What does this "uptime" number mean?

Hi,

I see on your status page that you show the uptime values for the last 5/10/15 minutes.
What exactly do these numbers mean.
Is this percentage of processor usage?
A complex calculation?
Something else?

Greetings from someone who really needs some holiday
  Meikel Weber
http://www.meikel.com

[Deb]

I see on your status page that you show the uptime values for the last 5/10/15 minutes.

The UPTIME is the number of Days Hours:Minutes showing... this is showing how long the server has been up since its last reset(reboot).

The numbers you are asking about are the Load Averages... separate in every way from UPTIME.

What exactly do these numbers mean. Is this percentage of processor usage? A complex calculation? Something else?

The true answer to this question is A complex calculation! But for the general idea....

The servers run multiple processes in memory.  Most of the processes that are in memory are 'sleeping' waiting for something to happen.  When a request comes in... it triggers one of those processes to 'wake up' and is put in the run queue to perform its task at hand.  The kernel has to now give the process CPU time. (e.g. someone just hit submit on a mail form)

Thinking of it layman's terms.. if you have a movie in the VCR and you press pause.. it's sleeping.  When you hit play it starts running again.  The load average is showing you how many processes are running and putting a load on the CPU. A VCR can only run one movie at a time.. The Servers of course can run many more than that.

A rule of thumb for our servers is as follows (noting that not all servers are setup exactly the same way and that the below is for the FIFTEEN MINUTE averages and *not* the one or five minute)

TAZ anything under three is fine... once over three it's considered high anything over 8 is considered dangerous

SIX --- anything under six is fine... once over six it's considered high and anything over fifteen to twenty is dangerous.

NINE -- same as six...

The one minute and five minute are basically irrelevant because any script etc (e.g. stats) can cause a quick spike.. If the average on the one or five minute is high but the fifteen minute average is showing normal then you can easily assume a particular script spiked up and has since completed its task. It is complex in nature to truly understand... but there for users reference type of thing.

Deb

[em9]

I was always told that when the load average goes above 1, that the processes are being cued, and so things are slowing down. Load averages should always be 0.something. Why is this not true with Fquest servers? They seem fast even though the load is 'high'....

[Deb]

It all depends on the type of software and hardware configurations used... as noted above, the TAZ server load needs to be lower than SIX and NINE's load averages simply because the TAZ build is not the same as SIX and NINE.

As I noted in the email to you concerning this question each server carries its own rules and the builds dictate what they can and can not handle.  A person is able to run many more processes on their PC when it has 64MB of RAM than they can if it only has 8MB of RAM.

Deb