A Crawl, Walk, Run Story of CI/CD
summary
A practical CI/CD pipeline uses drone for CI and harness for CD, with canary deployments, Datadog log verifications, and intelligent rollbacks tied to Elasticsearch/metrics. The build system runs about 300 builds 250 deployments per month, records 67 failed deployments and achieves a mean time to recovery of about one minute. The team implements inter-component dependency ordering, feature flags, and automated rollbacks, and the result is higher deployment velocity and much lower manual recovery effort.
AI-written summary · about
chapters (12)
- 0:00CI/CD journey introduction
- 1:46Why start CI/CD in phases
- 3:11Choosing Drone CI
- 6:41Drone CI workflow
- 8:24Two years with Drone CI
- 9:50Need for CD tool
- 10:50Harness CD features
- 13:58Harness CD deployment flow
- 16:48CI/CD impact on deployments
- 18:11Deployment metrics and recovery
- 20:16Benefits of simple CI/CD
- 21:20Conclusion
related talks
23mCI/CD: Conscious Uncoupling
33mCI/CD for Autonomous Drive and other complex systems - What we have learned about...
17mCI/CD - Build/Deploy Automation
23mCI/CD at Scale: Best practices with AWS DevOps Services
01h 01mA team's transition to Continuous Delivery
24mCI/CD Best Practices to Deliver Software Faster
48mDemystifying CI/CD
34mDeploy, Release, CI/CD, Oh My! DevOps for the Rest of Us