Skip to content

Latest commit

 

History

History
191 lines (153 loc) · 6.3 KB

README.md

File metadata and controls

191 lines (153 loc) · 6.3 KB

Image Analysis - Cloud Run Service using Vision API, Vertex AI and Langchain4j

This lab can be executed directly in a Cloud Workstation, Cloudshell or your environment of choice.

Open in Cloud Shell

Setup Java ecosystem

In order to build JIT or Native Java app images, please set up Java and GraalVM and the associated Java 21 distributions. A simple installer is available from SDKMan

sdk install java 21.0.4-graal

Clone the code:

git clone https://github.com/GoogleCloudPlatform/serverless-production-readiness-java-gcp.git
cd genai/image-vision-vertex-langchain/

Install the maven wrapper

The Maven Wrapper is an easy way to ensure a user of your Maven build has everything necessary to run your Maven build.

Run the command:

mvn wrapper:wrapper

Build the service code and publish images to the container registry

Build the JIT app image:

./mvnw package

Start the app locally:
java -jar target/image-analysis-1.0.0.jar

Build the Native Java executable:

./mvnw native:compile -Pnative

Test the executable locally:
./target/image-analysis

Build Native Java Tests:

./mvnw clean package -Pnative,nativeTest

Run the native tests locally:
./mvnw native-tests

Build a JIT and Native Java container Image

./mvnw spring-boot:build-image -Dspring-boot.build-image.imageName=image-analysis-jit

./mvnw spring-boot:build-image  -DskipTests -Pnative -Dspring-boot.build-image.imageName=image-analysis-native

Check the Docker image sizes:

docker images | grep image-analysis
image-analysis-native                                     latest                aa0f7b406966   44 years ago    372MB
image-analysis-jit                                        latest                67b3489cbec8   44 years ago    467MB

Start the Docker images locally. The image naming conventions indicate whether the image was built by Maven|Gradle and contains the JIT|NATIVE version

docker run --rm image-analysis-jit
docker run --rm image-analysis-native

Retrieve the Project ID, as it will be required for the next GCP operations

export PROJECT_ID=$(gcloud config get-value project)
echo $PROJECT_ID

Tag and push the images to GCR:

docker tag image-analysis-jit gcr.io/${PROJECT_ID}/image-analysis-jit
docker tag image-analysis-native gcr.io/${PROJECT_ID}/image-analysis-native

docker push gcr.io/${PROJECT_ID}/image-analysis-jit
docker push gcr.io/${PROJECT_ID}/image-analysis-native

Deploy and run workshop code

Enable the required APIs, if they are not already enabled:

gcloud services enable vision.googleapis.com
gcloud services enable cloudfunctions.googleapis.com
gcloud services enable cloudbuild.googleapis.com 
gcloud services enable run.googleapis.com
gcloud services enable logging.googleapis.com 
gcloud services enable storage-component.googleapis.com 
gcloud services enable aiplatform.googleapis.com

Retrieve the Project ID and Project Number

# get the Project_ID
export PROJECT_ID=$(gcloud config list --format 'value(core.project)')
  or 
export PROJECT_ID=$(gcloud config get-value project)

# get the Project_Number
export PROJECT_NUMBER=$(gcloud projects describe $PROJECT_ID --format='value(projectNumber)')

Create the GCS bucket

export BUCKET_PICTURES=vision-${PROJECT_ID}
gsutil mb -l us-central1 gs://${BUCKET_PICTURES}
gsutil uniformbucketlevelaccess set on gs://${BUCKET_PICTURES}
gsutil iam ch allUsers:objectViewer gs://${BUCKET_PICTURES}

Create the database

Instructions for configuring cloud Firestore available here

Set config variables

gcloud config set project ${PROJECT_ID}
gcloud config set run/region 
gcloud config set run/platform managed
gcloud config set eventarc/location us-central1

Grant pubsub.publisher to Cloud Storage service account

SERVICE_ACCOUNT="$(gsutil kms serviceaccount -p ${PROJECT_ID})"

gcloud projects add-iam-policy-binding ${PROJECT_ID} \
    --member="serviceAccount:${SERVICE_ACCOUNT}" \
    --role='roles/pubsub.publisher'

Deploy to Cloud Run

# deploy JIT image to Cloud Run
gcloud run deploy image-analysis-jit \
     --image gcr.io/${PROJECT_ID}/image-analysis-jit \
     --region us-central1 \
     --memory 2Gi --allow-unauthenticated

# deploy native Java image to Cloud Run

gcloud run deploy image-analysis-native \
     --image gcr.io/${PROJECT_ID}/image-analysis-native \
     --region us-central1 \
     --memory 2Gi --allow-unauthenticated

Set up Eventarc triggers

gcloud eventarc triggers list --location=us-central1

gcloud eventarc triggers create image-analysis-jit-trigger \
     --destination-run-service=image-analysis-jit \
     --destination-run-region=us-central1 \
     --location=us-central1 \
     --event-filters="type=google.cloud.storage.object.v1.finalized" \
     --event-filters="bucket=vision-${PROJECT_ID}" \
     --service-account=${PROJECT_NUMBER}[email protected]

gcloud eventarc triggers create image-analysis-native-trigger \
     --destination-run-service=image-analysis-native \
     --destination-run-region=us-central1 \
     --location=us-central1 \
     --event-filters="type=google.cloud.storage.object.v1.finalized" \
     --event-filters="bucket=vision-${PROJECT_ID}" \
     --service-account=${PROJECT_NUMBER}[email protected]

Test the trigger

gsutil cp images/GeekHour.jpeg gs://vision-${PROJECT_ID}
gsutil cp images/CloudRun.png gs://vision-${PROJECT_ID}

gcloud logging read "resource.labels.service_name=image-analysis-jit AND textPayload:GeekHour" --format=json
gcloud logging read "resource.labels.service_name=image-analysis-jit AND textPayload:CloudRun" --format=json

Log capture:

 gcloud logging read "resource.labels.service_name=image-analysis-jit AND textPayload:CloudRun" --format=json
 gcloud logging read "resource.labels.service_name=image-analysis-native AND textPayload:CloudRun" --format=json