Fix Production Rollouts on the Fly With Agentic AIOps
summary
Demonstration shows an Argo Rollouts + Agentic AI pipeline that runs a Kubernetes agent (Quarkus example) which analyzes logs, Prometheus metrics and pod state via A2A tools, decides pass/fail with a 50% confidence threshold, and can create PRs and GitHub issues (agent uses a GitHub token and assigns coding assistants like Google Jules). A presented failure triggers a rollback after limited impact; earlier runs generate a PR that fixes an index-out-of-range panic and adds a test, though the live demo suffers network/agent 500 errors.
AI-written summary · about
chapters (11)
- 0:00Agentic AIOps for rollouts
- 2:08Progressive delivery with Argo Rollouts
- 3:56Canary deployments with Argo Rollouts
- 5:23AI log analysis for anomalies
- 6:27Agent architecture for Kubernetes
- 7:30AI plugin integration with Argo Rollouts
- 9:18AI-powered rollback and fixes
- 11:01Demo: self-healing rollouts
- 16:26AI-generated GitHub issues
- 18:11AI benefits in rollout recovery
- 21:21Key takeaways: agentic AIOps
related talks
33mSelf-Healing Rollouts: Automating Production Fixes With Agentic AI
31mAIOps: (near) Zero-Touch Production Rollout Fixes
26mBeyond Argo Rollouts: Boosting Developer Experience With Intelligen...
27mUps, let's revert! 🔥 Automating Kubernetes Rollouts with Argo a...
27mIntegrating Agentic AI Into Internal Developer Platforms To Minimize Errors and Downtime
30mProgressive Delivery Made Easy With Argo Rollouts
18mTailored Deployment Strategies: Argo Rollouts for Diverse Applications
40mProgressive Delivery with Kubernetes and Argo Rollouts