Beyond Argo Rollouts: Boosting Developer Experience With Intelligen...
summary
A working Argo Rollouts plugin calls an agentic AI pipeline that analyzes canary pods by fetching logs and metrics, decides health, and triggers automatic rollbacks plus GitHub PRs or issues; demo reproduces a null pointer crash and a subtle memory leak and both cases are detected and handled. The system runs multiple parallel agents (diagnostic, analysis, scoring, remediation), uses a local Qwen model on OpenShift, is implemented in Java with Quarkus for speed, and is published in Argo Project Labs. Argo Rollouts metric-AI plugin Kubernetes Agent (Java, Quarkus)
AI-written summary · about
chapters (10)
- 0:00Introduction to AIOps
- 2:21Metrics-based rollouts
- 4:19Integrating AI into rollouts
- 5:17AI agent configuration
- 8:02Extending with agents
- 10:06Demo: Canary analysis
- 12:38Handling deployment failure
- 15:09Automated remediation demo
- 18:05Agent system architecture
- 20:44Takeaways and Q&A
related talks
33mSelf-Healing Rollouts: Automating Production Fixes With Agentic AI
24mFix Production Rollouts on the Fly With Agentic AIOps
31mAIOps: (near) Zero-Touch Production Rollout Fixes
18mTailored Deployment Strategies: Argo Rollouts for Diverse Applications
22mArgo Rollouts Reinvented: Unleash the Power of Custom Plugins for Progressive Delivery
11mKeynote: Smarter Progressive Delivery at Intuit
40mProgressive Delivery with Kubernetes and Argo Rollouts
22mReal Time Argo Rollouts Analytics Powered by Notification Engine