Deploy AI powered PetClinic with telemetry on Azure Container Apps

This is a quickstart that shows you deploy PetClinic App powered by OpenAI on Azure Container Apps, with OpenTelemetry enabled to collect and read logs and traces in Azure Monitor Application Insights.

Prerequisites

This quickstart requires the following to be installed on your machine.

• Unix-like operating system installed. For example, Ubuntu, Azure Linux, macOS, WSL2.
• Git
• Azure CLI 2.59.0 or higher version
• JDK 17
• Maven

Azure CLI setup

Check your Azure CLI version and upgrade to the required version if needed.

az --version
az upgrade

Install or update the Azure Application Insights and Azure Container Apps extensions.

az extension add -n application-insights --upgrade --allow-preview true
az extension add --name containerapp --upgrade --allow-preview true

Sign in to Azure CLI if you haven't already done so.

az login

Set the default subscription to use.

az account set --subscription "<subscription-id>"

Register the Microsoft.App and Microsoft.OperationalInsights namespaces if they're not already registered in your Azure subscription.

az provider register --namespace Microsoft.App
az provider register --namespace Microsoft.OperationalInsights

Define variables

Please replace placeholder values in <> with your own.

export AZURE_OPENAI_ENDPOINT="<azure-openai-endpoint>"
export AZURE_OPENAI_KEY="<azure-openai-key>"
export AZURE_SEARCH_ENDPOINT="<azure-search-endpoint>"
export AZURE_SEARCH_KEY="<azure-search-key>"
export AZURE_OPENAI_DEPLOYMENTNAME_CHAT="<chat-model-name>"
export AZURE_OPENAI_DEPLOYMENTNAME_EMBEDDING="<embedding-model-name>"

UNIQUE_VALUE=pc-$(date +%s)
LOCATION=eastus

RESOURCE_GROUP_NAME=${UNIQUE_VALUE}rg
ACA_ENV=${UNIQUE_VALUE}env
APP_INSIGHTS=${UNIQUE_VALUE}appinsights
CONFIG_SERVER=${UNIQUE_VALUE}configserver
CONFIG_SERVER_GIT_URI="https://github.com/seanli1988/petclinic.git"
ACA_AI_NAME=${UNIQUE_VALUE}ai
ACA_PETCLINIC_NAME=${UNIQUE_VALUE}petclinic
WORKING_DIR=$(pwd)

Create an Azure Container Apps Environment

Create a resource group and an Azure Container Apps environment.

az group create \
    -n $RESOURCE_GROUP_NAME \
    -l ${LOCATION}

az containerapp env create \
    --resource-group $RESOURCE_GROUP_NAME \
    --location $LOCATION \
    --name $ACA_ENV

Collect and read OpenTelemetry data in Azure Container Apps

OpenTelemetry agents live within your container app environment. You configure agent settings through the Azure CLI.

The managed OpenTelemetry agent accepts the following destinations:

• Azure Monitor Application Insights
• Datadog
• Any OTLP endpoint (For example: New Relic or Honeycomb)

In this tutorial, you use Azure Monitor Application Insights as the destination.

First, create an Azure Application Insights resource to receive OpenTelemetry data.

logAnalyticsWorkspace=$(az monitor log-analytics workspace list \
    -g $RESOURCE_GROUP_NAME \
    --query "[0].name" -o tsv | tr -d '\r\n')

az monitor app-insights component create \
    --app $APP_INSIGHTS \
    -g $RESOURCE_GROUP_NAME \
    -l $LOCATION \
    --workspace $logAnalyticsWorkspace

Next, enable OpenTelemetry for the Azure Container Apps environment and configure it to send data to the Azure Application Insights resource.

appInsightsConn=$(az monitor app-insights component show \
    --app $APP_INSIGHTS \
    -g $RESOURCE_GROUP_NAME \
    --query 'connectionString' -o tsv)

az containerapp env telemetry app-insights set \
  --name $ACA_ENV \
  --resource-group $RESOURCE_GROUP_NAME \
  --connection-string $appInsightsConn \
  --enable-open-telemetry-logs true \
  --enable-open-telemetry-traces true

Set up a managed Config Server for Spring microservices

Azure Container Apps provides a managed Config Server for Spring microservices. The Config Server is a centralized configuration service that serves configuration data to client applications.

Create a Config Server and configure it to use a Git repository as the backend.

az containerapp env java-component config-server-for-spring create \
  --environment $ACA_ENV \
  --resource-group $RESOURCE_GROUP_NAME \
  --name $CONFIG_SERVER \
  --configuration spring.cloud.config.server.git.uri=$CONFIG_SERVER_GIT_URI spring.cloud.config.server.git.refresh-rate=30

Deploy PetClinic AI

The PetClinic AI integrates with OpenAI service using LangChain for Java. Follow instructions below to deploy it on ACA.

First, prepare the source code for PetClinic AI by cloning the repository and target branch.

cd $WORKING_DIR
git clone --single-branch --branch ai git@github.com:seanli1988/petclinic.git petclinic-ai
cd petclinic-ai

Next, build and deploy the PetClinic AI service to Azure Container Apps.

