[toast0]

Deployment for Erlang needs to be a lot more flexible than other systems, because you have the opportunity to hotload code as a core feature [1], having your deployment solution invoke code loads is totally scriptable, but seems more likely to want human eyes watching than a traditional push, stop, start cycle.

OTP does provide the application construct which is supposed to provide some amount of configuration and start/stop/update support; although my team doesn't use the update support, we just push code and hotload it or stop and start beam depending on the changes.

[1] of course, you can hotload code in most languages if you try hard enough, I've done it in C and almost did it in perl; as long as you can load code and get a function pointer at runtime, you should be able to make it work.