why high load, but no cpu usage?

JustinHoMi · 01-09-2006, 10:26 PM

Hi,

Recently my computer has had a very high load (4.0-8.0) on average, but very little cpu or disk usage. Let me show you:

Code:

 22:25:04  up 2 days, 21:02,  2 users,  load average: 8.00, 8.01, 8.00
73 processes: 72 sleeping, 1 running, 0 zombie, 0 stopped
CPU states:  cpu    user    nice  system    irq  softirq  iowait    idle
           total    0.0%    0.0%    0.0%   0.0%     0.0%    0.0%  100.0%
Mem:   495800k av,  478588k used,   17212k free,       0k shrd,   44500k buff
                    380576k actv,   56796k in_d,    6660k in_c
Swap: 1012052k av,   57288k used,  954764k free                  228712k cached

Code:

[root@server]# vmstat
procs                      memory      swap          io     system         cpu
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 0  0  57288  17212  44420 228708    3    4   130    38  115    41  6  0 93  1

You can see that there is a little bit of swapping... but even when there was no swapping at all the load average was 4.0.

Any ideas why this would be? I do believe my system has an issue with PAM authentication, which is being discussed here, but I'm not sure what the relation is, if any.

Thanks!
Justin

linmix · 01-10-2006, 04:32 PM

Not sure wqhy it happened, but CPU usage and Load are two completetly different issues.
Load describes the system bus throughput i.e. how much data is being accessed and or moved and the CPU doesn't have to be a very active part of that process.

gilead · 01-10-2006, 05:44 PM

Hi linmix, do you have any sites where I can find out more about load average? The only stuff I found yesterday did link CPU usage with load which didn't explain what happened to JustinHoMi.

Typical of what I saw was http://www.luv.asn.au/overheads/NJG_...luvSlides.html, http://www.redhat.com/docs/manuals/l...h-rhlspec.html and even http://wiki.linuxquestions.org/wiki/Load_average

JustinHoMi · 01-10-2006, 07:22 PM

I wish I had time to read through those right now. I will next week, but for now I'll just tell you that I resolved the problem. If you read through the other thread you'll see what happened. Basically, the audit daemon was causing other programs to freeze. When they froze, they would stay open, and keep any files that they were accessing open. So, if load has anything to do with the number of processes that are waiting for execution, then I suppose the load average could be deceptive if a program freezes and cannot be killed.

linmix · 01-10-2006, 11:56 PM

I've got no sites to back up what I said, I just remember this happenening to me and posting a question at the forum (here). In my case it turned out to be mail writing messages like mad.
The LQ Wiki links the two: "load average is a measure used to tell how busy the CPU is. load average show the average number of processes waiting for execution." But it does explain about 'processes waiting for execution'.

I pulled this from a mail archive > In Linux, the load average is the average number of
> processes in the runqueue PLUS the average number of
> processes waiting on disk IO.

gilead · 01-11-2006, 02:04 AM

Thanks for that - I've been asked the question here at work and realised my understanding of it was a bit vague.

linmix · 01-11-2006, 10:43 AM

Quote:

Originally Posted by JustinHoMi

I wish I had time to read through those right now. I will next week, but for now I'll just tell you that I resolved the problem. If you read through the other thread you'll see what happened. Basically, the audit daemon was causing other programs to freeze. When they froze, they would stay open, and keep any files that they were accessing open. So, if load has anything to do with the number of processes that are waiting for execution, then I suppose the load average could be deceptive if a program freezes and cannot be killed.

What other thread??