Absolutely, Positively, Incredible!

[Vin DSL]

This is the most amazing, jaw-dropping thing I have experienced at JagPC in my 2+ years of being here...

 Copyright © 1996-2004 Lenon.com - All Rights Reserved.

This Website is Powered by an Enhanced version of PHP-Nuke
The Most Advanced Content Management System in the World.

Script generation time: 1.3823 seconds ( PHP 87% - MySQL 13% )
SQL queries: 116 | Page views last 5 minutes: 8 | Server load: 3294%

$ uptime
  6:43pm  up 12 days,  6:29,  0 users,  load average: 205.18, 149.15, 98.61
$ uptime
  6:44pm  up 12 days,  6:30,  0 users,  load average: 217.26, 159.29, 104.15
$ uptime
  6:44pm  up 12 days,  6:30,  0 users,  load average: 223.01, 166.24, 108.22
$ uptime
  6:45pm  up 12 days,  6:31,  0 users,  load average: 229.99, 173.30, 112.41
$ uptime
  6:46pm  up 12 days,  6:32,  0 users,  load average: 237.09, 184.83, 119.85

Now that's what I call stable! How the hell is the web server still working?

[Ron]

You need to understand what that number is: It's the number of runnable processes in the queue.

Here are some questions for you.
What are the processes in the queue? What are they doing?
How long or large a time slice will each require?
What is the priority of the jobs in the queue?
What is the priority of the Apache process?

Most of those processes may be so stupid that they do nothing but poll a resource then go back to the back of the run queue. Who knows?

It's a snapshot of an isolated number that means nothing in isolation or without context.

[Vin DSL]

Yeah, I know, Ron. However, all things considered, when is the last time you saw a web server functioning with a load of 200+?

I remember when the servers, here, crashed at a load of 60+...

No reason to hash it out, other than to remark on my utter amazement at the whole situation. Tech Services was on it like stink on sh!t, and they took care of it in a matter of minutes - literally.

You know, this place rocks - really! They isolated the problem[s] quickly. Basically, one of my 'neighbors' was stealing processes from the server, plus they found some "malicious scripts" running, blah, blah, blah.

They took care of all this, rebooted the server, and I was back in business, in short order. I challenge anyone to tell me where else you get service like this for pennies on the wholesale dollar.

[Vin DSL]

BTW, if you think this was inconsequential, the second time it happened, everything was so jacked up I couldn't even shell into my account. The connection was being refused. Server load was at 5999%, per my script (6.0 = 100%). A little quick math tells me SSH quit working at or around a server load of 375, but the web server kept on 'trucking'...

Amazing!

[JPC-Greg]

geez, actually im surprised your site was even accessible at a load like that. At a server load of 4+ everything becomes sluggish, at 10+ its probably inaccessible half the time... 30 and shes gone. 300 ? wow

[Ron]

There are some processes that add to the queue depth but take very little in way of CPU, memory or any other critical resource. There have always been rumors that some versions of UNIX counted some types of I/O waits as runnable.

While I agree that seeing a high "load" number would cause me to immediately drop everything and go diagnose the health of a system, and "load" was certainly threshholded on my monitoring systems, in the absence of CPU utilization and other vital signs..... well, now I'm just repeating myself.