Panel: Deployment
summary
Uses distillery and edeliver for releases, sometimes building artifacts inside Docker to target the right OS; CI runs full integration tests by launching the app and calling Ansible deploys into Docker review apps. Deployment is mostly continuous — teams deploy every day or push tagged releases with versions from git describe; one-person setups use a single Linode with no clustering, larger shops use GitLab runners, Kubernetes and Heroku. CI runs on spot instances at ~10% of on‑demand cost. Hot code upgrades exist but are rare and used more for experimentation than for critical uptime.
AI-written summary · about
chapters (13)
- 0:00Intro
- 0:50Deploying in the cloud
- 7:30Testing
- 9:00GitLab
- 12:20Review App
- 13:40Microservices
- 15:20Identity Microservices
- 18:57Use Cases
- 22:25Versions
- 24:20HTTP
- 26:25Elixir to Heroku
- 28:12New experimental feature
- 29:37Other options
related talks
52mFrom 0 to Continuous Deployments with Kubernetes, Helm and GitLab
48mHow to build solid CI-CD pipeline
58mDeveloper Empowerment – Production Deployment DIY
24mContinuous Deployment at Scale
29mRobust delivery with CI/CD
30mRelease Early and Release Often: Reducing deployment friction
46mBlue-green Deployments, A/B Testing, and Canary Releases on Kubernetes
40mDeploying at scale with a single click