Notify - Distributed Notification Delivery Service (Go + gRPC + GraphQL + Prometheus)

This project simulates a miniature version of how social media platforms (like Instagram or Twitter) notify users when someone they follow creates a new post.

It uses:

• ✅ gRPC to simulate users publishing posts
• ✅ Background worker queue to dispatch notifications
• ✅ Retry logic for failed deliveries
• ✅ GraphQL API to fetch a user’s notifications
• ✅ Prometheus metrics to monitor performance and failures

---

📦 Project Features Overview

Feature	Description
gRPC API	Simulates publishing a post (PublishPost)
Notification Queue	In-memory delivery queue for notifications
Worker Pool	Concurrent workers simulate delivery
Retry Logic	Retry 3 times for failed notifications (10% simulated failures)
GraphQL API	Fetch recent notifications for a user
Prometheus Metrics	Monitor sent, failed notifications, and delivery times
In-Memory Store	All data (posts, followers, notifications) is stored temporarily in Go maps

---

⚙️ How to Run (Locally Without Docker)

✅ Step 1: Start the System

make run

• gRPC Server → localhost:5050
• GraphQL Server → localhost:8081
• Prometheus Metrics → localhost:8081/metrics

---

✅ Step 2: Publish a Post (via gRPC)

grpcurl -plaintext -d '{"postId":"p1","authorId":"user1","content":"Hello world!"}' localhost:5050 post.PostService/PublishPost

🔄 Internally:

• Finds all followers of user1 (from mock data)
• Creates and queues notifications
• Background workers pick from the queue and deliver them
• Logs show success or failure

---

✅ Step 3: Query Notifications (via GraphQL)

URL:

http://localhost:8081/query

Example query:

{
  notifications(userId: "user2")
}

✅ Output:

{
  "data": {
    "notifications": ["New post from user1: Hello world!"]
  }
}

---

📊 Step 4: Monitor with Prometheus (Manual)

Visit raw metrics endpoint:

http://localhost:8081/metrics

Metrics:

• notifications_sent_total
• notifications_failed_total
• notification_delivery_duration_seconds

---

🖥️ Optional: Prometheus UI (Manual Mode)

📥 Download Prometheus

From: https://prometheus.io/download/

📁 Create prometheus.yml

global:
  scrape_interval: 5s

scrape_configs:
  - job_name: 'notify_service'
    static_configs:
      - targets: ['localhost:8081']

▶️ Start Prometheus

./prometheus --config.file=prometheus.yml

Open Prometheus UI:

http://localhost:9090

---

🐳 Docker + Prometheus Monitoring

Run the whole system using Docker Compose.

---

✅ Step 1: Build and Run with Docker Compose

docker-compose up --build

Runs:

• 🟣 notify (Go app):
  • gRPC: localhost:5050
  • GraphQL + Metrics: localhost:8081
• 🟡 prometheus:
  • UI: localhost:9090

---

✅ Step 2: gRPC Call (Inside Docker)

grpcurl -plaintext -d '{"postId":"p1","authorId":"user1","content":"Hello from Docker!"}' localhost:5050 post.PostService/PublishPost

---

✅ Step 3: Query Notifications (GraphQL)

URL:

http://localhost:8081/query

Query:

{
  notifications(userId: "user2")
}

---

📊 Step 4: Prometheus Metrics

Open Prometheus UI:

http://localhost:9090

Query metrics like:

• notifications_sent_total
• notifications_failed_total
• notification_delivery_duration_seconds

💡 Scraping happens every 5s.

---

🧱 Docker File Structure

Dockerfile

FROM golang:1.23

WORKDIR /app
COPY . .

RUN go mod tidy
RUN go build -o notify main.go

EXPOSE 5050
EXPOSE 8081

CMD ["./notify"]

docker-compose.yml

services:
  notify:
    build: .
    ports:
      - "5050:5050"
      - "8081:8081"

  prometheus:
    image: prom/prometheus
    ports:
      - "9090:9090"
    volumes:
      - ./prome_ui/prometheus.yml:/etc/prometheus/prometheus.yml

prometheus.yml

global:
  scrape_interval: 5s

scrape_configs:
  - job_name: 'notify-app'
    static_configs:
      - targets: ['notify:8081']

---

🧰 Under the Hood – Explained Simply

🟣 gRPC (PublishPost)

rpc PublishPost(Post) returns (NotificationResponse);

Simulates user post publishing and returns how many notifications were queued.

---

🔁 Queue + Worker Dispatcher

• Adds notifications to a channel (queue)
• Worker goroutines consume the queue
• Each "delivers" (logs) a notification
• 10% chance of simulated failure → retried 3 times with backoff

---

🧠 In-Memory Data

Used maps to simulate DBs:

map[string][]string        // followers
map[string][]Notification  // user notifications

---

📈 Prometheus Metrics Tracked

Metric	Description
notifications_sent_total	# delivered
notifications_failed_total	# failed
notification_delivery_duration_seconds	Timing histogram

---