Too Long; Didn't Read
I’d previously <a href="https://medium.com/@bparli/writing-goavail-a-cloud-monitoring-and-fast-dns-failover-agent-7c59254a5c45" target="_blank">written about motivations</a> for writing a decentralized IP/<a href="https://hackernoon.com/tagged/tcp" target="_blank">TCP</a> endpoint monitoring and DNS failover agent which I called Goavail. The long and short of it is the project is a cluster of lightweight monitoring agents which can be deployed to remote and disparate locations in order to either perform IP or, more recently added, TCP monitoring of specific endpoints (such as an AWS EIP). This came in handy when we decided we should have some layered availability monitoring and failover in place should something go wrong with one of our ingress procxies, an AWS zone go down or become partitioned. The service always performed well in failure testing, but fortunately, we never needed it in an unplanned Production failure situation.