> The server immediately started having problems, and every flaw in the message handling code suddenly became glaringly obvious. We quickly added another server and that seemed to hold off the worst of the problems (for the curious, we are running the server on Heroku, written in Java). I had to make changes to the servers on my phone at least once on the weekend, and we found a couple limitations in our database the hard way.
I would have wished for some more technical background about what actually broke and how you fixed it ("our service for X was written in Y, but that was bad because of Z" / "system slowed down at Y messages because we did Z" / ...).