A couple more ideas as I'm brainstorming about this:
This would be helped by a single process/daemon that starts up both the web service and the collector.
The collector should not end if it cannot connect after so many retries, but keep going.
Need to automatically run database migrations on startup.
Need to test how Pex (or similar tool) deals with the Pyramid static assets and Alembic migrations (i.e. how do paths work if everything has been zipped up).
Need a simpler way to deploy the Mesh Info application. A few ideas (there might be several parts to this):