Confused with CPU Load Average never below 1

[bROTHER]

Hi,

On a 2 cores system (AX86s) CPU idle but Load average never drops below 1.
I was expecting idle to be near 0 and 2 maximum processes working without delays.

Regards.

 

[ColinTaylor]

That's normal for that processor, the baseline is 1 rather than 0.

 

[Crimliar]

When you are running something like htop, it'll show you the individual processes. When little else is happening the simple act of ssh-ing into the router and then running htop itself will be a significant part of that very small processor load.

 

[bROTHER]

Hi,
So, if "1" is the minimun level, ¿should "3" be the "max" in a 2-core (AX86s) and "6" in a 4-core (AX86U)?
Regards.

 

[ColinTaylor]

So, if "1" is the minimun level, ¿should "3" be the "max" in a 2-core (AX86s) and "6" in a 4-core (AX86U)?

No. CPU load doesn't work that way, there is no "max" value. CPU load is not the same CPU usage. (Google for a more detailed explanation of CPU load.)

 

[bROTHER]

Hi,

Yes, I know. That's why I wrote "max" between quotation marks. Thanks for the note anyway.
I'd like to stick to configurations that avoid leaving the system in a continuous high CPU Load when there is really no need.
But to acomplish this, I need to know the numbers, let it be 0 to 1 per core or 1 to 2 per core. That's really what I am asking because I used to the 0 to 1 per core rule until I noticed this.

Regards.

 

[ColinTaylor]

I'd like to stick to configurations that avoid leaving the system in a continuous high CPU Load when there is really no need.

The things I keep an eye on are the per processor CPU usage and the iowait time. If both of those are low (i.e. near zero) then I'm happy.

 

[bROTHER]

When you are running something like htop, it'll show you the individual processes. When little else is happening the simple act of ssh-ing into the router and then running htop itself will be a significant part of that very small processor load.

Hi,

That's it!
Noticed after a reboot that the 15 min CPU Load average was below 1.

So, its 0 to 1 per core.

Thanks.

 

[ColinTaylor]

Noticed after a reboot that the 15 min CPU Load average was below 1.

So, its 0 to 1 per core.

No it isn't [EDIT: Or to be clearer, the load number is 1 + (0 to 1 per core) ]. The only reason the 15 minute load average was below 1 was because you had recently rebooted your router. So the data for some of the time period didn't exist (and hence were calculated as zero).

Leave your router on for more than 30 minutes then see if after that the load average ever goes significantly below 1.00.

admin@RT-AX86U:/tmp/home/root# top
Mem: 733356K used, 200608K free, 3312K shrd, 3648K buff, 385152K cached
CPU:  0.0% usr  0.0% sys  0.0% nic  100% idle  0.0% io  0.0% irq  0.0% sirq
Load average: 1.00 1.00 1.00 2/171 21927
  PID  PPID USER     STAT   VSZ %VSZ CPU %CPU COMMAND
 1685     1 admin    S    15020  1.6   1  0.0 networkmap --bootwait
 1696     1 admin    S    19444  2.0   0  0.0 conn_diag
  345     1 admin    S    18532  1.9   3  0.0 /bin/swmdk
 1717  1696 admin    S    18332  1.9   1  0.0 amas_portstatus
 1694     1 admin    S    18332  1.9   0  0.0 roamast
 1740     1 admin    S    18292  1.9   2  0.0 amas_lib
 1519     1 admin    S    15260  1.6   3  0.0 /sbin/netool
 1721     1 admin    S    14440  1.5   2  0.0 cfg_server
    1     0 admin    S    14232  1.5   3  0.0 /sbin/init