We stopped using Cube seriously over 3 years ago. It continued to be used to capture business metrics (including payments, user signups etc) but was never seriously used for higher frequency system metrics. We have a lot of operational investment in MySQL (and Redis and Hadoop) and chose not to invest in MongoDB as well.
I tried a couple of times to keep it going as a personal research project and also to merge the infochimps fork back into the main branch, but I was not able to prioritize the effort given other commitments.
Cube remains a bunch of interesting ideas, but its node dependencies are outdated and it's never been restructured to take advantage of newer mongo features either.
Hope that gives some context.
Tom.