Distributed Automation using Kubernetes, Docker, Traefik and Helm
summary
Kubernetes, Docker, Traefik and Helm power a Selenium Grid rebuild that actually creates hubs on demand, spins up hundreds of browser pods (examples show 50–500) and destroys them after runs, cutting provision time vs EC2; Expedia runs ~90,000–200,000 UI tests a day on mixed solutions and proves containerized grids work. Key specifics: Traefik routes hostnames to per-hub services, use warm IP target to avoid IP exhaustion, and favor one browser per pod for predictable autoscaling and QoS.
AI-written summary · about
chapters (20)
- 0:00Intro
- 1:00Out current solution..
- 2:40Pure EC2 vs Containers
- 3:33Micro-services
- 6:12Multiple AWS A/C
- 7:19Cost
- 10:53Big Picture
- 14:49End to End flow
- 17:34Creating - Autoscaling - Terminating Selenium Grid
- 20:39Horizontal Pod Autoscaler
- 24:11Warm Worker Node Pool
- 25:07Https based Hub
- 25:38Resource QoS
- 27:19Reserve Compute Resource
- 29:50Hub and Node in different subnet
- 30:29OKHttp to ApacheHttp
- 31:07To Selenium Committers
- 32:43Headless vs Regular
- 33:23Prevent IP Address bleeding - EKS
- 34:49Job type Deployment
related talks
45mSelenium in Kubernetes: Lessons Learned
25mAutomating Chaos Attacks at Expedia
43mInfinite Space for Your Tests
47mBuilding Highly Scalable Appium Infra with Terraform &Kubernetes
35mOptimizing Kubernetes Deployments with Helm
36mEffective Test Automation in Kubernetes
46mDemystifying Kubernetes and Containers Orchestration for Developers
44mFrom single container to applications at scale