Panel: From Signals To Safety Nets: Self-Healing Progressive...
summary
Shows a working self‑healing pipeline that uses Argo Rollouts, Linkerd, and Prometheus to convert time‑based canaries into signal‑driven rollouts. A Linkerd sidecar injects metrics, Prometheus answers PromQL queries from an AnalysisTemplate, and Argo advances or auto‑rolls back when success rate falls below 95% (two failures, one‑minute checks). Demo deploys v2 at 30% traffic, spikes errors, and Argo reverts to v1 returning to 100% success—templates and YAML snippets for the closed‑loop are provided.
AI-written summary · about
chapters (18)
- 0:00Panel introduction
- 1:47Modern deployment challenges
- 2:30Faster deployments vs outages
- 3:56CrowdStrike outage & canaries
- 5:20Safer deployment methods
- 6:04Signals in distributed systems
- 7:50GitOps & Argo Rollouts
- 8:55Linkerd observability
- 9:58Templates & automated rollbacks
- 11:48Service mesh signals
- 13:32Self-healing loop
- 15:44AI ops beyond rollbacks
- 17:52Key signals to monitor
- 19:58Self-healing demo
- 23:34Summary of self-healing
- 24:36Merge Forward initiative
- 25:21Accessibility working group
- 27:08Wrap-up & resources
related talks
30mProgressive Delivery Made Easy With Argo Rollouts
23mStop Deploying Blind! Using Observability and Argo Rollouts to Light the Way
40mProgressive Delivery with Kubernetes and Argo Rollouts
27mUps, let's revert! 🔥 Automating Kubernetes Rollouts with Argo a...
35mProgressive Delivery with Argo Rollouts
33mSelf-Healing Rollouts: Automating Production Fixes With Agentic AI
25mPrometheus Data analysis and Event Notifications for Progressive Delivery
25mObservability In ArgoCD/Rollouts Using Streaming ML For Reducing MTTR