0

My WHM current service status is

  • Server load 31.53 (1 CPU) (red icon)
  • Memory Used 36.48% (1,122,544 of 3,076,964)
  • Swap Used 0.81% (8,540 of 1,048,568)

I'm under the impression that if the server load is above 1, I would notice some lags, let alone if it's 30+.

I was wondering why my sites/WHM/SSH seems to work just fine despite this?

7
  • What is the output of egrep 'processor|model name' /proc/cpuinfo, free and uptime?
    – jaume
    Mar 18, 2013 at 11:50
  • @jaume processor: 0 model name: Intel(R) Xeon(R) CPU E5-2630 0 @ 2.30GHz | 800MB | 19:54:24 up 58 min, 1 user, load average: 30.21, 30.41, 30.98
    – IMB
    Mar 18, 2013 at 11:55
  • Note that I am noticing some lags but I was expecting a "catastrophic" lag instead of a minor one.
    – IMB
    Mar 18, 2013 at 12:00
  • Thanks, but the output of egrep and free is not complete. Could you please add the complete output of both commands to your answer? Could you also add uname -a and top -n 1 | head -n 20?
    – jaume
    Mar 18, 2013 at 12:05
  • I'm not sure I understand. Why would 30% cpu usage be a problem?
    – terdon
    Mar 18, 2013 at 12:58

1 Answer 1

1

I do not think that load average means what you think it means.

A load of 30 means that there are 30 processes ready to run.

That is not the same as 30 simultaneously active processes, all trying to continuously use the CPU. Some of those 'ready to runs' might only want to do something for a fraction of a second.

E.g. if I have this situation:
Two processes are ready to run. (Lets call them process A and process B).

Both will check a single value, then decide they do not need to act and go back to sleep for a second.

Since there are two processes waiting to run the load at this time is 2.

Time line
Start --------------------------0.5 sec--------------------------1.0 sec
A runs and sleeps
                   B runs and sleeps
                                     CPU is idle CPU is idle  CPU is idle 

Notice that the load is 2, yet the CPU is idle most of the time.

If realize that this is a very fabricated example. If you had a RAM disk and two processes waiting to bzip some data on that RAM disk then the situation would be quite different. The CPU would be maxed out, the system would slow down and feel sluggish. But that is a complete opposite constructed situation.

In your case whatever your system is doing is closer to the first example.

(This is the best answer I can give with the information you posted. For a more detailed answer you will need to post much more information).

4
  • You pretty much corrected the way I understand server loads. I think I get it now. Thanks! I think what was happening to me was a combination of both examples. Most of the time everything's quite normal then at some time slower than usual, both having 30+ server load.
    – IMB
    Mar 18, 2013 at 14:18
  • Additional question though: What does it mean if the load is below 1? Like right now, I have 0.59 (2 CPUs) green icon. Load Averages: 0.85 0.75 0.67. (I upgraded 1 more cpu core just a few hours ago).
    – IMB
    Mar 18, 2013 at 14:20
  • Then there is on average less then one process ready to run. The on average is important here. Obviously there is never a 0.85th of a process.
    – Hennes
    Mar 18, 2013 at 15:28
  • But if you have had - on average - 30 processes ready to run for the last minute, and you don't have 30 cores on your machine, then you can probably benefit from some more resources for your current workload.
    – rich remer
    Jul 28, 2016 at 17:55

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .