Skip to content

Challenge for those applying to the platform team as devops/sre

Notifications You must be signed in to change notification settings

onrunning/platform-eng-challenge

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

29 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Platform engineering challenge

To be used in the live problem challenge

Context

This code repository contains the blog rails app based on the Getting Started with Rails guide.

Problems

There are some improvements and bug fixes that can be made in:

  • Dockerfile
  • .github/workflows/build.yml
  • The automation of repetitive tasks

Tasks

Containers

  • Looking at the Dockerfile, how can we avoid busting the cache for the dependency update everytime we update the source code ?
  • Ideally the image size for running in production should be less than 1GB (bonus)
  • Use build arguments so we can use the same ruby version as defined in file blog/.ruby-version as opposed of an hardcoded image version

Continuous Deployment

Using .github/workflows/build.yml:

  • Add the neccessary steps to tag the image based on the git commit hash
  • Uploaded to AWS ECR using 000000000000.dkr.ecr.us-east-1.amazonaws.com/blog as registry and github actions secrets AWS_SECRET_ACCESS_KEY and AWS_ACCESS_KEY_ID
  • Add another job, that depends on the build job, which would call kubectl to set the image for the k8s deployment named blog (assume the kubectl config file exists locally with the name k8s.config)
  • Alternatively, use the manifests under kubernetes to apply the change. There's a TAG placeholder that can be used in to replace with the actual commit hash.
  • Add a step on this job that would monitor the deployment progress (bonus)

Infra with terraform

Looking at the infra directory, databases are defined in rds.tf In monitor.tf there are two resources with an imaginary type that define database monitors

  • How can we simplify the change of values that keep repeating ?
  • If we get a request to create 20 more databases, how can we avoid duplicating some of the code for the required 40 extra monitors ? (hint: we might want to adjust the values for each monitor)

Kubernetes

Looking at the kubernetes directory we can see the definition of a deployment and a service

  • What kubectl command would you use to create an HorizontalPodAutoscaler with 1 minimum and 20 max replicas triggered by a cpu usage threshold of 70% ?
  • Assuming the subdomain blog.info how would you expose this application to the world ?
  • Write the necessary configuration manifest to make it happen

About

Challenge for those applying to the platform team as devops/sre

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published