It's been a few months, but I finally figured out how I've been misusing node and the mongodb JS plugin.blog comments powered by Disqus
Essentially it's my PHP background to blame, where every new request to the server is exactly that - a new request - and everything has to be reinitialized anew.
Instead, node is a true server - it runs, maintains state, and waits for new input, be it a page request or ... something else. As such, it is not necessary to reinitialize the database, the connections, and most importantly the data collection that mongo maintains. Instead you run new finds/saves against a collection that's already loaded - much faster, seems to me.
Initially the mongo allowed these repeated loads and I'm sure the performance suffered; but at some point, the plugin author decided to stop the madness and simply started throwing errors when repeated (and redundant) connections to an already open DB were requested. I didn't really have the headspace to grok this until now, and I freely admit it's much more elegant to well, behave as node expects. Hopefully that means more updates, on the bloggy at least. DCSB will still have to wait. :)