SG-1000 CPU usage numbers



  • I'm a little confused with some of the usage numbers, and hoping someone can educate me..

    I understand the CPU is a single-core (https://forum.pfsense.org/index.php?topic=126872.msg700496#msg700496), so a CPU utilization of '1' would mean the single-core is at capacity. And that's what I'm seeing on my Diagnostic>System Activity page (and dashboard), with load averages of ~0.70.

    But when I go to the Status>Monitoring page, the graph indicates that I'm barely pushing 20%.

    So.. is my SG-1000 nearing capacity (~70%) or is it pretty lax (~20%)? Confused, looking for comments/understanding.




  • Banned

    Part of the first screen shot is cut off but I think that's telling you that your CPU is 79% idle (21% in use).

    Also, accessing the webgui causes a spike in CPU usage. Since you aren't using it 99% of the time its more accurate to look at the RRD graph averages than real-time usage while accessing the webgui.



  • Thanks for pointing out the 'Idle' record - I was meaning to ask about that also! I also hadn't considered the load of the webgui itself. I'd suppose there'd be even less usage if I used SSH instead of the gui.

    OK, so just to round-out my understanding (referring to the image below): Why do I have Load Averages in the 0.50-area, but the Idle process is in the 75% area? (note: I let it sit like this for 5 minutes to be sure).

    I'd expect these numbers from a dual-core system, not a single-core. Any idea?



  • Banned

    I believe that the difference you're seeing is just because you're using the wrong tool to figure out your CPU usage data.

    The three load averages are the averages from the last 1, 5 and 15 minutes.

    The WCPU column, or Weighted CPU usage uses a formula to try to figure out what the systems average use is.
    I don't know what that formula is or how it works.

    I think that if you go to status monitoring, select your CPU for 8 hours @ 1 minute intervals, then deselect the interrupts and processes on the graph it will give you a better idea as to what's going on.
    Add up the three utils for your total system use.

    Notice that on my system The load average is ~1.3%, but the max is ~26%.

    Looking at my graphs you can see that my user utils spends the vast majority of its time under 2%, but occasionally spikes up in the 15-20% range for a short period (probably a cron job running).

    Basically, I think that staring at a live update feed of top is a waste of your time (in this case). That information is already summarized and visualized for you in the RRD graphs, one of the great features of pfSense!  ;D

    If my understanding/explanation of any of this is wrong I hope someone smarter than myself steps in and corrects me!




  • @pfBasic:

    I think that if you go to status monitoring, select your CPU for 8 hours @ 1 minute intervals, then deselect the interrupts and processes on the graph it will give you a better idea as to what's going on.

    Basically, I think that staring at a live update feed of top is a waste of your time (in this case). That information is already summarized and visualized for you in the RRD graphs

    Well that does make sense. Sounds like I'm over-complicating things by looking at the Top averages instead of utilizing the RRD graphs. Perhaps I should just use the Top info when diagnosing an issue (if any!).

    Thanks for the clarification!


Log in to reply