cloudbreadpapa / lstm-project

lstm-project

lstm-project in Seoul

This is sample workflow for lstm-project

AWS S3 Bucket to Azure Blob Storage

Azure Function S3 to Blob project.

LSTM model

Set up DSVM - Linux Ubuntu, execute jupyter notebook on it
https://docs.microsoft.com/en-us/azure/machine-learning/data-science-virtual-machine/dsvm-ubuntu-intro

https://<VM DNS name or IP Address>:8000/
# remind "https"

Test Jupyter notebook version info.

import sys
sys.version

To execute LSTM model, clone and execute python file.

git clone https://github.com/CloudBreadPaPa/lstm-project.git
# setup conda env or python3.x
python lstm.py

Check lstm.py file and execute on DSVM or local python virtual env.

(Optional) Install azure ml service and workbench
https://docs.microsoft.com/en-us/azure/machine-learning/preview/quickstart-installation

Prepare K8S dhackfest

• Azure CLI 2.0 https://docs.microsoft.com/en-us/cli/azure/install-azure-cli?view=azure-cli-latest

• Docker https://docs.docker.com/docker-for-windows/install/ https://docs.docker.com/docker-for-mac/install/ Enable Shared Drives - https://docs.docker.com/docker-for-windows/#shared-drives

• Install kubectl

az aks install-cli

• Helm https://github.com/kubernetes/helm/releases

• Bash On Windows https://docs.microsoft.com/en-us/windows/wsl/install-win10

• Azure Cloud Shell https://docs.microsoft.com/en-us/azure/cloud-shell/overview?view=azure-cli-latest

• Azure Service Principal https://docs.microsoft.com/en-us/azure/azure-resource-manager/resource-group-create-service-principal-portal Enable the service principal as a ‘Contributor’ on the subscription and save your Tenant Id, Subscription Id, Client Id (User id), Client Password (Secret)

• Docker Hub Create a user - https://hub.docker.com

• GitHub Create a user - https://github.com

• Visual Studio Code https://code.visualstudio.com/

• Recommended VS Code extensions YAML Docker ESLint C# Python Beautify Nuget Package Manager GitHub - Repo

• Clone the following repo to run through the workshop: https://github.com/torosent/containers-k8s-workshop git clone https://github.com/torosent/containers-k8s-workshop.git

Azure Container Registry

• Create Resource group

az group create --name myResourceGroup --location eastus

• Create a container registry

az acr create --resource-group myResourceGroup --name myContainerRegistry007 --sku Basic

• Login to ACR

az acr login --name <acrName>

• Pull test docker image

docker pull microsoft/aci-helloworld

• Get ACR Server info

az acr list --resource-group myResourceGroup --query "[].{acrLoginServer:loginServer}" --output table

• Tag docker image and push

docker tag microsoft/aci-helloworld <acrLoginServer>/aci-helloworld:v1
docker push <acrLoginServer>/aci-helloworld:v1

• List repo in Registry

az acr repository list --name <acrName> --output table

Azure Kubernetes Service - AKS

• Check az cli version

az --version

2.0.21 or higher

• Enable AKS preview

az provider register -n Microsoft.ContainerService

• Create resource group

az group create --name myResourceGroup --location eastus

• Create AKS cluster

az aks create --resource-group myResourceGroup --name myAKSCluster --node-count 1 --generate-ssh-keys

