[nzlug] Load stuck at 1

[Nick 'Zaf' Clifford]

Steve Holdoway wrote:
> On Wed, 07 May 2008 19:15:16 +1200
> "Nick 'Zaf' Clifford" <zaf at nrc.co.nz> wrote:  
>   
>> Ummm... no.
>> By definition.
>> Load averages are, as they say, the average system load for the past 1,
>> 5 and 15 minutes.
>> If you have a system with a 15 minute load average of 250, you have
>> something very, very, broken (or an extremely busy system).
>>     
>
> Sorry, Zaf but you're wrong. I even quoted example figures. Yes, it's an extremely busy system, but if it were true, then there would be no chance at all of even logging in, let alone using it as an interactive platform.
>   
Yup, as I said, it could be an extremely busy system, that said.....
> I've sent you an image of said system, where I just logged in, and ran up top. It's debian stable, and has been rebooted twice in the last 18 months. One of them was to move the server to another data centre.
>   
Yup, ok, looked at the images, a large number of processes (400+), with
what looks like almost all of them sendmail processes that are in the D
state (waiting on Disk). So if you'll go back, you'll notice I did
mention that processes waiting for disk also raise the load average.

> I do wonder why you think I'd make something like this up, though (:
>   
Didn't suggest you made it up, I said that I find it hard to believe.
Maybe I didn't speak well, and maybe I did come across too
confrontational, my apologies.

As it stands, I look at that, and I'm amazed. To me it seems totally
broken (this time I'm referring to sendmail, not your system) to have
400 or so processes all waiting for disk. IMHO sendmail shouldn't be
trying to deliver once it saturates the disk. But I'll concede that I'm
not a busy mail admin, and the busiest mail system I've worked on had 10
emails per second on average, so I may be wrong (again!).

But either way, I don't plan (or want) to start a flame war about load
averages, sendmail, or vim (did I mention that its better than emacs?
*grin*)

Nick