From dev to prod. One Year of Using Kubernetes in Production
summary
One year running Kubernetes in production across five clusters aligned into a CI/CD pipeline yields concrete practices: use Azure DevOps for repos and enforced builds, deploy with Helm without Tiller to get templated releases and rollbacks, ship logs via EFK stack (Fluent Bit → Fluentd → Elasticsearch → Kibana), collect metrics with Prometheus + exporters, add tracing via OpenTracing/Jaeger (Zipkin-compatible), back up resources and PVs with Velero (formerly Ark) but handle cross-region snapshots manually, clone clusters for safe upgrades, enforce RBAC and image scanning (Anchore/Clair), and automate everything.
AI-written summary · about
chapters (8)
- 0:00Introduction and speaker intro
- 3:54Kubernetes basics and context
- 14:25Release management challenges
- 17:38Logging and monitoring setup
- 25:18Cluster backup and recovery
- 28:31Kubernetes upgrades and scaling
- 31:24Security in Kubernetes
- 37:05Key takeaways and summary
related talks
41mKubernetes — A Year in Production
45mLearning How to Pronounce Kubernetes to Production in 3 Months!
34mKubernetes from Dev to Prod at GoEuro
35mKubernetes journey
29mOrchestrating 10,000+ Kubernetes deployments with 1 pipeline
29mHow to make Kubernetes rhyme with prod readiness
37mRunning Kubernetes in a Manufacturing Line
20mKubernetes 101