Skip to content

Step 2: Understanding Metrics

Learn how Prometheus collects and stores metrics from your Django app

📊 What Are Metrics?

Metrics are numeric measurements that help you understand your application's behavior:

Metric Type Example What It Tells You
Counter django_http_requests_total Total number of requests
Histogram django_http_requests_latency_seconds Distribution of request times
Gauge node_memory_MemAvailable_bytes Current memory usage

🔍 How Metrics Work

Step 1: Django Exposes Metrics

Django automatically exposes metrics at /metrics:

# View Django metrics
curl http://localhost:9000/metrics | head -20

You'll see output like:

# HELP django_http_requests_total Total number of HTTP requests
# TYPE django_http_requests_total counter
django_http_requests_total{method="GET",status="200",view="/todo/"} 42.0

# HELP django_http_requests_latency_seconds HTTP request latency
# TYPE django_http_requests_latency_seconds histogram
django_http_requests_latency_seconds_bucket{le="0.005",method="GET"} 30.0

Step 2: Prometheus Scrapes Metrics

Prometheus automatically scrapes metrics every 15 seconds:

# prometheus/prometheus.yml
scrape_configs:
  - job_name: "django"
    static_configs:
      - targets: ["obs-django:9000"]
    metrics_path: /metrics

Step 3: Query Metrics in Prometheus

Open Prometheus UI: http://localhost:9090

Try these PromQL queries:

# Check if Django is up
up{job="django"}

# Request rate (requests per second)
rate(django_http_requests_total[5m])

# 95th percentile latency
histogram_quantile(0.95, rate(django_http_requests_latency_seconds_bucket[5m]))

# Database query rate
rate(django_db_execute_total[5m])

🎯 Key Metrics to Monitor

HTTP Metrics

Metric Description Query
django_http_requests_total Total requests sum(rate(django_http_requests_total[5m]))
django_http_requests_latency_seconds Request duration histogram_quantile(0.95, rate(...[5m]))
django_http_responses_total_by_status Response status codes sum by (status) (rate(...[5m]))

Database Metrics

Metric Description Query
django_db_query_duration_seconds Query duration histogram_quantile(0.95, rate(...[5m]))
django_db_errors_total DB errors rate(django_db_errors_total[5m])
django_db_execute_total Total queries rate(django_db_execute_total[5m])

System Metrics

Metric Description Query
node_cpu_seconds_total CPU usage 100 - (avg(rate(node_cpu_seconds_total{mode="idle"}[5m])) * 100)
node_memory_MemAvailable_bytes Available memory node_memory_MemAvailable_bytes
node_filesystem_avail_bytes Available disk node_filesystem_avail_bytes

🔬 Hands-On Exercises

Exercise 1: Generate Traffic

Create some requests to generate metrics:

# Make 10 requests to the home page
for i in {1..10}; do
  curl -s http://localhost/ > /dev/null
  echo "Request $i sent"
done

# Make requests to the todo page
for i in {1..5}; do
  curl -s http://localhost/todo/ > /dev/null
  echo "Todo request $i sent"
done

Exercise 2: Query Prometheus

Open http://localhost:9090 and run these queries:

# See total requests
sum(django_http_requests_total)

# See request rate over last 5 minutes
sum(rate(django_http_requests_total[5m]))

# See requests by status code
sum by (status) (rate(django_http_requests_total[5m]))

Exercise 3: Check Target Health

  1. Go to http://localhost:9090/targets
  2. Look at the "django" target
  3. Verify it shows "UP" status

📈 Visualizing in Grafana

  1. Open http://localhost:3000
  2. Login with admin/admin
  3. Click "Explore"
  4. Select "Prometheus" datasource
  5. Enter a query:
rate(django_http_requests_total[5m])
  1. Click "Run Query"
  2. You'll see a graph of request rates!

🔧 Troubleshooting

No metrics appearing

Check if Django is exposing metrics:

curl http://localhost:9000/metrics | head

Check if Prometheus can reach Django:

docker exec obs-prometheus wget -qO- http://obs-django:9000/metrics | head

Check Prometheus targets:

curl http://localhost:9090/api/v1/targets | jq '.data.activeTargets[] | select(.labels.job=="django")'
Prometheus shows target as DOWN

Verify containers are on same network:

docker network inspect django_app_observability_network | grep -A5 obs-django

Restart Prometheus:

docker compose -f django_app/docker-compose.yml restart obs-prometheus

📚 What You Learned

In this step, you learned:

  • ✅ How Django exposes metrics at /metrics
  • ✅ How Prometheus scrapes metrics every 15 seconds
  • ✅ How to query metrics with PromQL
  • ✅ Key metrics to monitor (HTTP, DB, System)
  • ✅ How to visualize metrics in Grafana

🎓 Next Steps

Now that you understand metrics, let's learn about logs:

Ready for Step 3? 👇

Step 3: Logs →