Skip to content
/ gepp Public

Developer's Helper to Docker, Kubernetes, and Terraform. Fully automatic, without any config or question 🙌

License

Notifications You must be signed in to change notification settings

gurayops/gepp

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GEPP - Developer's Helper to K8s

GEPP takes your app and Dockerize it, sets up a Kubernetes cluster and runs your app in it, configure K8s resources and produce Terraform file for Azure deployments, and more! All are happening in seconds!

GEPP Demo

How and Why

Requirements:

  • Running Docker
  • main.py file
  • app variable for WYSGI

Run:

docker run -it --rm -v /var/run/docker.sock:/var/run/docker.sock -v $PWD:/project -e PROJECT_NAME=${PWD##*/} guray/gepp:0.1-alpha5

GEPP provides developers to easily run their apps as containerized apps in Kubernetes. Here is what it is aiming:

  • You run the tool in a directory with Python/Go/NodeJS code
  • It will generate a Dockerfile depending on the framework your code uses
  • It will generate YAMLs for Kubernetes deployment, pod autoscaling, service definitions (it will ask some questions like service type) + ingress
  • (optional) It can generate a plan for Terraform for your cloud provider for a K8s cluster and deploying your app (in JSON, using tfcdk)
  • (optional) It will insert environment variable, secretmap, configmaps to the generated YAML (interactively asks) - WIP
  • (optional) It can get storageclasses from your cluster and define some storage and insert it to your YAML - WIP
  • (optional) It can generate servicemonitor and podmonitor YAML files for Prometheus operator - WIP
  • (optional) It can generate network policies for some of the network providers - WIP

It is mostly a repetitive task to Dockerize your project and generating YAMLs that look like ones in other projects/microservices. In addition to replaying the same steps, it may easily consume a lot of time to set up to Dockerize your app, write YAMLs for Kubernetes deployment(and define resources), configure autoscheduler, write Ingress manifests for L7 routing/LB, create Kubernetes services etc. After all of these, you should deploy a Kubernetes cluster, either locally or on a remote target; than connect to it, upload & deploy your app.

GEPP is being developed to shorten these procedures to a fully(or mostly) automatic, yet useful process. Just run gepp inside a directory and wait for your localhost-port pair to connect your app.

For interactively using GEPP run gepp -i or gepp --interactive. (Currently works for only additional ports)

You will find Dockerfile, .dockerignore, Kubernetes YAMLS, AKS definition for deploying it with Terraform in your directory.

Status

Currently, it is in 0.1-alpha stage. It should work without any problems but has not tested in wild, nor all the intended features has been implemented yet. If you are encountering a problem, please open an issue.

PR's are more than welcome!

The proposal of this project consists of these steps:

  • Auto detect Flask or other frameworks
  • Generate Dockerfile if not exists
  • Create Deployment YAML
  • Create autoscaler YAML
  • Create service (type optional, clusterip default)
  • Create ingress definition (optional)
  • Deploy to K8s (optional)
  • Test with port-forward
  • Generate Terraform file for K8s integration + GKE/AWS/Azure/...
  • Ask secret usage, getting names from K8s (optional)
  • Ask disk type, size, mountpoint (optional)
  • Create local cluster with kind/minikube/k3d and deploy the app onto it
  • Use skaffold/telepresence to provide a complete dev environment
  • Get status, logs, metrics
  • Ask metric exports
  • Re-running should be possible, and it should check the YAMLs and other files to be idempotent

Releases

No releases published

Packages

No packages published