I've used both TC (TeamCity) and Hudson (Jenkins if you will) with RoundhousE.
The process is actually quite simple. You need to include RH in your build output or have it installed on the box (install being a loose term here, just meaning on the PATH somewhere available).
Then I would recommend you split your build from your deploy. I can go into many details but it amounts to having a CI build that does nothing to affect your environments (kind of the same concept as command query separation, you don't mix a build and a deploy). But you can have an automated deploy for an environment, say DEV, when the CI build is successful. And you can define your other environments, but have them with a manual trigger (so a QA person could say trigger a deploy to the TEST environment).
For the particulars of how with RoundhousE:
- As previously mentioned, rh.exe is reachable somewhere.
- Your scripts are checked in and getting built with your code.
- Depending on whether this is greenfield DB or maintenance, you would do a drop/run or a restore run. Also depending on your ability to affect an environment.
- If you have credentials necessary to a database, the optimal solution is to have someone set those in environment variables and then just put those into the connection string. i.e. if your database is named Northwind and you are hitting the DEV environment, you might set an environment variable named %Northwind_DEV_PWD%. Or you would go a step further and put the whole connection string into an environment variable. Whatever makes the most sense between you and your development team (DBAs included).
Here is an example (from a recent project in Jenkins):
Please feel free to hit the group up with any followup questions.
Note: So sorry about not approving this sooner. It got lost in my inbox.