. The first CPU is what SNMP reports, and that happens to be the one used primarily for spectrum analysis. That utilization is normal.
We are having the same issue described by Steve.
Is there a workaround? Any plans to change this? Reports all cores, report the average etc?
All our Mimosas are at the top of CPU Util list. Certainly they doesn’t look nice there! Removing the CPU monitor I don’t think its an option.
The values you observed are completely 100% by design, but we understand that it has caused unnecessary concern for you and others, so I will provide additional explanation that should reduce that.
The spectrum analyzer process is running at low priority on just one of several cores, and it does not affect the data path (no performance impact). In fact, having spectrum data allows accurate on-channel and co-channel noise calculations for modulation changes, and enables smart channel change decisions. Turning off spectrum analysis is not a viable option for radio performance reasons, and we believe it is important enough to (mostly) dedicate a core for this purpose.
One proposed solution to alleviate the concern is to remove the included generic MIBs which report CPU utilization only on that CPU, but I have been advocating against that because the MIB also includes other more useful device health information. I’m afraid that might happen if users insist that this is a problem after being presented with the explanation.
We currently have no plans to expose all core utilization information in the Mimosa-specific MIB, but I will present that request. If accepted, it is likely to be much lower priority than other feature development in the queue.
Thank you very much for the quick response and explanation.
I agree with you regarding removing the MIBs. However, since the reason is how Mimosa reports CPU utilization, it will impact all SNMP tools including Solarwinds. I think for all of us, it is important to monitor the health of the Mimosas and should have a higher priority on your pipeline. Even reporting the average (of the 4 Cores) could be an improvement. Its hard to rely on equipment we cannot obtain accurate health information from.