Web won't start up until each of the workers has

[jonnor]

web crashes because binding the queues don't work, as they queues don't exist yet.

This affects first-run, say when using Heroku button to deploy, or just using git push. Need a well-documented way to deal with it. Probably using msgflo-setup as part of a postdeploy step?

Workaround on Heroku is to manually start each of the workers by enabling them under Resources in dashboard, then restarting the service.