reboot an instance that is triggered off SNS via lambda
This was inspired by a bug in a developer's code. The cycle was app crashes -> manually run a script to ssh + reboot
This was tiresome as there was no idication of when app was about to crash and I couldn't have any agents on the box for monitoring.
Devops to the rescue! I noticed this pattern while looking at EC2 graphs : as network packets drop to X, app crashes shortly after. So created a SNS topic that alerts on lower than X network packets. This lambda runs when the SNS topic triggers.
As this lambda depends on Crypto libraries, it's important to compile and package the lambda on Amazonlinux. I deploy mine from a docker container.