Тестовое задание для поступления в DevOps Cloud.ru Camp

Результаты выполнения тестового задания следует опубликовать на GitHub или захостить на любой открытой платформе (например, Github Pages) и отправить на почту devopscloudcamp@cloud.ru. Также следует указать свои контактные данные для получения обратной связи.

1. Ansible playbook

How to setup ansible.cfg and launch your playbook

• Add the your private key to the directory you need
• Specify it in the your_repo/ansible.cfg file under the private_key_file option.
• Set the inventory file in the inventory option.
• Run the command ansible-playbook playbook.yml in directory your_repo/playbook.

How to fast check ansible playbook if you don't have inventory

To fast check playbook i used Vagrant

Execute next commands:

cd ~/your_repo/playbook
vagrant up

After starting the virtual machine, you will see the execution of the playbook with the keys that vagrant creates for his needs. You should see next output

PLAY RECAP *********************************************************************
tag_cloud                  : ok=7    changed=5    unreachable=0    failed=0    skipped=0    rescued=0    ignored=0

After this you can check idempotence:

vagrant provision tag_cloud

expected result:

PLAY RECAP *********************************************************************
tag_cloud                  : ok=5    changed=0    unreachable=0    failed=0    skipped=1    rescued=0    ignored=0

You also can use inventory from vagrant inventory file in directory your_repo/playbook/.vagrant/provisioners/ansible/inventory/vagrant_ansible_inventory to set up your ansible.cfg and test ansible-playbook playbook.yml command

However, for optimal idempotency testing, it's advisable to use specialized tools like Molecule.

2. Web application on Python

• A Python script has been created using Fastapi to assign endpoints.
• Created endpoints: /id /hostname /author readiness liveness
• A dependency file requirements.txt has been created for launching applications

How to setup application

setup dependecies

cd ~/your_repo/app
pip install --upgrade -r requirements.txt

After launch your app

AUTHOR=some_name UUID=some_valid_uuid_v4_or_not_valid_to_check_readiness_probe  uvicorn main:app --host 0.0.0.0 --port 8000 --reload

How to check application

curl http://localhost:8000/created_endpoints

3. Dockerfile

A Dockerfile was created in the your_repo/app directory. To create it was used:

• Images are based on alpine.
• Specific versions of base images are pinned during the build.
• Dependencies in the requirements.txt file are pinned to specific versions required for the application's execution.
• Multi-stage build is employed.
• Layers that are likely to change frequently, such as app source code, are positioned towards the end of the Dockerfile.
• The final image is created as rootless.

I am aware of the existence of distroless images, https://github.com/GoogleContainerTools/distroless but the documentation separately states:

The following images are also published on gcr.io, but are considered experimental and not recommended for production usage:
gcr.io/distroless/python3-debian11

As a result, I chose not to utilize it.

I also chose not to set a default value for the environment variable UUID, because it's challenging for people to comprehend and remember entries like 54015250-c765-48a3-9921-6bc47bb40e11. If you happen to forget to specify this variable in a Kubernetes Deployment, your application will fail the readiness check. But if you have a default value set, this might go unnoticed.

How to create docker image

To create the final image, run the following commands:

cd ~/your_repo/app
docker build -t your_hub_name/image_name:tag .
# If necessary, push to your custom Docker registry or Docker Hub
docker push your_hub_name/image_name:tag

How to check your image

To test your image, run the following commands:

# You can try different scenarios by passing a valid or invalid UUID to the variable.
docker run -e UUID=some_valid_or_not_UUID -d -p 8000:8000 your_hub_name/image_name:tag

curl http://localhost:8000/endpoints_in_app

Kubernetes manifest

• A deployment.yml manifest has been created, which sets up the namespace cloudru and the deployment itself.
• A service.yml manifest with a ClusterIP type has been created to serve the deployment.

The deployment has 3 replicas. The UUID variable is assigned the value of meta.uid during the deployment rollout. This variable is used in the readiness probe within the application code. The AUTHOR variable is also used for the readiness probe validation, ensuring you didn't forget to set the variable in the Dockerfile.

How to rollout your deployment

If you don't have kubernetes cluster use minikube

minikube start  --kubernetes-version=v1.26.3

rollout your deployment

cd ~/your_repo/manifest
kubectl apply -f ./

How to check your deploy

Check readiness your pods and your service

kubectl get pod -n your_namespace
kubectl get svc -n your_namespace
# or inspect events to catch errors
watch kubectl events -n your_namespace
# or inspect pod logs
kubectl logs pod_name -n your_namespace --watch

Check working your app

kubectl port-forward your_pod_name 8000:8000
# open new terminal
curl http://localhost:8000/endpoints

Helm chart

In the /helm directory, a chart has been created that sets up a deployment and a service for it, as well as a README.md file generated using the helm-docs utility.

How to setup and deploy helm chart

cd ~/your_repo/helm
# Ensure your manifests are correctly generated
helm template .
# Deploy
helm upgrade --install your_release_name . --namespace your_namespace --create-namespace

How to check your deployed chart

# Check created service name
kubectl get svc -n your_namespace
# Launch pod with curl and sh/bash/etc
kubectl run pod_name --image image_with_curl_and_sh -n your_namespace
k exec -it debug -n your_namespace -- sh
# within container check your app endpoints
curl http://service_name:8000/endpoints

Endpoints such as /id /hostname can be checked several times in a row to ensure that your service is load balancing between different pods, and you can receive a different UUID and /hostname from each pod.