How can one tell if a server is overloaded?

[eurorocco]

I use the "top" command and I see load average, and also cpu idle time.

For example...

10:43:22 up 2 min, 1 user, load average: 4.00, 1.63, 0.60
130 processes: 129 sleeping, 1 running, 0 zombie, 0 stopped
CPU states: cpu user nice system irq softirq iowait idle
total 2.6% 0.6% 6.2% 0.0% 0.0% 0.0% 190.0%
cpu00 1.5% 0.5% 4.9% 0.0% 0.0% 0.0% 92.8%
cpu01 1.1% 0.1% 1.3% 0.0% 0.0% 0.0% 97.2%
Mem: 2068952k av, 267676k used, 1801276k free, 0k shrd, 32104k buff
47156k active, 190196k inactive
Swap: 1052248k av, 0k used, 1052248k free 80708k cached

This is an Intel IV with Hyperthreading, so it seems to have two CPUs, cpu00 and cpu01.

Load average in the last minute is 4, which I think is high and pretty bad. Is it?

Idle time is 190% (out of 200%), which I think is low and pretty good. Is it?

So, would you say this server needs help or not? Maybe someone is abusing the server when this snapshot was taken?

I have seen worse, maybe load average around 200, but that would be a server that definitely needs immediate and urgent help, since the server will, at that level, not be responsive, and it will look really, really slow.

So, what should the limit be in terms of load average or idle time, or both, to say this server is okay, or this server is not okay.

Also, if I see cpbackup or cpanellogd (ie backups or web log processing like webalizer or awstats) then I tolerate higher values for load average and lower idle times, since I know these two tasks are quite demanding on CPU and Disk Usage.

Thanks in advance to your help responding these questions, providing your professional opinions.

ER

 

[ZapX.net]

This is only my personal opinion. I don't think you can decide just with one snapshot like that; though with a load of 4 I'd definitely look into which processes are creating the load.

If the norm is 4, though, then I'd definitely say you should consider splitting off onto another server. I personally don't like when load gets over 1.0 as the norm.

 

[PWSowner]

Best thing to do is check the load several times a day for a few days and see how it is on average. Server load always goes high for some operations, but should be below it's maximum most of the time.

 

[chirpy]

I agree that you need to observe it over a period of time. One thing you did not mention is your OS version and your kernel version. If you're running RHE or CentOs you should make sure that you've got the latest kernel installed and running.