not_IOmap reducer
gerryflapYeah that checks out. Now you’re ready for that massively parallel big data (or sandwich?) processing
JuanPeeceAnd then you fuck up your asynchronous timing, leaving a race condition where the sandwich has bread on the inside. Oh, and QA can only reproduce sometimes
With mapreduce you should generally have your code written in such a way that that cannot happen. Unless the reduce step is improperly programmed.