Hello all,

I would like you know if there is a rule of thumb or a chart that could feel me how the server is coping.

I have a Bandwidth hourly monitor, and in shell using "TOP" i can see the load and the CPU usage and Memory usage.

What i would like to know is, for example:

the server is using 1Mbit of the connection constantly, the load fluctuates from 0.85 - 1.57 and the CPU is about 30% including Hyper-Threading

How would i know, maybe by a calculation if the server is coping well or not with the values given. I do not have a problem with the server, but i would like to know more on how to see if there is a problem with the capabilities of the server at anytime if a problem comes up, and if its the server doing to much or not.

I hope this is explained ok.

Thank you.

Average usage is 1, top of the line is 2, anything above and beyond 2 you can leave until it reaches 3/4, at which point you should investigate what is causing the high load (top and free will tell you).


Ok make sure you have built an SMP kernel because HT will look like 2 CPUs.

If it is a web server make sure to check stuff like total apache processes:

Compare it to MaxServers in your httpd.conf, if too high look into the status of the threads kill the bad ones and hopefully that will settle your server.

There is a lot of other process handling tips required but from the stats you are showing here, nothing untoward is going on.

Cool, ok thank you znx

okay... so... is there a way to find out how it's running overall, I mean, i understand what you said, and its not a web server, but its a multi-funcation server.

But i know what the load should be under... but is there a way to know how the server is running, as i said like a calculation?

Load * connection / cpu

or something. I need to know what to look out for as a whole, not just the load.

I have a connection graph, and i have 10Mbit to play with, so i know if there is to much traffic or not, but how do i know if the server is coping with the traffic or not.

Is that where LOAD comes in, and CPU usage ? this is what im not 100% sure on.

thanks

I haven't heard of any calculation with regards to working out the status of a box as a whole. Some obvious things to ensure are:

Server Checks
Connection - Can you connect (ping testing is the simplest)
CPU - What is the loadavg at anything > 3 investigate. (notice CPU% isn't required)
Disks - What is the space free on the system, if >90% look into getting more storage.
Processes - Simple, count the number of processes. Like I expect httpd, mysqld and ftpd to be running so a quick count of the number will tell me that all is well.
Logs - Grep the logs for information about errors that are occuring.

Service Checks
DNS - FTP - HTTP - IMAP / SMTP / POP3 - SSH - ETC!
- Run similar checks on these, for instance with nc you can test several of those, expect will do the others.

I know this maybe isn't what you are looking for but in short there is not an easy way to quantify all of those checks.

znx said everything, there is no such formula. In fact it wouldn't make any sense as every server is built different for its own purpose. You might wanna have a look at some monitoring software like Nagios and/or Cacti.