[lug] Server health monitoring

[Nate Duehr]

Orion Poplawski wrote:

 > Does anyone have any recommendations for packages to monitor server
 > functions/health like process status, disk space, etc?
 >
 > Thanks!

Using bb and nagios both here, they're fairly different beasts.

Things I like about BB:
- LOTS of third-party plugins/scripts written for it.
- Setup is easy -- run a script, copy a tar file, extract, start 
program.  Add host to bb-hosts file.  90% of what you need to monitor 
pops up.

Things I like about nagios:
- Greater flexibility across the board.
- Better web interface, by far.

Things I don't like about BB:
- Purple status.  When it loses connectivity to what it is trying to 
monitor it pages the CRAP out of you.
- No way to schedule system downtime easily.  It has a setting, but you 
have to manually edit a configuration file.  And no way to automatically 
restart monitoring later.  Could be scripted, but nagios handles this 
better.

Things I don't like about nagios:
- Little too much emphasis on silly graphical interfaces of all sorts... 
tons of options for the display (fine if you're trying to impress 
customers in a NOC, but useless for the average admin).  VRML view?! 
Who needs THAT?!
- Not as many available plugins... scripting necessary.   Younger in its 
development cycle.

Not sure if it's a plus or minus:
- Object oriented configuration and/or XML configuration for nagios is 
becoming popular.  Seems like this will be neat for large corps with 
XML-based tools that need to feed nagios, but the average joe trying to 
monitor ten servers will find it too complex.

There's probably other things I've forgotten....

Some other useful network/system monitoring tools:

mon
fping
nanog-traceroute
net-snmp
mrtg
ttcp

probably others...

-- 
Nate Duehr, nate at natetech.com

Quando Omni Flunkus Moritati
"When all else fails, play dead."