feat: add healthcheck endpoint + reconfigure ci cd

.github/workflows/ci-production.yml

@@ -36,8 +36,11 @@ jobs:
         id: create_model
         run: |
           MODEL_ID=$(gcloud ai models upload \
+            --container-ports=8080
+            --container-predict-route="/predict" \
+            --container-health-route="/health" \
             --region=${{ secrets.GCP_VERTEX_AI_REGION }} \
-            --display-name="interview-ai-detector-model" \
+            --display-name=interview-ai-detector-model \
             --container-image-uri="${{ secrets.GCP_REPO_REGION }}-docker.pkg.dev/${{ secrets.GCP_PROJECT_ID }}/interview-ai-detector/model-prediction:latest" \
             --format="value(model)")
           echo "MODEL_ID=${MODEL_ID}" >> $GITHUB_ENV
@@ -47,15 +50,15 @@ jobs:
         run: |
           ENDPOINT_ID=$(gcloud ai endpoints create \
             --region=${{ secrets.GCP_VERTEX_AI_REGION }} \
-            --display-name="interview-ai-detector-endpoint" \
+            --display-name=interview-ai-detector-endpoint \
             --format="value(name)")
           echo "ENDPOINT_ID=${ENDPOINT_ID}" >> $GITHUB_ENV
 
       - name: Deploy model to endpoint
         run: |
           gcloud ai endpoints deploy-model ${{ env.ENDPOINT_ID }} \
-            --region ${{ secrets.GCP_VERTEX_AI_REGION }} \
-            --model ${{ env.MODEL_ID }} \
-            --display-name interview-ai-detector-deployment \
-            --machine-type n1-standard-4 \
-            --accelerator count=1,type=nvidia-tesla-t4
+            --region=${{ secrets.GCP_VERTEX_AI_REGION }} \
+            --model=${{ env.MODEL_ID }} \
+            --display-name=interview-ai-detector-deployment \
+            --machine-type="n1-standard-4" \
+            --accelerator=count=1,type=nvidia-tesla-t4

Dockerfile

@@ -13,5 +13,5 @@ RUN pip install --no-cache-dir -r requirements.txt
 # Make port 8080 available to the world outside this container
 EXPOSE 8080
 
-# Run gunicorn with Uvicorn workers
-CMD ["gunicorn", "-k", "uvicorn.workers.UvicornWorker", "-w", "4", "-b", "0.0.0.0:8080", "prediction:app"]
+# Run uvicorn
+CMD ["uvicorn", "prediction:app", "--host", "0.0.0.0", "--port", "8080"]

prediction.py

@@ -1,4 +1,4 @@
-from fastapi import FastAPI
+from fastapi import FastAPI, Response, status
 from pydantic import BaseModel
 from hypothesis import BaseModelHypothesis
 from random_forest_dependencies import RandomForestDependencies
@@ -24,6 +24,11 @@ class RequestModel(BaseModel):
     instances: List[PredictRequest]
 
 
+@app.get("/health")
+async def is_alive():
+    return Response(status_code=status.HTTP_200_OK)
+
+
 @app.post("/predict")
 async def predict(request: RequestModel):
     responses = [process_instance(data) for data in request.instances]

requirements.txt

@@ -6,5 +6,4 @@ textstat
 scikit-learn==1.4.1.post1
 transformers
 fastapi
-uvicorn
-gunicorn
+uvicorn