[TriLUG] "Light" monitoring

[Brian McCullough]

I have been working on how to ask this question, but I guess I'll go
ahead anyway.


I have a system that seems to be becoming more fragile, and I would like
to monitor it, and send myself e-mail messages when it needs attention.


I know about Nagios, but it seems to add more load to the target system
than I would like, with it's polling several times per second, depending
on what services it is monitoring.

I also wondered about something like MRTG, and read the graphs remotely.
I don't know whether I can set up alarms that way, though.

I can also do something like "ping -c 3" from an outside site.


Primarily, to begin with, I am interested in load levels and web server
"aliveness" over time, with the ability to alarm ( via e-mail and
possibly SMS ) when some threshold ( say high load over three minutes )
is passed.

I have had the web server apparently just go away two or three times
this month, and have seen some very high "top" values at more than one
point. 

Side question -- since Top and friends only show one value, what is it saying about a multi-cpu system?



Any suggestions, or roll my own?  I'm sure that that is not the answer;
there have to be multiple tools to help me with this problem.


Thanks,
Brian