hello-world-service

This service greets the entire world with hellos.

Result Link: https://hello-zeal.uploadfilter24.eu/

CI-Link: https://woodpecker.uploadfilter24.eu/repos/5
CD-Link: https://argo.uploadfilter24.eu
Grafana Dashboard: https://grafana.uploadfilter24.eu/d/85a562078cdf77779eaa1add43ccec1easdas/zealview?orgId=1&refresh=5s

Done

• Automated and re-producible provisioning of infrastructure
• Possibility to scale compute resources to more than just one instance
• Fault-tolerance for single VM outages
• Automated deployment process including (CI/CD pipeline)
  • Building the application
  • Executing tests
  • Infrastructure provisioning
  • Deploying the application
  • Basic verification of a successful deployment
• Monitoring/alerting capabilities to detect future downtimes proactively

Read Me

Everything is automated and operated via CI/CD
for updates on the website, just edit the Dockerfile push it to the git.
Create a new version tag and push the tag to git, then just watch the CI/CD doing the magic.

Requirements

• Kubernetes
• Helm
• git
• Bash

Conclusion

completed the case-study in a slightly modified way.
due to my skills, my trust in kubernetes, my will for continuous improvement and the task to:

improve their setup for deploying and operating hello-world-service with the goal to protect the service against future downtimes

i decided to build a full automated pipeline: from building the image, over a security scan, building the helmchart, releasing everything, upto a fully automated deployment to kubernetes.

some things i have added:

• trivy scan of the image
• multi arch build
• rootless nginx environment
• tolerace to high load scenarios, auto-scaling capabilitie
• build image and helm-chart aliged to the tag version
• trigger the ArgoCD to rollout the new tag version
• wait until the healthy state from ArgoCD returns
• test the Public Endpoint with the tests.sh script

information about the infrastructure:

• the Kubernetes that presents you this complete case-study, is my own freetime project.
• the Kubernetes cluster is deployed via KOPS and runs on 1 ControlPlane, 3 Nodes. The Architecture of all Servers is ARM and located at Hetzner

alternativ solution that was also in my mind:

• build the container in CI/CD
• use terraform/packer/cloudformation to spin up 3 Nodes at the CloudProvider in an Autoscaling Group
• use terraform/packer/cloudformation to place a public loadbalancer infront of those 3 nodes with an activ healthcheck for the running application by port
• provision the node with some provisioning and config management tool (ansile/salt/chef) and deploy the container to the nodes

Local development

Build

• Execute docker build . -t hello-world-service

Run

• Execute docker run -it --rm -d -p 18080:80 --name hello-world-service hello-world-service
• Open your browser and navigate to localhost:18080 and be greeted with a hello

Test

• Execute ./tests.sh localhost:18080