medic / upgrade-service-kubernetes

Upgrade service for k8s

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

upgrade-service-kubernetes

Create self upgrading deployments

Tests

Tests

This tool provides an image that you can use to upgrade your pods in your kubernetes cluster.

Endpoint

The default port for upgrade-service is 5008. This can be overridden by setting the environment variable: UPGRADE_SERVICE_PORT To upgrade pods, send a POST request to /upgrade endpoint for sending upgrade commands with the payload structure below.

Payload Structure

This is the payload structure:

{
  "containers": [
    { 
      "container_name": "medic-api",
      "image_tag": "medic-api:3.05" 
    }
  ], 
  ...
}

Multiple upgrade requests can be sent in the same payload by adding more elements to the array using the same format above.

Security Considerations - RBAC (Roles, RoleBindings)

Role based access controls have been configured such that the upgrade-service can only manipulate deployments within the same namespace.

Deployment

Deploy your favorite kubernetes cluster. As an example let's use k3d.

Create a cluster like this:

k3d cluster create cht-upgrade-service --port 5008:30008@loadbalancer

Deploying the upgrade-service:

  1. Create a namespace if you don't already have one. You can also deploy to an existing namespace.

kubectl create namespace k8s-cht-deployment

  1. Deploy the upgrade-service.

kubectl -n k8s-cht-deployment apply -f kubernetes/

  1. Apply RBAC policies

kubectl -n k8s-cht-deployment apply -f kubernetes/rbac

How to quickly see it in action

To see this in action, the quickest way is to deploy a simple pod in the same namespace. Let's use nginx to demonstrate this:

Let's deploy nginx with version tag 1.20 and we'll have the upgrade-service upgrade it to 1.21

  1. Deploy nginx in the same namespace

kubectl create deployment -n k8s-cht-deployment nginx --image=nginx:1.20

  1. Send the following curl command
curl -X POST -H "Content-Type: application/json" -d '{
  "containers": [
    { 
      "container_name": "busybox", 
      "image_tag": "busybox:1.35" 
    }
  ], 
  ...
}'  localhost:5008/upgrade

You should get the response from the server stating that upgrade succeeded.

{"message":"Successfuly upgraded 1 containers"}

Confirm that the nginx pod is now using image 1.21.

Environment Variables Overwrite Available

These don't need to be set - the kubernetes setup picks them up automatically. However, if you need to overwrite any of the environment variables for whatever reason, you can use the config.json file. It contains the following variables:

    "KUBECONFIG_DEFAULT_PATH": "",
    "CHT_DEPLOYMENT_NAME": "",
    "CHT_NAMESPACE": ""

It's a good idea to not change these settings unless you know what you're doing. Here's a short description of what each one of them are for.

"KUBECONFIG_DEFAULT_PATH": This helps you specify where the kubeconfig file exists on disk. If this is specified, it will be taken as the default. "CHT_DEPLOYMENT_NAME": The name of the deployment you would like to use for the setup. "CHT_NAMESPACE": The name of the namespace you would like to use for the setup.

About

Upgrade service for k8s


Languages

Language:TypeScript 91.4%Language:JavaScript 7.6%Language:Dockerfile 1.0%