CPU usage vs Idle process usage



  • The System Information widget of the dashboard display usually tells me that overall CPU usage of my Atom is 50% or greater. I do have a busy dashboard with 10 widgets, including the Traffic Graphs which updates at 1 sec and displays 4 of my 5 interfaces (1 physical, 4 VLANs). Dropping widgets and slowing refresh intervals does cause CPU usage to drop beneath 50% (say, down to 20%), but it seems to climb back to 50% or greater.

    At the same time, I have monitored using Diagnostics : System Activity and see that the two idle processes indicate much more idle time than expected given the CPU Usage. The update rate is faster (refreshed every 2 seconds vs 10 seconds on System Information widget), but it does not seem to be even close to the expected value (i.e. if I see 50% usage, I expect the last 10 values of WCPU to average 50%, if I see 20% CPU usage, then I expect the last 10 values of WCPU for the idle processes to average 80%, etc.)

    Can anyone explain the discrepancy between these two numbers?


  • Banned

    I'd suggest to use something reasonable… like, top from shell.



  • @doktornotor:

    I'd suggest to use something reasonable… like, top from shell.

    Why is it unreasonable to trust the diagnostic system activity?
    https://doc.pfsense.org/index.php/System_Activity

    Besides, none of these different views of CPU activity should be inconsistent. CPU usage is the opposite of CPU idle, so they better add up to 100%.


  • Banned

    Sigh. I'm merely suggesting proven tools designed for the purpose and providing pretty much realtime information. Javascript dashboard gimmick refreshed every X seconds is not one of them, not to mention that they - as you have already noticed - heavily affect the system resources.


Log in to reply