This is a continuation of the hacking to get Jepsen work with InfluxDB. I’m reviewing both technologies at the same time and logging it on my blog in the hope that readers will learn from it, that both Kyle Kingsbury and InfluxData will benefit from my feedback, if not other than one of the many feedbacks that they have produced products that are great - and lastly that I’ll manage to organize my experimentation around a memorable narrative.
Before the holidays I promised myself and my wife that I won’t code during the holidays (starting tomorrow). So I won’t. However, I’m setting up an environment for January on my Mac so they will wait for me when I come back for my next train #CommuteProject: a Jepsen test for InfluxDB!
Testing parallel algorithms or distributed systems is notoriously hard. If you deal with either of them - and there is higher and higher chance that you are already, or will in the future - then you should look at formal methods to have confidence in your design! The world of formal methods is evolving fast, and already becoming useful for cutting edge tech companies dealing with complex problems.