I try to log most errors and exceptions, and send them to a log-management tool. I use Loggly (they make it easy to run reports or get automated notifications b

I try to log most errors and exceptions, and send them to a log-management tool. I use Loggly (they make it easy to run reports or get automated notifications based on error-type) but have heard good things about Splunk as well. To monitor the actual servers, I tend to use CloudWatch (since we are on AWS), but New Relic and DataDog are popular options. I also use Pingdom to hit all our servers routinely (once a minute or so), and let us know if something goes down or is taking too long to respond. This is a bit of a redundant check considering the other metrics we have, but I like having some redundancy there. Hope that helps.