advanced-analytics-apps-on-azure-appservice
Instructions on how to deploy containerized Data Analytics on Azure App Service.
Overview
Required Tools
Please make sure you have the following tools installed on your machine.
- Code Editor: Visual Studio Code https://code.visualstudio.com/
- git source-code management https://git-scm.com/
- azure command line interface 'az' https://docs.microsoft.com/en-us/cli/azure/install-azure-cli?view=azure-cli-latest (Azure CLI)
- Docker for Windows https://docs.docker.com/docker-for-windows/install/
Create an Azure DevOps project
We will use Azure DevOps to host the source-code and secondly utilize Azure Pipelines to automate build and deployment tasks.
- login at https://dev.azure.com using your corporate credentials.
- Please refer to https://docs.microsoft.com/en-us/azure/devops/organizations/projects/create-project?view=azure-devops for further guidance.
Clone the repository
In your DevOps project please proceed to "Repos".
Clone the repository to your local computer.
Download the example project for Bokeh
Please proceed to the following URL.
Please just perform the ZIP download and extract the content in the previously cloned git repository
https://github.com/dariustehrani/bokeh-on-docker
Test the Dockerfile
- Open Powershell (in Visual Studio Code use STRG+ö on a german keyboard, type powershell)
- CD into the project folder
- run
docker build -t bokeh-on-docker:latest . - test the image by running
docker run -p 8080:8080 -it bokeh-on-docker:latest - Open a browser and point it to http://localhost:8080 You should see a simple bokeh demo app.
Did this work for you?
Excellent. Within your project directory perform:
git add .
git commit . -m "your commit message"
git push
Visit the repository view in Azure DevOps to check if your code has arrived there.
Azure Container Registry (azurecr.io)
Typically we would not want to build and persist Docker images locally but have Azure DevOps manage this for us. Let's create a Container Registry. You can do this in the portal or using the following Azure CLI commands (in PowerShell):
az account show
If you are not logged in, run:
az login
Create a resource group:
az group create -l westeurope -n bokehondockerYOURNAME
Create your azure container registry
az acr create -n bokehYOURNAME -g bokehondockerYOURNAME --sku Standard --admin-enabled
azure container registry docker login
az acr login --name bokehYOURNAME
(optional) show the current username and password
az acr credential show -n bokehyourname
Azure Pipeline Setup
Create a new service principal (SP)
The SP will be used by Azure DevOps to connect to your Azure subscription and manage resources on your behalf.
You need to replace the values with your subscription ID and Resource Group Name.
az account show
az ad sp create-for-rbac -n "bokehondocker" --role contributor --scopes /subscriptions/{SubID}/resourceGroups/{ResourceGroup1}
Setup Azure DevOps Service Connections
AzureRM Service connection
Insert the credentials you receive here:
- Proceed to your Azure DevOps project.
- Click "Project settings" -> "Service connections" -> "New service connection" -> Select "Azure Resource Manager".
- Define a meaningful name. Select
"Scope:Subscription"
"Subscription: YOURSUBSCRIPTION"
"Resource Group: NAMEOFRGYOUCREATED" - Click "use the automated version of the service connection dialog."
- Insert the AppID from the shell output into "Service principal client id" .
- Insert the key into "Service principal client ID".
- Click "Verify connection".
- Make sure you tick the "Allow all pipeline to use this connection" box.
Azure Container Registry Service connection:
- Click "New servic connection" and select "Docker Registry".
- Select "Azure Container Registry" and fill out the remaining information.
- Take note of your service connection names.
Bring up the Azure DevOps Pipeline
- Click on "Pipelines" -> "Builds"
- Click "New Pipeline"
- Select "Azure Repos Git" and click "Continue".
- Choose "Hosted Ubuntu 1604" as Agent pool.
- Select YAML and specify the path to your existing azure-pipelines.yml file.
- Save the build pipeline
- Proceed to the "Pipelines" overview and choose to edit the newly created pipeline.
- Replace the containerRegistry with your service connection name.
- Click "Run"
Azure Web App
- Create a new Web App on https://portal.azure.com
- Choose your subscription and resource group and select a compact but meaningful name.
- Choose "Docker Image", "Linux" and the location "West Europe"
- Create a new App Service Plan with the SKU "P1v2"
- Leave the rest as is and create the App Service
-
Once the service has been created proceed to your App Service and click on the Container settings
-
Choose your Azure Container Registry with the image listed Tag "latest" and Continuous deploy: OFF
-
Scroll down to "Diagnostic logs" in the Monitoring section Enable Application Logging filesystem 100MB quota 30 days retention. Click save.
-
Head over to “Deployment Slots” Create a DEV slot. Make sure to copy the settings from the other slot.
-
Proceed to azure DevOps, uncomment the settings in your azure-pipeline.yml (second section) by removing the # signs. Modify the code to correspond with your Subscription, Resource Group, WebApp, Slot Name as well as the service connection you have created.
-
Save your settings and run the pipeline.
-
Find the URL of your Web App where you should see your bokeh demo app!
Shiny R on Microft Open R (ADVANCED USERS ONLY)
As building the R Packages take a lot of time, the following example separates the creation of the runtime container from the actual Shiny app container. This will save 18 Minutes on average per deployment. Secondly you might want to set up a scheduled trigger, e.g. on a weekly basis so that your base images always includes latest updates and patches.
Microsoft R Open Shiny BASE Image
Build a CI/CD Pipeline for the base Image. https://github.com/dariustehrani/mro-shiny-base
Microsoft R Open Shiny APP Image
The following example will build shiny app container. You will need to modify the FROM entry to reference your azure container registry.
FROM YOUR.azurecr.io/mro-shiny-base:latest