# Build the artifact
mvn clean package -DskipTests=true

# Deploy the artifact to ACA
az containerapp create \
    --resource-group $RESOURCE_GROUP_NAME \
    --name $ACA_AI_NAME \
    --environment $ACA_ENV \
    --artifact ./target/demo-0.0.1-SNAPSHOT.jar \
    --target-port 8080 \
    --ingress 'internal' \
    --env-vars \
    AZURE_OPENAI_ENDPOINT=$AZURE_OPENAI_ENDPOINT \
    AZURE_OPENAI_KEY=$AZURE_OPENAI_KEY \
    AZURE_SEARCH_ENDPOINT=$AZURE_SEARCH_ENDPOINT \
    AZURE_SEARCH_KEY=$AZURE_SEARCH_KEY \
    AZURE_OPENAI_DEPLOYMENTNAME_CHAT=$AZURE_OPENAI_DEPLOYMENTNAME_CHAT \
    AZURE_OPENAI_DEPLOYMENTNAME_EMBEDDING=$AZURE_OPENAI_DEPLOYMENTNAME_EMBEDDING \
    --min-replicas 1

Then, bind the PetClinic AI service to the Config Server.

az containerapp update \
  --name $ACA_AI_NAME \
  --resource-group $RESOURCE_GROUP_NAME \
  --bind $CONFIG_SERVER:myconfigserver

Alternatively, you can bind the PetClinic AI service to the Config Server from the Azure Portal.

• Go to the Azure Portal, navigate to the Azure Container Apps environment, and select Services > Services (Preview).
• Select service with type Spring Cloud Config.
• Under section Bindings from the popup window, select and bind the PetClinic AI service to the Config Server by following the wizard.
• If you don't see number of Connected Apps increase by 1, wait for a while, refresh the page and try again.

Deploy PetClinic app

The PetClinic app talks to PetClinic AI service. Follow instructions below to deploy it on ACA.

First, prepare the source code for PetClinic app by cloning the repository.

cd $WORKING_DIR
git clone git@github.com:seanli1988/petclinic.git
cd petclinic

Next, build and deploy the PetClinic app to Azure Container Apps.

# Build the artifact
mvn clean package -DskipTests=true

# Deploy the artifact to ACA
az containerapp create \
    --resource-group $RESOURCE_GROUP_NAME \
    --name $ACA_PETCLINIC_NAME \
    --environment $ACA_ENV \
    --artifact ./target/spring-petclinic-3.2.0-SNAPSHOT.jar \
    --target-port 8080 \
    --ingress 'external' \
    --env-vars \
    PETCLINIC_AI_HOST=http://${ACA_AI_NAME} \
    SPRING_DATASOURCE_URL=jdbc:otel:h2:mem:db \
    SPRING_DATASOURCE_DRIVER_CLASS_NAME=io.opentelemetry.instrumentation.jdbc.OpenTelemetryDriver \
    --min-replicas 1

Then, bind the PetClinic app to the Config Server.

az containerapp update \
  --name $ACA_PETCLINIC_NAME \
  --resource-group $RESOURCE_GROUP_NAME \
  --bind $CONFIG_SERVER:myconfigserver

Run the following command to output the URL of the PetClinic, and open the URL in your web browser to talk with OpenAI!

PETCLINIC_APP_URL=https://$(az containerapp show \
    --resource-group $RESOURCE_GROUP_NAME \
    --name $ACA_PETCLINIC_NAME \
    --query properties.configuration.ingress.fqdn \
    --output tsv)
echo "PetClinic app URL: $PETCLINIC_APP_URL"

Clean up

Congratulations! You have successfully deployed the AI powered PetClinic to Azure Container Apps.

Now you can clean up the resources to avoid incurring charges if they are not needed. Run the following command to delete the resource group and all resources created in this tutorial.

az group delete \
    --name $RESOURCE_GROUP_NAME \
    --yes --no-wait

Run the PetClinic locally with OpenTelemetry disabled

You can also run the PetClinic locally with OpenTelemetry disabled, to test it before deploying it to Azure Container Apps.

Run PetClinic ai locally

Run the following commands to start the PetClinic ai locally.

cd $WORKING_DIR/petclinic-ai
mvn spring-boot:run -Dspring-boot.run.arguments="--otel.sdk.disabled=true" 

Note that the PetClinic ai requires the following environment variables to be set. If you followed the instructions above and keep the terminal open, you should have them set already.

export AZURE_OPENAI_ENDPOINT="<azure-openai-endpoint>"
export AZURE_OPENAI_KEY="<azure-openai-key>"
export AZURE_SEARCH_ENDPOINT="<azure-search-endpoint>"
export AZURE_SEARCH_KEY="<azure-search-key>"
export AZURE_OPENAI_DEPLOYMENTNAME_CHAT="<chat-model-name>"
export AZURE_OPENAI_DEPLOYMENTNAME_EMBEDDING="<embedding-model-name>"

Run PetClinic app locally

Open a new terminal window, locate the PetClinic app directory which is $WORKING_DIR/petclinic, and run the following commands to start the PetClinic app locally.

mvn spring-boot:run -Dspring-boot.run.arguments="--otel.sdk.disabled=true --server.port=8081"

Open a web browser and navigate to http://localhost:8081 to access and test the PetClinic app powered by AI.

References

Refer to the following links for more information.

1. LangChain for Java: Supercharge your Java application with the power of LLMs
2. Retrieval Augmented Generation (RAG) in Azure AI Search
3. From zero to - nearly - hero with Azure OpenAI NLP and vector-based search in Azure Cognitive Search
4. Understanding how the Azure OpenAI "use your data" feature works
5. Collect and read OpenTelemetry data in Azure Container Apps
6. OpenTelemetry Collector
7. How to instrument Spring Boot with OpenTelemetry
8. Create an Azure Monitor Application Insights using Azure CLI
9. Tutorial: Connect to a managed Config Server for Spring in Azure Container Apps (preview)
10. Connect applications in Azure Container Apps