In a similar vein, if you need to extract metrics from log files you don't control, collectd has a plugin for that. From collectd you can send the metrics to graphite, amqp, mongo, redis, riemann, etc. 
If you don't mind, I'm curious about a statement at the bottom of the page:
"This article was motivated by many success and failures experienced with production systems at Heroku."
Heroku is, as far as I know, more expensive than EC2. Maybe even double? With this being the case, at what point does it make sense to just run your own EC2 and set things up how you like instead of fighting against the platform?
I've only used Heroku moderately, but I've read many instance of folks spending lots of time trying to get Heroku to work how they want. It seems to me, they could've possibly spun up an EC2 instance and gotten things configured more quickly and easily.
So, to pose my question in another way: if it's possibly faster, and more cost-effective to run an EC2 instance, why do people continue to appear to be fighting against Heroku?