Continuous Delivery | Hacker News Reader

Continuous Delivery

rdegges.com

11 pointsrdegges8 comments

Nice post, Randall.

But I had trouble buying the premise of:

    30 minutes later I had configured my CD software, Jenkins [..]

Nothing in Jenkins can be configured in just a half hour. ;)

Thanks! I wrote this post a long time ago, but this is how I've always done it: http://www.rdegges.com/simple-continuous-integration-deploym...

Usually takes me 20 minutes ish.

Brilliant write-up on the benefits of continuous delivery.

Especially the part on delivering value as fast as possible (& drastically changing Cost of Delay) is a very powerful concept that generates enormous economic value (something that engineers often don't immediately think about in comparison to productivity).

The new bamboo 5 beta takes steps towards managing and monitoring deployments. I just thought I'd give it a mention.

More info from: http://blogs.atlassian.com/2013/05/bamboo-beta-program/

Good post! Though from experience it's never worked as smoothly as described when in a team setting. "Oh, you forgot to check features 1-40 in all major browsers?" Rollback button : engage

I use rollback liberally in team settings -- I also do a lot of my work on Heroku, so if anything breaks we do a `heroku rollback` immediately.

Usually though -- we're very careful about deploying stuff live. We autodeploy our develop branch to a testing Heroku project, and if that's good, we merge to master.

So if I'm working on a feature branch I'm assuming that the build and deploy would only happen after I merge into master and push.

Is that how continuous deployment is usually set up?

Depends on what you need. I tend to have CD setup for both my develop and master branches, and everything else gets worked on / tested locally only (until those changes are merged).

Reply on news.ycombinator.com