How a Unicorn Migrated From an In-House Solution to Helm
summary
400+ services move from an in-house Python/CDK generator to Helm with Argo CD handling deployments; a single reusable base chart called chartmander lives as a dependency in each service repo while service charts override defaults (metrics off, replica counts, etc). Requirements: self-serve, composable, extensible. Python pipeline emits a values.yaml with cloud items (IAM roles, ECR URLs) so Helm and existing infra stay synchronized; base chart later splits into security, monitoring, networking modules.
AI-written summary · about
chapters (22)
- 0:00Intro
- 1:07Where I work
- 2:00Compute environment
- 3:05Migration to Kubernetes
- 4:40Usability
- 5:20Agenda
- 5:45About Mey
- 6:43The Helm Survey
- 8:47The Helm Requirements
- 11:10The Helm Structure
- 11:51Composable
- 12:53Extensible
- 14:37Evolution of Helmchart
- 15:02Helm features
- 18:21Common Helm solutions
- 21:10Single Helm chart
- 23:38Umbrella chart
- 28:10Multiple umbrella charts
- 29:22Inspiration from Jenkins
- 31:17Base Chart Solution
- 35:30Base Chart Benefits
- 36:30Existing Infrastructures Code
related talks
43mHow a Unicorn Migrated From an In-House Solution to Helm
32mHelm your way with Kubernetes
26mHelm your way with Kubernetes
05mLightning Talk: New Helm Chart Installation Experience with Visual and Repeatable...
32mManaging Helm Deployments with Gitops at CERN
38mHelm: The Charts and the Curious
43mAt The Helm of Kubernetes: Repeatable Infrastructure Creation for Mere Mortals
27mAre you Helming it the right way? Helm challenges and its solutions