Stop Deploying Blind! Using Observability and Argo Rollouts to Light the Way
summary
Argo Rollouts automates progressive delivery and start with 10% traffic canaries, querying Prometheus or a custom analysis job to enforce concrete success criteria such as 95% requests return 2xx within about 15–30 minutes. Requirements: application can run two instances, trusted RED metrics (rate errors duration) or Linkerd-provided metrics, and full automation so Argo makes rollbacks automatically. UI promotion via an Argo CD extension exists; statefulsets are currently unsupported.
AI-written summary · about
chapters (19)
- 0:00Argo Rollouts introduction
- 1:06Deployments gone wrong
- 2:12Automated rollbacks goal
- 3:37Argo Rollouts standalone
- 5:00Progressive delivery with Argo Rollouts
- 6:02Adopting Argo Rollouts requirements
- 7:26Success metrics in Argo Rollouts
- 9:08Choosing metrics for apps
- 10:55Missing metrics solutions
- 12:00Adoption mistakes
- 13:24Automating deployments steps
- 14:47Observability resources
- 15:30CNCF accessibility support
- 17:12Q&A: StatefulSets support
- 18:20Q&A: Metrics for canary
- 19:44Q&A: Argo CD rollouts
- 20:47Q&A: Fast deployments handling
- 21:51Q&A: Multiple deployments
- 22:53Q&A: Rollback behavior
related talks
30mProgressive Delivery Made Easy With Argo Rollouts
35mProgressive Delivery with Argo Rollouts
26mProgressive Delivery Made Easy with Argo Rollouts
18mTailored Deployment Strategies: Argo Rollouts for Diverse Applications
40mProgressive Delivery with Kubernetes and Argo Rollouts
24mArgo Rollouts from the Trenches – Next Level Progressive Delivery Best Practices
22mAnalysis in Action: Real-World Metrics Driven Argo Roll...
25mJuggling Argo Rollouts for Progressive Delivery Across Multiple-Services