• Install kubectl for Azure cli

az aks install-cli

• Get kubernetes cluster credential and config

az aks get-credentials --resource-group myResourceGroup --name myAKSCluster

• List kube node

kubectl get nodes

• Create Azure Vote YAML

    apiVersion: apps/v1beta1
    kind: Deployment
    metadata:
      name: azure-vote-back
    spec:
      replicas: 1
      template:
        metadata:
          labels:
            app: azure-vote-back
        spec:
          containers:
          - name: azure-vote-back
            image: redis
            ports:
            - containerPort: 6379
              name: redis
    ---
    apiVersion: v1
    kind: Service
    metadata:
      name: azure-vote-back
    spec:
      ports:
      - port: 6379
      selector:
        app: azure-vote-back
    ---
    apiVersion: apps/v1beta1
    kind: Deployment
    metadata:
      name: azure-vote-front
    spec:
      replicas: 1
      template:
        metadata:
          labels:
            app: azure-vote-front
        spec:
          containers:
          - name: azure-vote-front
            image: microsoft/azure-vote-front:v1
            ports:
            - containerPort: 80
            env:
            - name: REDIS
              value: "azure-vote-back"
    ---
    apiVersion: v1
    kind: Service
    metadata:
      name: azure-vote-front
    spec:
      type: LoadBalancer
      ports:
      - port: 80
      selector:
        app: azure-vote-front

• Run application

kubectl create -f azure-vote.yaml

• Get service status

kubectl get service azure-vote-front --watch

• Monitor

kubectl proxy

• Delete cluster

az group delete --name myResourceGroup --yes --no-wait

• Scale AKS node

az aks scale --resource-group=myResourceGroup --name=myAKSCluster --node-count 3

• Manually scale pods

kubectl get pods
kubectl scale --replicas=5 deployment/azure-vote-front
kubectl get pods

• Autoscale pods

Helm

• Install Helm https://github.com/kubernetes/helm/blob/master/docs/install.md

• Initialize Helm

helm init

• Find helm chart

helm search

• Helm repo update

helm repo update

• Install from Helm

helm install stable/nginx-ingress

Jenkins Deployment

• Deploy Jenkins VM to Azure https://azuremarketplace.microsoft.com/en-us/marketplace/apps/azure-oss.jenkins?tab=Overview

• Proxy connection

ssh -L 127.0.0.1:8080:localhost:8080 [email protected]

• Login to Dashboard localhost:8080

sudo cat /var/lib/jenkins/secrets/initialAdminPassword

• Error handle

unix /var/run/docker.sock: connect: permission denied Build step 'Execute shell' marked build as failure Finished: FAILURE
sudo usermod -a -G root jenkins
sudo service jenkins restart
sudo chmod 664 /var/run/docker.sock

How to create a development infrastructure on a Linux VM in Azure with Jenkins, GitHub, and Docker

Continuous deployment with Jenkins and Azure Container Service

https://docs.microsoft.com/en-us/azure/container-service/kubernetes/container-service-kubernetes-jenkins

ACS Engine

Config ACS Engine to deply GPU on cluster node
Microsoft Azure Container Service Engine - Kubernetes

• Install acs-engine and basic usage https://github.com/CloudBreadPaPa/acs-engine/blob/master/docs/acsengine.md#install-acs-engine

• Deploy acs-engine Kubernetes cluster https://github.com/CloudBreadPaPa/acs-engine/blob/master/docs/kubernetes/deploy.md

az login
az account list -o table
acs-engine deploy --subscription-id <SUBSCRIPTION_ID> --dns-prefix <DNS_NAME> --location eastus --auto-suffix --api-model examples/kubernetes.json

acs-engine Kubernetes workthrough

• acs-engine Kubernetes GPU acs-engine GPU Kubernetes json script

Monitoring - Datadog

Persistent Volumn

Azure file or Azure Disk Reference link : https://kubernetes.io/docs/concepts/storage/persistent-volumes/

• Create PVC

kubectl create -f deployment-storage.yaml

• Get PVC

kubectl get pvc

• Deploy MySQL w/ PVC

kubectl create -f deployment-mysql.yaml

• Deploy Wordpress w/ PVC

kubectl create -f deployment-wordpress.yaml

• Login to Wordpress and set data

• Delete Pod and Services

kubectl delete pod wordpress
kubectl delete pod wordpress-mysql
kubectl delete services wordpress
kubectl delete services wordpress-mysql
kubectl delete deployment wordpress
kubectl delete deployment wordpress-mysql

• Re-deploy and check the PV data

flask Machine Learning inference application

sample flask application using keras
https://blog.keras.io/building-a-simple-keras-deep-learning-rest-api.html