RRDtool is a specialized database to log, graph and analyze data collected over time, typicaly scalar values of various kinds. The RRDTool web site has a gallery with many examples.
Archiveopteryx provides various kinds of monitoring data. This page explains how to integrate Archiveopteryx and RRDtool.
The use-statistics configuration variable has to be enabled. You may want to change statistics-address or statistics-port.
Once Archiveopteryx provides monitoring data, you need to set up a cron job to fetch that data and store it using rrdtool update. Archiveopteryx contains a suitable script in lib/contrib/rrdglue. You may want to modify it or use a script of your own.
At the very least, you have to edit it to store the RRD files in a suitable directory. If you store other RRD files in e.g. /usr/local/rrd, we suggest setting RRDDIR in rrdglue to e.g. /usr/local/rrd/aox.
We suggest running rrdglue using cron every 5, 10 or 15 minutes. rrdglue fetches all data since its previous invocation and stores them in RRD files, so the only difference is how old the latest graphed data can be.
The RRD files created by rrdglue may not suit your policy. Here is the rrd create command:
rrdtool create login-failures.rrd -s 1 DS:login_failures:COUNTER:30:0:2147483647 \ RRA:AVERAGE:0.5:120:2880 \ RRA:AVERAGE:0.5:3600:672 \ RRA:AVERAGE:0.5:86400:720
This stores data at second resolution. If you prefer to have smaller RRD files, you could change -s 1 to e.g. -s 10 or -s 60.
The three RRA lines specify that RRDtool should:
We'll add some example graphs when we have a server that's been running Archiveopteryx, rrdglue and RRDtool for a few months.
What kind of graphs? I'm not sure.
In case of questions, please write to email@example.com.
Last modified: 2010-11-19