Experience has taught me that if you want to get something done, you need to stay off the bleeding edge, or at the very least only drag yourself to the bleeding edge at controlled increments.

The only thing that should be changing often is YOUR code. Everything else should be as stable as possible.

What I find works best is playing leap frog: * your code is unstable / upgrade nothing * your code is stable, stop changing it and upgrade things one at a time

test suites / regressions are a must.

I suffer from the same trouble on occasions. I primarily deal with the Microsoft ecosystem which is volatile and fragmented at the best of times.

I would kill to do something in good old java/mysql.

A lot of the new ideas and principles are a mess. Don't get too absorbed - just use what lets you get the job done quickly and efficiently. a lot of it is just noise.

I tend to find that a complete break from everything for a day helps. That and cider :)

Try postgreSQL next time, at least it's ACID compliant ;)

Try using less things... How about just using Node and Mongo, and building the rest yourself?

Node being in its infancy is a pretty good reason in itself to not tie yourself to the frameworks which are likewise infantile.

I've got several production back-ends and a full-stack e-commerce site running with Node and MySQL only.

A month ago I finished a project for university, involving Node.js - nothing more. At the beginning, I've started with Node.js, mongoDB, socket.io. express and connect. Half way through the project, I realized I would never be able to deliver in time. So I started to cut out technologies... first mongoDB, socket.io, express, and in the end, even connect. Even dealing with Node alone, I had a hard time finishing my project... but I did. I still got quite a good grade because in contrast to my fellow students, I was able to deliver.