Not all microservices are bug free. Actually, most of them have some hidden bugs. Some of those bugs may even cause your microservice to hang, or to return unexpected responses. Will you wait until the next production crisis just to find out about it?

We, at eBay, took the proactive approach. Meet Watchtower, a monitoring/alerting system that periodically checks each host in the pool for expected responses as well as latency.

In this session, I'd like to describe Watchtower's high level architecture, its pipeline, and its visualization. In a nutshell, it saved our asses (more than once!)