[mrtg] MRTG + Squid Implementation

[Warren Togami]

(please CC replies to warren at togami.com)

I'm working on building a low budget squid server for my high school.  It
has been running for over 2 months now with great results, but now I am
currently working on getting MRTG to graph the squid statistics to give us
some real time statistics on our traffic.

My project notes are available here.  The administrators suggested for me to
write up about the project for submission to this local high school
techonlogy contest, kinda like a science fair type of thing.
http://www.ultrapenguin.com

Unless I missed something, it appears to me that the documentation on the
subject appears to be somewhat out of date.  After I finish sorting out all
these problems, I'll write up some new documentation for the community,
along with new sample configuration files and additional notes reflecting
the newer versions of this software.

squid-2.3.stable1
MRTG 2.8.12

First I have to get these things working correctly.  After 8 hours of
skimming through the old documentation and mailing list archives, I've
managed to get it working.  Sorta.

Current issues - http://www.ultrapenguin.com/squidmrtg/
1) MRTG graphs seem to be working, except it flattens out at the max
variable value at completey idle times during the night.  Yes, I have read
the documentation saying that MRTG flattens at times when packets arn't
coming back.  I checked this running mrtg manually, and it isn't complaining
about the lack of server responses so I'm guessing that this isn't the
problem.  The /var/log/cron only displayed errors when the squid process was
shutdown during one of the 5 minute updates.
2) cacheSysPageFaults also flattened out at a high, non-zero number.  This
demonstration of MRTG showing squid data has zero page faults, so I'm
guessing that this isn't good.
http://www.ircache.net/Cache/cache-snmp/mrtg-demo/demo/
Is this a squid problem, or a MRTG problem?  I think I read somewhere else
that page faults are normal.... is this true?
3) This MRTG sample configuration
http://www.ircache.net/Cache/cache-snmp/mrtg-demo/mrtg-squid-snmp.cfg seems
to work for me with the following modifications.
    a) LoadMIBs: /opt/squid/etc/mib.txt
    b) MRTG doesn't understand Options[^]: perhour .... should it?
    c) Commented out ICP sections, because I don't use it and it returns
errors.

What can I add/implement to better represent the data (when it actually does
work) on the MRTG graphs?  I read some about KB, MB and other unit
representations in some documentation, but I'm not sure how and where to
implement these effectively.

I was also wondering if there has been any work to further customize MRTG to
better represent squid specific data.  It says "Bytes per Second" along the
y-axis of all graphs.  Is it possible to make graph specific labels, say for
squid memory, CPU usage, etc.  If so I could make this my next project,
perhaps a standardized squid + MRTG package to make the job easier for
future admins.

Thanks,
Warren Togami
warren at togami.com



--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:mrtg-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/mrtg