export RUNTIME=compose

# Option 1: Quick start with pre-built images (recommended for first run)
COMPOSE_FILE="./infra/compose/docker-compose.ci.yml" make start

# Option 2: Build all services locally and start everything
make start

# Option 3: Start infrastructure first, then selective services
make start-infra
make start SERVICES="python-otel-service go-otel-service csharp-otel-service"
make start SERVICES="rust-otel-service cpp-otel-service"   # slow on first build

# Generate traffic + open Grafana
make traffic
make open-grafana   # http://localhost:3000 (admin/admin)

# Tear down (compose: down + prune)
make stop

export RUNTIME=k8s

# Terminal A — deploy
make start

# Terminal B — port-forward everything (observability + services)
make forward

# Terminal A — generate traffic + open Grafana
make traffic
make open-grafana   # http://localhost:3000 (admin/admin)

# Tear down
make stop

make open-grafana
# Navigate to: Explore → Jaeger → Select service

# Request rate (Go, C# services)
otel_http_server_request_duration_seconds_count
rate(otel_http_server_request_duration_seconds_count[5m]) # per-second request rate based on the otel_http_server_request_duration_seconds_count counter, averaged over the last 5 minutes

# Response size (Go service)
otel_http_server_response_body_size_bytes_count
rate(otel_http_server_response_body_size_bytes_count[5m]) # per-second increase of the response body size counter over the last 5 minutes

# 95th percentile latency (Go service)
histogram_quantile(0.95, rate(otel_http_server_request_duration_seconds_bucket[5m])) # histogram_quantile(0.95, rate(otel_http_server_request_duration_seconds_bucket[5m])) calculates the 95th percentile request duration over the last 5 minutes

# Browse all HTTP metrics
{__name__=~"otel_http.*"}

# Note: Not all services export the same metrics due to varying auto-instrumentation support:
# - Go: ✅ Full HTTP metrics (Gin auto-instrumentation)
# - C#: Partial (client metrics only, server metrics missing)
# - Python: Different metric names (use response_size instead of duration)
# - Rust: ❌ No HTTP metrics (Actix has no auto-instrumentation)
# - C++: Manual counter only
#
# TODO: Add manual HTTP metric instrumentation for Rust/C++ services.
# Refer to OpenTelemetry examples for your language:
# - Rust: https://github.com/open-telemetry/opentelemetry-rust/tree/main/examples
# - Python: https://opentelemetry.io/docs/languages/python/instrumentation/
# - C++: https://github.com/open-telemetry/opentelemetry-cpp/tree/main/examples

# In Grafana Explore → Loki, try:
{service_name="rust-service"}
{service_name=~".*-service"} |= "error"

OpenTelemetry Observability Stack PoC

  PROJECT_ROOT = /Users/marvingajek/Documents/poc-repos/otel-poc
  RUNTIME      = compose

Usage:
  make <target> [RUNTIME=compose|k8s] [SERVICES="svc1 svc2"]

  help                   Show available targets
  open-grafana           Open Grafana in browser
  open-jaeger            Open Jaeger in browser
  open-prometheus        Open Prometheus in browser
  lint                   Run pre-commit hooks
  start                  Start the platform (compose: SERVICES="svc1 svc2" optional)
  start-infra            Start only infrastructure (observability stack)
  stop                   Stop the platform
  restart                Restart the platform
  logs                   Follow platform logs
  build                  Rebuild service images (compose only; k8s rebuilds via deploy-to-kind.sh)
  status                 Show platform status
  traffic                Generate test traffic
  test                   Run service + telemetry tests
  forward                Port-forward everything (k8s only)
  forward-obs            Port-forward observability only (k8s only)
  forward-svc            Port-forward services only (k8s only)
  forward-bg             Background port-forward (k8s only; writes PID to /tmp/otel-pf.pid)
  forward-stop           Stop background port-forwards