We are using heroku_san with mongoid_rails_migrations (fixed version at git://github.com/quantopian/mongoid_rails_migrations) for a Ruby on Rails application built on top of ruby 2.0.0, rails 3.2.13, and MongoDB.

When we run "rake staging deploy" or "rake production deploy", it does the entire deploy process four times.

When we do the deploy with "-t", it claims that all four deployments are happening within a single invocation of the deploy rule.

We thought perhaps it was running the deployment separately on each dyno, but our staging environment has only 2: 1 web dyno and 1 worker. On the other hand, our production environment has 3 web dynos and 1 worker. I don't know if any of this is relevant.

Oddly, it doesn't always do this. For example, it doesn't seem to do it in automated builds run by our Jenkins server, even though those builds are using the same "rake staging deploy" command we type on the command line.

Has anyone else seen this before? Any pointers for how to troubleshoot it further?