node.js daemon to classify log file events and upload them to Google Analytics. This means that by running this daemon you will be able to analyse some events that show up in your syslog using Google Analytics, alongside your web site access statistics.
Log output recognised by this daemon is sent to Google Analytics by means of the Measurement Protocol. This means you need to have Universal Analytics enabled for your property.
Out of the box this daemon supports syslog events from OpenSSH and scanlogd, as well as standard "combined" format HTTP logs and kippo honeypot log files. Adding more log formats is a matter of simply adding more regular expressions. Additionally, CPU and network usage is translated to events that are also logged.
Use npm to install the package; the package should be installed globally so as to be usable from the command line:
# npm install -g analyticsd
If you get an error about npm not being a valid command, install node.js by following this guide: https://github.com/joyent/node/wiki/installing-node.js-via-package-manager
You should run analyticsd as root; it will drop privileges automatically - by default to the user 'daemon' and the group 'adm' which should be able to read log files. To run the programme, use a command like this:
# analyticsd --tid UA-XXXXX-Y
The --tid parameter specifies the Google Analytics property to send data to. Have a look at the Google Analytics Admin panel to get this ID if you've misplaced yours.
analyticsd will not fork to the background by default. Use the --daemon flag for that.
# analyticsd --tid UA-XXXXX-Y --daemon
To launch the daemon at boot time, add a line like the previous to your /etc/rc.local - before any exit; instructions, if there are any.
Further options may be documented in the daemon's man page:
$ man analyticsd
NOTE: You should use a separate ID from your 'normal' Google Analytics ID, at least while testing. If you later decide to use a single ID for both the events processed by this daemon and your website, you should set up different views with different event types - the SSH events alone might drown out your actual web site accesses otherwise.