What is "load average"?

[RenoWV]

Following a recent 15 hour server outage, I installed a cgi script that tells me some data about my server, including "load average".

What does that mean? Is a higher number worse than a lower number? When does it get to be a concern?

Here is an example of what the script displays:

-----------------------------------------------------------------
10:20am up 11:51, 0 users, load average: 4.51, 4.37, 4.42
-----------------------------------------------------------------

Thanks for any illumination...


__________________

 

[jimbopalmer]

this looks like the uptime command, which shows the number of programs which wish to use computer time for 'recent' time periods 1 minute 5 minutes and 15 minutes as I recall. If you have 5 or more CPUs, then over 4 of them are in use recently, if you have less than 5 CPUs, you could have kept more CPUs busy than you have. (On my sun with 4 CPUs, if the load rose over 35, I was doomed. It would never recover.)

the rup command shows this info for all unix boxes on your network

sar -q shows a similar number, but it only shows the amount OVER the number of CPUs you have.

I tried to remain child-like, all I acheived was childish.

 

[RenoWV]

Thanks Jimbo. Since posting my question, I've done a little research and found this page:

http://www.teamquest.com/html/gunther/ldavg1.shtml

which is *highly complex* and therefore totally over my head, however Dr. Gunther does confirm what you said:

"The [load average] triplet is intended to provide you with some kind of information about how much work has been done on the system in the recent past (1 minute), the past (5 minutes) and the distant past (15 minutes)."

Most of his article is very mathematics oriented, but in general I think I have the drift -- thanks for your feedback....

 

[jimbopalmer]

Load average is the number of jobs that wish they were running (including those that are). In your example that is over 4 but less than 5.

if you have over 4 CPUs then this load average does not imply jobs are waiting, but if you have 4 or fewer CPUs, then jobs could run faster with more or faster CPUs, because all this numbers 'agree' , this is a sustained load, not just a spike.

the sar -q command shows daily averages but in a twist, it only shows the jobs which could not get a CPU to run on, not the running jobs.

I tried to remain child-like, all I acheived was childish.