The Old Joel on Software Forum: Part 4 (of 5) - Monitoring Server Resources in Unix

Monitoring Server Resources in Unix

What would commands to use to monitor system resources in UNIX?

I have used 'top' and 'src' - I was wondering if there are any others that can be useful.

Thanks.

Alfred E Neuman
Thursday, February 26, 2004

ps - list of processes running on system
lsof - list open files for a process
netstat - various network statistics

Rhys Keepence
Thursday, February 26, 2004

I use glance all the time. It will show you disk usage, memory usage, cpu usage, and network i/o (I think). It also breaks down each number into System consumed and User consumed. You can check the CPU stats (C) to view the issue queue (should be at 5 or below for system responsiveness). It's a pretty handy tool.

Also, if you have certain types of processes you want to check, for instance you want to see what processes are operating under a particular user, use ps -ef | grep <username>.

Lou
Thursday, February 26, 2004

Have a look at Nagios http://www.nagios.org/ or Ganglia http://ganglia.sourceforge.net/

Ganglia is more for clusters, but Nagios can be configured in many combinations. You can also check windows machines remotely, giving facts like CPU Load, Disk Usage, Uptime, Service States, Process states, Memory Usage, File Age and Most Perflib counters.
(with a http://nsclient.ready2run.nl/ plugin. I have not used the windows check.. )

Your whole network can be displayed in a nice overview, and you can configure mail warnings, pager warnings etc if a service fails.
Runs with apache and can be accessed by any browser.

Fredrik Svensson
Friday, February 27, 2004