Load testing suite for Tara API using k6.
Prerequisites#
Install k6#
bash
# macOS
brew install k6
# Ubuntu/Debian
sudo gpg -k
sudo gpg --no-default-keyring --keyring /usr/share/keyrings/k6-archive-keyring.gpg --keyserver hkp://keyserver.ubuntu.com:80 --recv-keys C5AD17C747E3415A3642D57D77C6C491D6AC1D69
echo "deb [signed-by=/usr/share/keyrings/k6-archive-keyring.gpg] https://dl.k6.io/deb stable main" | sudo tee /etc/apt/sources.list.d/k6.list
sudo apt-get update
sudo apt-get install k6
# Docker
docker pull grafana/k6
Running Tests#
Basic Run#
bash
# From the api directory
k6 run test/load/api-load.k6.js
With Custom Configuration#
bash
# Set API URL and test credentials
k6 run -e API_URL=http://localhost:3000 \
-e TEST_USER_EMAIL=test@example.com \
-e TEST_USER_PASSWORD=testpass123 \
test/load/api-load.k6.js
Run Specific Scenarios#
bash
# Run only smoke tests
k6 run --scenario smoke test/load/api-load.k6.js
# Run only load tests
k6 run --scenario load test/load/api-load.k6.js
# Run only stress tests
k6 run --scenario stress test/load/api-load.k6.js
# Run only soak tests
k6 run --scenario soak test/load/api-load.k6.js
With Docker#
bash
docker run -i grafana/k6 run - < test/load/api-load.k6.js
Test Scenarios#
Smoke Test (30s)#
- Basic functionality check
- 1 virtual user
- Verifies endpoints are working
Load Test (5m)#
- Typical expected load
- Ramps up to 50 concurrent users
- Simulates user journey (browse, search, view profile)
Stress Test (12m)#
- Beyond normal capacity
- Ramps up to 200 concurrent users
- Tests system limits and recovery
Soak Test (10m)#
- Sustained load over time
- 30 constant users
- Checks for memory leaks and degradation
Performance Thresholds#
| Metric | Threshold |
|---|---|
| p95 Response Time | < 500ms |
| p99 Response Time | < 1000ms |
| Error Rate | < 1% |
| Health Check p95 | < 100ms |
| Meditation Endpoints p95 | < 500ms |
| Search Endpoints p95 | < 800ms |
Outputs#
Console Output#
Summary is printed to console after test completion.
JSON Report#
Results are saved to load-test-summary.json.
InfluxDB (Optional)#
Send metrics to InfluxDB for Grafana visualization:
bash
k6 run --out influxdb=http://localhost:8086/k6 test/load/api-load.k6.js
CI Integration#
Add to GitHub Actions:
yaml
- name: Run Load Tests
run: |
docker run --rm -i \
-e API_URL=${{ secrets.API_URL }} \
-e TEST_USER_EMAIL=${{ secrets.TEST_USER_EMAIL }} \
-e TEST_USER_PASSWORD=${{ secrets.TEST_USER_PASSWORD }} \
grafana/k6 run - < apps/tara/api/test/load/api-load.k6.js
Interpreting Results#
Key Metrics#
- http_req_duration: Total request time
- http_req_blocked: Time spent blocked before sending
- http_req_connecting: Time spent establishing TCP connection
- http_req_sending: Time spent sending data
- http_req_waiting: Time spent waiting for response (TTFB)
- http_req_receiving: Time spent receiving data
Custom Metrics#
- error_rate: Percentage of failed checks
- authenticated_requests: Count of authenticated requests
- meditation_request_duration: Duration for meditation endpoints
- auth_request_duration: Duration for auth endpoints
- search_request_duration: Duration for search endpoints
Troubleshooting#
API Not Reachable#
Ensure the API is running and accessible at the configured URL.
Authentication Failures#
Check that test user credentials are correct and the user exists.
Rate Limiting#
If seeing 429 errors, the API rate limiter is working. Adjust VU count or add delays.