One thing guaranteed to bring on swearing is an OS that’s tanking. Some process sucks up CPU or disk and pulls down the whole OS with it. *nix flavored operating systems make this easier to deal with by exposing OS internals and providing you with tools to process that information.

What’s bothered me until now is that I had no idea how to audit things like who’s using up the most CPU cycles, who’s taking up all the network bandwidth and so on. My rough equivalent was ps -eo c,args | tail -n +2 | sort, wrapped up in a convenience function. I was looking at NetBSD stuff and stumbled across a few tools under the configuration, administering, and tuning section. These tools surely have Linux counterparts, if they aren’t mostly the same.

Of particular interest to me were:

• iostat(1)
• lastcomm(1)
• sa(8)

sysstat appears to be a collection of tools in Linux.

I still haven’t found a good network auditing tool. I suppose wireshark is a viable choice, coming as it does with command line tools. I’ll have to play around— I don’t really want low-level packet dumps, and a list of open connections from netstat(8) isn’t quite what I want, either. I’ll have to play around. And of course if you have any suggestions, feel free to drop ‘em in the comments.