Building an AI-Powered Spring Boot Microservices Application: My Journey, Issues & Fixes

In this post, I’ll walk you through my experience building an AI-powered fitness recommendation app using Spring Boot microservices, integrated with MongoDB Atlas, Eureka Service Discovery, and Google Gemini for personalized AI recommendations.

🎯 In this post, I’ll highlight:

• 🧱 The architecture I used

• 🔨 Key steps in building it

• ❗ Challenges I faced

• ✅ How I fixed them

---

🏗️ Project Architecture

My application is made up of three microservices (so far):

1. User Service – Handles user data (uses PostgreSQL)

2. Activity Service – Stores activity logs (uses MongoDB Atlas)

3. AI Recommendation Service – Generates personalized fitness plans using Google Gemini

All services are registered with the Eureka Server for service discovery.

---

🔧 Tech Stack

• Java 17

• Spring Boot 3.1.9

• Spring Cloud 2025.0.0

• MongoDB Atlas

• PostgreSQL

• Google Gemini (AI API)

• Eureka Server

• Docker (for future deployment)

---

🛠️ Steps I Followed

• ✅ Created microservices using Spring Initializr

• ✅ Integrated JPA & PostgreSQL for user-service

• ✅ Set up MongoDB Atlas for activity-service

• ✅ Integrated Google Gemini API in ai-service

• ✅ Added Eureka Server and registered all services

• ✅ Tested all APIs using Postman

• ✅ Used Spring Cloud OpenFeign for inter-service communication

---

🧱 Problems I Faced & How I Solved Them:

1. ❌ MongoSocketReadException — MongoDB Atlas Connection Issue

• Issue: While connecting activity-service to MongoDB Atlas, I got a MongoSocketReadException.

• Root Cause: I had not whitelisted my current IP address in the MongoDB Atlas dashboard.

• Fix:

  • Go to MongoDB Atlas Console → Network Access

  • Click Add IP Address

  • Use "Allow access from anywhere" (0.0.0.0/0) or your specific IP

  • Save and reconnect

---

2. ❌ Eureka Client Not Registering Microservices

• Issue: Microservices were not appearing in the Eureka Dashboard.

• Root Cause: I forgot to add the Eureka Client dependency in the pom.xml.

• Fix:

  • Add the following to each microservice:

    <dependency> <groupId>org.springframework.cloud</groupId> <artifactId>spring-cloud-starter-netflix-eureka-client</artifactId> </dependency>

  • Also add these properties:

    eureka.client.register-with-eureka=true eureka.client.fetch-registry=true eureka.client.service-url.defaultZone=http://localhost:8761/eureka/

  • Make sure every service has a unique spring.application.name

---

💡 Learnings

• ✅ Always whitelist your IP in MongoDB Atlas

• ✅ Never miss essential dependencies like Eureka Client

• ✅ Use Spring Profiles for environment-specific configs

• ✅ Test each microservice individually before full integration

---

📌 Conclusion

This project helped me:

• 🧩 Build scalable microservices using Spring Boot

• 🔐 Securely connect to external databases

• 🧠 Integrate AI-powered services using Google Gemini

• 🛠️ Troubleshoot real-world issues with networking and microservice registration

---

🔗 What’s Next?

• ⚙️ Setting up asynchronous communication using RabbitMQ for better service decoupling

• 🛡️ Implementing authentication & authorization using Keycloak

• 💻 Adding a frontend using React or Next.js

---