Issue - Between the hour of (08:05 — Jan 12) and (19:40 — Jan 12) , approximately ~21 Million clicks were dropped over the whole day, i.e. either 500 was returned or the requests timed out (460 status code).
Monitoring - We are monitoring for any repeat signs. This was fixed by adding a local cache to each instance so our global cache is not overloaded whenever new servers get added in large numbers or lots of new offers get added.