Automating and Testing Production Ready Kubernetes Clusters in the Public Cloud
summary
Built cloud-agnostic Kubernetes clusters deployed to AWS and Google using Chef for bootstrapping, an overlay network with Calico, and farm-style provisioning via Scaler. Auto-scaling, HAProxy-based ingress, Vault-backed secrets, and CI server-spec + canary clusters validate changes. Production claims: ran 100% of election-night traffic, ~170 successful deployments that night, 1,500 deployments during hurricanes, and reduced full-site deployment time from ~2 hours to 25 minutes. Ongoing pain points: node drain/Calico cleanup, cloud PV parity, and rapid Kubernetes version churn.
AI-written summary · about
chapters (20)
- 0:00Intro
- 0:39Who is Gannett
- 1:18What is a Platform as a Service
- 4:48Requirements
- 6:48Sprint Tasks
- 7:34Kelsie Hightower Effect
- 8:12Network Requirements
- 10:10Scaler
- 11:17Architecture
- 12:59Paz Labs
- 14:54Kubernetes Architecture
- 19:31Canary Clusters
- 20:03Runbooks
- 20:42Observability Engineering
- 22:14Challenges
- 24:44Contract Limits
- 25:31Persistent Volumes
- 27:44Election Success
- 29:00Deployment Success
- 30:15Whats Next
related talks
35mKubernetes journey
32mRunning Kubernetes in a Manufacturing Line – What Could Possibly...
17mOur Journey to Kubernetes at Scale
41mKubernetes — A Year in Production
40mFrom dev to prod. One Year of Using Kubernetes in Production
36mOptimizing Resource Usage in Kubernetes
29mHow to make Kubernetes rhyme with prod readiness
45mLearning How to Pronounce Kubernetes to Production in 3 Months!