Server Load

SN3 · 11-01-2003, 10:46 AM

I was looking at the server load for my server and it was at 8.17 (4 cpus). isnt it suppose to be under 1.8 or something?

fishfreek · 11-01-2003, 10:52 AM

It really depends on wha its doing. A load of that level would be ok for short bursts but if its like that for prolonged periods of time then it could cause the server to become unresponsive.

Ideally server load should be under 1.00 over a period of time.

Various things like CPU and memory utilization go into account when calculating load averages.

SN3 · 11-01-2003, 11:07 AM

heres the whole thing:

Server Status

apache (1.3.28 (Unix)) up
bind (9.2.1) up
exim (exim-4.24-30_cpanel_stmpcontrol_antivirus_rewrite_mailman2) up
exim-26 up
ftpd up
imap up
mysql (4.0.15-standard) up
postgresql up
syslogd up
webmail up
Server Load 11.43 (4 cpus)
Memory Used 21.1 %
Swap 0.929 %
Disk sda3 (/) 71 %
Disk sda1 (/boot) 41 %
Disk sdb1 (/backups) 81 %

SN3 · 11-01-2003, 11:07 AM

the server load is at 11.43!

SN3 · 11-01-2003, 11:08 AM

now its at 7

SN3 · 11-01-2003, 11:09 AM

its bouncing at about 11 and 10. thats not good

SN3 · 11-01-2003, 11:09 AM

right?

fishfreek · 11-01-2003, 11:10 AM

To really have an idea as to what is going on and what is using more resources than others you need to get into a terminal session and run TOP

this will show the process's that are using the most resources, etc.

SN3 · 11-01-2003, 11:12 AM

how do i do that?

fishfreek · 11-01-2003, 11:17 AM

well first off you need either telnet or ssh access to the server. Alot of shared hosts dont give this access out.

And when i talk about over a period of time i mean more like over the course of an hour the average or over the course of a few hours. having four or five samples over a few minute time frame is not a good sample as someone might be doing maintance work on a large database or something that would artifically increase load. Or the server might be compiling stats at that time frame.

hatchet · 11-01-2003, 11:42 AM

damn.. mine's at .09 with 1cpu... oh wait .06

SN3 · 11-01-2003, 12:49 PM

uptime stuff:

Since: 06/10/2002
Outages: 61
Uptime: 96.221%

SN3 · 11-01-2003, 12:50 PM

all the specs:

Since: 06/10/2002
Outages: 61
Uptime: 96.221%
Year Outages Uptime
2003 20 99.015%
2002 41 92.065%

Year Month Outages Downtime Uptime

2003
November 0 0 hrs, 0 mins, 0 secs 100.000%
October 0 0 hrs, 0 mins, 0 secs 100.000%
September 0 0 hrs, 0 mins, 0 secs 100.000%

is it ok at least?

Jazzee · 11-01-2003, 12:53 PM

<OT> That's my birthday 6-10

</OT>

SN3 · 11-01-2003, 12:59 PM

Happy late birthday

Jazzee · 11-01-2003, 02:21 PM

LOL, Thanks, but I think you told me that on my birthday

fishfreek · 11-01-2003, 03:16 PM

Sept and Oct with no down time is good. Looks like they had some issues in 02 and early 03 though.

SN3 · 11-01-2003, 03:56 PM

Quote:

Originally Posted by Jazzee

LOL, Thanks, but I think you told me that on my birthday

well, happy birthday - late

and fishfreak... i think they had a problem with the other servers having outages... my server was down for like 20 min for upgrades - twice

I, Brian · 11-11-2003, 04:50 AM

With regards to server loads - the only real way to know if your server is overloaded is if the server is being unresponsive.

My chronicles-network site is hosted at Ventures Online - the server loads seem to average 11-15. Most servers cannot handle this, but I didn;t actually note a problem with performance. I e-mailed the managers there, who explained that their servers are basically souped up dual xenons, that can easily handle such loads. And that if I were to have any real problems with servers being sluggish, they would be happy for me to page them.

So, basically, server load figures can only be applied on individual servers - there is no general index on what constitutes a good and/or bad server load.