Skip to content

DEPLOYMENT.md

Latest commit

 

History

History
72 lines (53 loc) · 1.61 KB

DEPLOYMENT.md

File metadata and controls

72 lines (53 loc) · 1.61 KB
Raw

Deployment Guide

Live API: https://java-sentiment-api.onrender.com

Docker

docker build -t sentiment-api .
docker run -p 8080:8080 sentiment-api

With custom configuration:

docker run -d \
  -p 8080:8080 \
  -e SPRING_PROFILES_ACTIVE=production \
  -e MAX_HEAP=1g \
  sentiment-api

Local Development

mvn clean package -DskipTests
mvn spring-boot:run

Requires production model in models/production/ (run ./scripts/promote_to_production.sh first).

Configuration

Variable Default Description
SPRING_PROFILES_ACTIVE (none) Set to production for stricter limits
SENTIMENT_SVM_MODEL /app/models/production/sentiment_model.ser Model path
MAX_HEAP 512m JVM max heap
SERVER_PORT 8080 API port

Rate Limits:

  • Default: 100 single/min, 20 batch/min
  • Production profile: 60 single/min, 10 batch/min

Health Check

curl http://localhost:8080/api/v1/health

Actuator endpoints: /actuator/health, /actuator/prometheus

Performance

Metric Value
Startup time < 10 seconds
Memory 512MB - 1GB
Latency <10ms mean, <3ms p99
Batch (100 texts) 50-100ms

Troubleshooting

Issue Solution
Model not found Run ./scripts/promote_to_production.sh
Out of memory Increase -e MAX_HEAP=1g
Slow responses Increase CPU: --cpus 2

Production Checklist

  • Model exists: models/production/sentiment_model.ser
  • Health endpoint responds: /api/v1/health
  • Memory limits configured