Monitoring multi core / (multiple CPUs) servers
with top, tload and on Linux
The default GNU / Linux
top command does allow to see
statistics on servers and systems with multiple CPUs.
This is quite beneficial especially on Linux systems which are not
equipped with
htop which does show statistics to the
multiple-core system load.
To examine the multiple CPUs statistics with the default
top
command available on every Linux system and part of the
procps -
/proc file system utilities
1. Start top:
linux:~# top
When the top system load statistics screen starts up
refreshing,
2. press simply
1
You will notice all your system cpus to show up in the top
head:
As I have started talking about
top, a very useful way to
use top to track processes which are causing a system high loads
is:
linux:~# top -b -i
This command will run top in batch mode interactively and will show
you statistics about the most crucial processes which does cause a
server load, look over the output and you will get an idea about
what is causing you server troubles.
Moreover if you're a Linux console freak as me you will also
probably want to take a look at
tload
tload command is a part of the
procps - /proc file
system utilities and as you can read in the tload manual
tload -
graphic representation of system load average
Here is a picture to give you an idea on the console output of
tload :
Another tool that you might find very usefel is
slabtop it's
again a part of the
procps linux package.
slabtop - displays a listing of the top caches sorted by one of
the listed sort criteria., in most of the cases the
slabtop kernel cache monitoring tool won't be necessary for
the regular administrator, however on some servers it might help up
to the administrator to resolve performance issues which are caused
by the kernel as a bottleneck.
slabtop is also used as a tool by kernel developers to write
and debug the Linux kernel.