blatherings on software…
We built an event driven system in my team over 2 years ago that sometime back ran into database transaction deadlock problems which puzzled me enough to want to take a deeper look into what happened. This mini-post is a bit of documentation of that investigation.
Recently a planned AWS Aurora database failover/master reboot caused database connection failures in one of our services which caused me to delve deeper into how MySql driver manages connections under the hood.
In this somewhat of a long post, I will go over some of the lessons I and my team learned while building an event driven architecture at work, why did we build it, recommendations and watch-outs for anyone else thinking of doing the same. This is more of a field report from a real life environment so there is a generous helping of pragmatism around architecture and careful and deliberate avoidance of dogma and purism.
Artineering 