Hi,

I need to redirect the output of top command to a text file for monitoring the CPU utilization later on. I am using RHEL 3 AS (2.4.21-32.EL).
I used the following command:

#top -b -n 1>top.txt

but the command does not exit by itself and the text file size keeps on increasing until killed manually using Ctrl-D.

Please help!!!!

P.S. Is there a way by which I can take only the first 10-15 lines of top's output and discard the rest?

Hi,

it works as expected.

'top' command goes through one iteration, then exits.
Using 'top' - version 3.2.7


To use only first 15 lines, try 'head' command.

Other tools for collecting historical data are sar and uptime.

Sar is set to run in 10 minute intervals by default (RH). To see if sar is running simply type:
sar
this will show data for the previous day only. Lots of other options though, as sar is very powerful. man sar for more info.

Load Average is a very useful tool for telling how busy a server really is. ie: if the cpu is at 90% but your load average is only 1.3 0.4 0.08 then you have nothing to worry about.
Load average by default is showing you the load average for the past: 1 minute, 10 minutes & 15 minutes. I will run uptime via cron every 10 minutes just to monitor the load average on a machine. ie:
/usr/bin/uptime >> uptime.out060607
Just redirect the output to a file, as above. Then, if desired you could email the file yourself at midnight, and zero out the file for the next day of stats.
A general rule of thumb is any load average under 2 is fine. Load average equates to a line at a movie theatre. If more than 2 people are waiting in line, service is not so fast. By the same token a load average of 2 means two processes are waiting for service by the CPU. Anything over that, if sustained, your response time will be affected.

"sar" is a good tool, but sometimes it is necessary to get the detail "top" provides. Running in batch mode should work fine.

@DotHQ: sar will give you the load averages, so you are collecting this twice.
Oft quoted but not necessarily true.
The load average also includes tasks in "uninterruptible sleep". Generally this means (disk) I/O. Can be the major contributor to the loadavg numbers, especially if the runqueue is short.

I do believe that is why it's called a 'Rule of thumb'. It is more true than untrue in my opinion.

Thank you!! One and all. The info provided was really helpful.