Smoke Testing

Purpose

Minimal confidence checks that the system is alive and functional after deployment. Not exhaustive — just enough to confirm nothing is fundamentally broken.

When to Use

• After azd deploy to Azure.
• After docker compose up --build.
• After container image updates.
• As post-deployment CI gate.

Flow

Step 1 — Backend health

# Local
curl -sf http://localhost:8001/healthz | jq .

# Azure (replace with deployed URL)
curl -sf https://<backend-url>/healthz | jq .

Pass: Returns {"status": "ok"} with HTTP 200.

Step 2 — Sync chat round-trip

curl -sf -X POST http://localhost:8001/api/v1/chat \
  -H "Content-Type: application/json" \
  -d '{"message": "smoke test"}' | jq .

Pass: Returns JSON with session_id, message_id, and non-empty response.

Step 3 — SSE streaming works

curl -sN -X POST http://localhost:8001/api/v1/chat/stream \
  -H "Content-Type: application/json" \
  -d '{"message": "smoke test"}' | head -20

Pass: Output contains event: message_start, event: delta, and event: done.

Step 4 — Frontend loads

# Local
curl -sf http://localhost:5173 | grep -i "<html"

# Azure
curl -sf https://<frontend-url> | grep -i "<html"

Pass: Returns HTML content.

Step 5 — Storage connectivity (if Cosmos enabled)

curl -sf -X POST http://localhost:8001/api/v1/chat \
  -H "Content-Type: application/json" \
  -d '{"message": "persistence check", "session_id": "smoke-test-session"}' | jq .session_id

Pass: Returns the same session_id sent in the request. No 500 errors in backend logs.

Step 6 — Telemetry emitting (optional)

# Check backend logs for structured output
docker compose logs backend 2>&1 | grep "chat_stream_complete"

Pass: Structured log entries appear with expected fields.

Automated Smoke Script

#!/usr/bin/env bash
set -euo pipefail

BASE_URL="${1:-http://localhost:8001}"
FRONTEND_URL="${2:-http://localhost:5173}"

echo "=== Smoke Tests ==="

echo -n "Health check... "
curl -sf "$BASE_URL/healthz" | grep -q '"ok"' && echo "PASS" || { echo "FAIL"; exit 1; }

echo -n "Chat endpoint... "
curl -sf -X POST "$BASE_URL/api/v1/chat" \
  -H "Content-Type: application/json" \
  -d '{"message": "smoke"}' | grep -q '"response"' && echo "PASS" || { echo "FAIL"; exit 1; }

echo -n "SSE streaming... "
STREAM=$(curl -sN -X POST "$BASE_URL/api/v1/chat/stream" \
  -H "Content-Type: application/json" \
  -d '{"message": "smoke"}' --max-time 10 2>/dev/null || true)
echo "$STREAM" | grep -q "event: done" && echo "PASS" || { echo "FAIL"; exit 1; }

echo -n "Frontend loads... "
curl -sf "$FRONTEND_URL" | grep -qi "html" && echo "PASS" || { echo "FAIL"; exit 1; }

echo "=== All Smoke Tests Passed ==="

Checklist

• Backend /healthz returns 200
• Chat endpoint returns valid response
• SSE stream emits all 4 event types
• Frontend serves HTML
• No 500 errors in backend logs
• Storage writes succeed (if Cosmos enabled)

Source: pablosalvador10/template-agentic-ai-apps-azure — distributed by TomeVault.