deploy-backend

name	deploy-backend
description	Deploy do backend Django para Cloud Run (GCP). Usar quando precisar publicar mudancas da API em producao.
disable-model-invocation	true
allowed-tools	Bash, Read, Grep

Deploy Backend (Cloud Run)

Deploy do {{BACKEND_DIR}} para o Cloud Run na GCP.

Pre-requisitos

• Docker rodando localmente
• gcloud autenticado (gcloud auth login)
• Credenciais de Docker configuradas (gcloud auth configure-docker {{GCP_REGION}}-docker.pkg.dev)

Passos

1. Verificar mudancas pendentes:

   cd {{BACKEND_DIR}} && git status
   

   • Se houver mudancas nao commitadas, avisar o usuario

2. Verificar se ha migrations pendentes:

   cd {{BACKEND_DIR}} && source venv/bin/activate && python manage.py showmigrations | grep "\[ \]"
   

   • Se houver migrations nao aplicadas, avisar (precisam ser aplicadas no Cloud SQL tambem)

3. Build da imagem Docker:

   cd {{BACKEND_DIR}}
   IMAGE="{{GCP_REGION}}-docker.pkg.dev/{{GCP_PROJECT}}/{{ARTIFACT_REPO}}/{{BACKEND_SERVICE}}:latest"
   docker build -f Dockerfile.cloudrun -t $IMAGE .
   

4. Push para Artifact Registry:

   docker push $IMAGE
   

5. Deploy no Cloud Run:

   gcloud run deploy {{BACKEND_SERVICE}} \
     --image $IMAGE \
     --region {{GCP_REGION}} \
     --platform managed \
     --allow-unauthenticated \
     --min-instances 1 \
     --max-instances 10 \
     --memory 1Gi \
     --cpu 1 \
     --timeout 3600 \
     --concurrency 80 \
     --no-cpu-throttling
   

6. Verificar deploy:

   gcloud run services describe {{BACKEND_SERVICE}} --region {{GCP_REGION}} --format="value(status.url)"
   

CRITICO

• SEMPRE usar --memory 1Gi — 512Mi causa OOM em producao
• URL de producao: https://{{PROD_API_URL}}
• Cloud SQL: {{GCP_PROJECT}}:us-central1:{{CLOUD_SQL_INSTANCE}}