Sylva, Taming Complexity With FluxCD Dependency Management
summary
Sylva is a Kubernetes-as-a-service for telco use cases that uses Cluster API and Flux to get predictable, fully declarative installs; teams report ~30 deployments and management clusters with ~100–150 customizations (tens or hundreds total resources). Flux dependsOn and ready expressions serialize reconcilations, custom per-resource timeouts detect CI failures, and a custom silva-watch CLI builds dependency graphs to pinpoint failing customizations. Upgrades use a temporary root dependency Helm job to force correct ordering; deletion uses a pre-delete job and wait-for-termination policy.
AI-written summary · about
chapters (9)
- 0:00Silva Kubernetes as a Service
- 1:25Silva’s architecture and FluxCD
- 3:33Parallel reconciliation challenges
- 7:24Chaos solved with Flux dependencies
- 9:34CI and deployment testing
- 13:28Infrastructure upgrades handling
- 17:06Deleting resources complexity
- 20:17Silva’s impact and future
- 22:26Q&A
related talks
41mSylva Project, What Is It, What We Achieved, and Where Are We Going?
07mFluxCD - Gitops For All Sizes
40mTreating Kubernetes as a Linux Distro: APT-Style Packaging with FluxCD
33mFlux Ecosystem Evolution
36mManaging Thousands of Clusters and Their Workloads with Flux
42mIntro: Flux
21mIntroduction to FluxCD
21mFlux AIO - a Distribution of Flux Made with Timoni and CUE Lang