/klustr.dev

Primary LanguageJavaScript

"Ohana is a full-stack management tool that abstracts self-service K8s workflow for engineering/DevOps teams by giving developers the freedom to create namespace/virtual clusters on-demand achieving strong multi-tenancy in a cluster through RBAC."

REQUIREMENTS:

  • Node.js version 12.18.3+
  • Google Cloud SDK
  • vClusters
  • Helm

INSTALL:gi

  • With NPM:

TO GET STARTED:

  • Run npm install , we need to do this because our npm dependencies
  • Run ohana-*** with the "nix", "mac", "win", "m1" flags depending on your operating system or command line interface
  • Please note that if you are using WSL and Windows, please install using the ohana-nix flag; if you are using powershell, please install using the ohana-win flag
  • Google cloud may need to be installed locally for macOS: https://cloud.google.com/sdk/docs/quickstart#mac

FOR ADMINS TO CONNECT:

  • utilize gcloud init to set up a configuration; regions are currently locked to us-west1 or us-west1-a. functionality to choose or create will come in a later update;
  • Run gcloud config set account <account-email-address> to change accounts if managing more than one cluster; functionality in UI to be included later
  • Connect to the cluster with your google cloud account using gcloud auth login; you may be asked to verify in a browser

FOR ADMINS TO MANAGE TEAMS:

  • Use an admin account and log in with your credentials. You will be taken to the admin panel where you can create teams, users, and have a view of the existing teams and users
  • Create users within the UI; functionality to use a .yaml configuration for RBAC configuration
  • Once users have their own account, they'll be able to log in

FOR USERS:

  • Once you have your account and can log in, the UI is intuitive to connect and create
  • Input the necessary fields in the vCluster page:
  • Cluster : Original Cluster name to connect to
  • vCluster: the vCluster name you want
  • Host Namespace:

// necessary to create a cluster if it doesn't already exist; be aware of regional resource availability gcloud.create = (clusterName, numNodes, gcloudRegion) => gcloud container clusters create ${clusterName} --num-nodes=${numNodes} --region=${gcloudRegion} // 'gcloud container clusters get-credentials : gcloud.getCredentials = (clusterName) => gcloud container clusters get-credentials ${clusterName} --region=us-west1 // 'gcloud config set account ' gcloud.switchAccount = (gcloudUserEmail) => gcloud config set account ${gcloudUserEmail}

const kubectl = {};

kubectl.createNamespace = (hostNamespace) => kubectl create namespace ${hostNamespace} // can create spaces, accounts, configurations, namespaces etc based on the config file passed in kubectl.createFromConfig = (configFile) => kubectl apply -f ${configFile} // can create spaces, accounts, configurations, namespaces, roles, etc based on the config file passed in impersonating a user; admin only kubectl.createFromConfigAs = (configFile, userName) => kubectl apply -f /yamlConfigs/${configFile}.yaml --as=${userName} // get additional detail on pods kubectl.describe = (hostNamespace) => kubectl describe pods -n ${hostNamespace} // kubectl create deployment --image=<insert image file/link> kubectl.deployImage = (deploymentName, hostNamespace, imageFile) => kubectl create deployment ${deploymentName} -n ${hostNamespace} --image=${imageFile} // expose the deployment for kubernetes kubectl.expose = (deploymentName, hostNamespace) => kubectl expose deployment ${deploymentName} -n ${hostNamespace} --type LoadBalancer --port=80 --target-port=8080 // deploy the pod in a specific namespace with the image configuration kubectl.deploy = (hostNamespace, configFile) => kubectl apply -n ${hostNamespace} -f /Users/fenris/Desktop/Codesmith/klustr.dev/yamlConfigs/${configFile}.yaml // deploy the pod in a specific namespace with the image configuration impersonating a user; admin only kubectl.deployAs = (hostNamespace, configFile, userName) => kubectl apply -n ${hostNamespace} -f /yamlConfigs/${configFile}.yaml --as=${userName}

Stage 1: successfully initialize node, git repo, webpack and running express backend server.

  • webpack:

    • current loaders: preset-react, preset-env
    • npm run build and npm start scripts work

  • server/express

    • production server listening on 3000

  • backend database

    • not yet implemented

  • Assumes you already have a cluster that is currently running on GKE

  • Assumes you have admin privileges for said cluster (Admin) -- the Admin would then grant you rights to the namespace, which would allow you to create a vcluster

Commands used to get started:

  1. npm init -y ---> creates package.json file
  2. touch README.md ---> creates a blank README file
  3. git init ---> initialize git local repository
  4. git remote add origin <github repo link> ---> link a remote repository from github to local repo
  5. git checkout -b master ---> creates a master branch and checkouts at same time
  6. npm install webpack webpack-cli --save-dev ---> install webpack locally and install webpack-cli (tool used to run webpack on command line)
  7. touch index.html ---> creates an index.html file
  8. npm install -D babel-loader @babel/core @babel/preset-env @babel/preset-react webpack ---> install babel-loader NodeJS module, as well as presets which are just loaders which transpile code to something that can be understood by browser, e.g. ES6/7/8 -> ES5, React JSX -> JS/HTML https://webpack.js.org/loaders/babel-loader/ https://stackoverflow.com/questions/47721169/babel-vs-babel-core-vs-babel-loader-vs-babel-preset-2015-vs-babel-preset-react-v
  9. npm install --save-dev react ---> install react nodeJS module 10)npm install --save-dev react-dom ---> install react-dom nodeJS module 11)npm install --save-dev express ---> install express nodeJS module

Stage 2: get webpack dev server running

  • installed nodemon to run server and re-compile upon saves
  • installed cross-env and concurrently to allow for npm run dev script to open 8080 and reroute server requests to 3000
  • installed webpack-dev-server
  • in script 'dev' and 'build', we set the NODE_ENV to be 'development' and 'production' respectively
  • in webpack.config.js, we add a devServer property which allows us to set up config for development mode, set port to 8080, set proxy to localhost/3000
  • in webpack.config.js, we set 'mode' to process.env.NODE_ENV, to dynamically use the correct configurations depending on environment (production vs. development)
  1. npm install -g nodemon or npm install --save-dev nodemon ---> install nodemon nodeJS package
  2. npm install --global cross-env or npm install --save-dev cross-env ---> install cross-env nodeJS package
  3. npm install -g concurrently or npm install -save-dev concurrently ---> install concurrently nodeJS package
  4. npm install --save-dev webpack-dev-server ---> install webpack-dev-server, allows

Stage 3: fixing bug with serving bundle.js in production environment

  • Apparently you need to use app.use to catch the request sent in index.html for /bundle/build.js
  • need to investigate why app.get('/bundle/build) does not work.