Implement Metrics Collection - The Tracer | Build Distributed Systems

                      TASK
                    

Implementation

Metrics quantify system behavior: how many requests, how fast, how much memory. Three types cover nearly everything: counters (monotonically increasing), gauges (can go up or down), and histograms (distribution of values like request duration).

Implement a node that records and queries metrics:

// Counter: increment by 1 on each request
{ "type": "counter", "msg_id": 1,
  "name": "http_requests_total", "value": 1,
  "labels": {"method": "POST", "service": "api"} }
-> { "type": "metric_recorded", "in_reply_to": 1,
    "name": "http_requests_total", "value": 1 }

// Gauge: current heap memory in bytes
{ "type": "gauge", "msg_id": 2,
  "name": "memory_usage_bytes", "value": 1073741824,
  "labels": {"service": "api", "type": "heap"} }
-> { "type": "metric_recorded", "in_reply_to": 2,
    "name": "memory_usage_bytes", "value": 1073741824 }

// Histogram: record one request duration observation
{ "type": "histogram", "msg_id": 3,
  "name": "request_duration_ms", "value": 123,
  "labels": {"endpoint": "/api/users"} }
-> { "type": "metric_recorded", "in_reply_to": 3,
    "name": "request_duration_ms", "count": 1, "sum": 123 }

Sample Test Cases

Counter metricTimeout: 5000ms

Input

{
  "src": "service",
  "dest": "metrics",
  "body": {
    "type": "counter",
    "msg_id": 1,
    "name": "http_requests_total",
    "value": 1,
    "labels": {
      "method": "POST",
      "service": "api"
    }
  }
}

Expected Output

{"type": "metric_recorded", "in_reply_to": 1, "name": "http_requests_total", "value": 1}

Gauge metricTimeout: 5000ms

Input

{
  "src": "service",
  "dest": "metrics",
  "body": {
    "type": "gauge",
    "msg_id": 1,
    "name": "memory_usage_bytes",
    "value": 1073741824,
    "labels": {
      "service": "api",
      "type": "heap"
    }
  }
}

Expected Output

{"type": "metric_recorded", "in_reply_to": 1, "name": "memory_usage_bytes", "value": 1073741824}

Histogram metricTimeout: 5000ms

Input

{
  "src": "service",
  "dest": "metrics",
  "body": {
    "type": "histogram",
    "msg_id": 1,
    "name": "request_duration_ms",
    "value": 123,
    "labels": {
      "endpoint": "/api/users"
    }
  }
}

Expected Output

{"type": "metric_recorded", "in_reply_to": 1, "name": "request_duration_ms", "count": 1, "sum": 123}

Hints

Hint 1▾

Counter: only increases — use for request counts, error counts, bytes sent

Hint 2▾

Gauge: can go up or down — use for memory usage, queue depth, connection count

Hint 3▾

Histogram: records value distribution — use for request duration, response size

Hint 4▾

Labels are key-value pairs that add dimensions (filter by service, endpoint, status)

Hint 5▾

p95 = value at the 95th percentile: 95% of requests complete faster than this

Resources

Prometheus Metric Types

Counter, gauge, histogram, and summary metric types explained

                      OVERVIEW
                    

Theoretical Hub

Concept overview coming soon

Key Concepts

countergaugehistogramlabelspercentile

main.py

python