One Chart to Rule Them All: Continuous Deployment with Helm at Ticketmaster
summary
Ticketmaster builds and maintains a single, 251-line Helm webservice chart that deploys apps across a hybrid fleet (about 15 clusters) using GitLab CI, with per-namespace Tiller, namespace-creator enforced AD access and quotas. The chart injects sidecars for Jaeger and Fluentd, adds Prometheus scraping annotations, emits ALB or nginx ingress annotations for AWS/on‑prem, creates CPU-based HPA (min/max replicas) and PodDisruptionBudgets; this central chart reduces duplicated charts and “works well so far” for multiple product teams.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 0:50Background
- 1:37Agenda
- 2:19Kubernetes
- 4:28Helm
- 5:51Hybrid Cloud
- 8:03Web Service Chart
- 12:34Basic Structure
- 12:59Values File
- 15:29Deployments
- 17:39Sidecars
- 20:04Metrics
- 20:46Conditionals
- 24:34Config Maps
- 25:14HPA
- 26:54Pod disruption budget
- 30:09Demo
related talks
32mManaging Helm Deployments with Gitops at CERN
27mMore Charts, More Problems Lets Talk Bringing Sanity
27mAre you Helming it the right way? Helm challenges and its solutions
26mLessons Learned from Maintaining Our Own Chart Repository
26mHelm your way with Kubernetes
35mOptimizing Kubernetes Deployments with Helm
28mGet Your Helm Charts Out There!
36mDelve into Helm: Advanced DevOps [I]