Decurb - Scalable Deployment with Docker and Kubernetes

General

Names: Brian Seo, Lucas Kim

Project 5 Video Demo Link:

https://youtu.be/z5wcuFpePEw

The video is not an updated version that includes the AI chatbot (used open ai) for interactive response, general Q&A based on the database, and other questions. Due to safety issues, attempt to modify or harm database through AI chatbot is prevented.

Instruction of Deployment:

1. Set up a Kubernetes cluster on AWS using kOps with 1 control-plane and 3 worker nodes.
2. Deploy a single MySQL server to handle both master and slave configurations.
3. Monolithic version: Use kubectl apply -f decurb.yaml and kubectl apply -f ingress.yaml to deploy the decurb application.
4. Access the decurb site using the AWS ELB URL.
5. For multi-service architecture, stop the original deployment using kubectl scale, then apply decurb-multi.yaml and ingress-multi.yaml.

Collaborations and Work Distribution:

• Brian Seo: Kubernetes ingress configuration, JWT-based session architecture, MySQL read-write splitting, cart and order processing, Docker image publishing, cluster debugging, full-stack integration, and final video demo.
• Lucas Kim: Docker setup, static asset routing, test deployment pipeline, pod logging validation.

---

Task 1: Run the decurb application in a Docker container

• Followed Murphy Movies "Docker" branch to build and verify decurb as a Docker image.
• Created two images:
  • decurb-login: handles user authentication and session
  • decurb-movies: handles browsing, search, cart, checkout
• Docker images pushed to Docker Hub:
  • bseo97/decurb-login:v3
  • bseo97/decurb-movies:v3

---

Task 2: Set up a Kubernetes (K8s) cluster on AWS

• Created Kubernetes cluster on AWS using kOps.
• Cluster consisted of 1 control-plane and 3 worker nodes.
• Verified cluster status using kops get all and kubectl get nodes.

---

Task 3: Deploy decurb to the K8s cluster

• Wrote decurb.yaml and ingress.yaml to deploy decurb application.
• Moved MySQL database into the cluster using StatefulSet (MySQL master + slave).
• Verified:
  • Functional login
  • Movie browsing and search
  • Cart and order placement
• Confirmed sales data was inserted into MySQL by querying the database from within a pod.

---

Task 4: Revise decurb using a multi-service architecture

• Split the decurb app into two services:
  • /api/login, /api/logout, /api/session-check → decurb-login
  • All other /api/... and static files → decurb-movies
• Used Maven multi-profile system to compile separate .war files.
• Wrote decurb-multi.yaml and ingress-multi.yaml:
  • Set Ingress rules to route requests to the appropriate service
  • Configured session stickiness using cookies
• Verified:
  • JWT cookie is issued and stored on login
  • Requests to /api/login are handled by login pod
  • Other endpoints (search, browse, cart, place order) handled by movie pods
  • Logs confirm correct routing per service

---

Files Included in the Monolithic Directory

• Dockerfile
• decurb.yaml
• ingress.yaml
• pom.xml

Files Included in the multi-service Directories

• Dockerfile
• decurb-multi.yaml
• ingress-multi.yaml
• pom.xml

---

Demo Checklist (Video Demonstration Includes):

• AWS EC2 cluster setup shown (control + worker nodes)
• Docker Hub with both images verified
• kops get all and kubectl get all executed
• Application tested end-to-end:
  • User login
  • Movie search and add to cart
  • Checkout and DB insertion confirmed via kubectl exec mysql-secondary-0
• JWT verified in browser cookies
• Multi-service routing confirmed by analyzing pod logs

---

Notes

• ELB URL from Ingress remains consistent between single-service and multi-service deployments.
• Cluster and resources were removed after demo to prevent charges.
• HTTPS and reCAPTCHA were considered optional and were not included.