This isn't a criticism: it's inevitable in a young platform, and I'd happily help develop the libraries. But it didn't seem a feasible option for a project that (a) had deadlines and (b) needed features beyond the (impressive) node.js core.
Node devs tend to take a different approach to Rails/Django in this regard, there's no single ORM with lots of database backends (ala ActiveRecord). This is probably related to the fact that Node devs tend to use tools like Redis, Mongo and Couch rather than traditional relational databases like MySQL and Postgres which all have similar SQL interfaces. There are some really nice libraries and modeling tools available but they're not database agnostic SQL ORMs, for example recently I've been using Mongoose: http://mongoosejs.com/
I've come from the Django world so I empathise with the "How do I do [insert Django/Rails feature here]?" mindset, but a lot of the time the Django/Rails just doesn't work well Node. Node.js is a single-threaded, event-based, non-blocking server; it really is a totally different paradigm so you have to be prepared to rethink the modus operandi for developing web applications.
Also, I didn't find a lot of help online with how to do this modus operandi rethink. (Again, no whining implied).
0.4.0 brings quite a few API changes from 0.2.6, so if you haven't been keeping up, you should really scour the API docs. You can also check out the wiki for a guide on migrating. https://github.com/ry/node/wiki/Migrating-from-v0.2-to-v0.3
It seems like that under this assumptions node.js is stable enough, but what are you doing with it? Maybe it's not going to be "mission critical", but good enough to run production web applications.