Nova Rewards — k6 Load Tests

Issue #638

Overview

k6 load test scripts for the three critical API endpoints:

Script	Endpoint
scenarios/webhook-actions.js	POST /api/webhooks/actions
scenarios/user-balance.js	GET /api/users/:id/balance
scenarios/campaigns-list.js	GET /api/campaigns
full-suite.js	All three concurrently

Load profile

• Ramp from 0 → 100 VUs over 5 minutes
• Sustain 100 VUs for 10 minutes
• Ramp down to 0 over 1 minute

SLA thresholds

Metric	Target
p95 response time	< 500 ms
Error rate	< 0.1 %

Prerequisites

1. k6 installed — https://k6.io/docs/get-started/installation/
2. Backend running locally or pointing to a staging environment
3. DB seeded with load-test fixtures (see below)

Quick start

# 1. Seed the DB and get credentials
npm run load:seed
# Follow the printed export instructions, e.g.:
export LOAD_TEST_USER_TOKEN="eyJ..."
export LOAD_TEST_MERCHANT_KEYS="load-test-merchant-key-1,load-test-merchant-key-2,load-test-merchant-key-3"

# 2. Start the backend (separate terminal)
npm run dev

# 3. Run individual scenarios
npm run load:webhook
npm run load:balance
npm run load:campaigns

# 4. Or run the full suite (all three concurrently, ~300 peak VUs)
npm run load:all

Environment variables

Variable	Default	Description
BASE_URL	http://localhost:3001	Backend base URL
WEBHOOK_SECRET	test_secret	INBOUND_WEBHOOK_SECRET value
USER_TOKEN	(required)	JWT for GET /users/:id/balance
USER_ID_MIN	1	Lower bound of user ID pool
USER_ID_MAX	500	Upper bound of user ID pool
MERCHANT_API_KEYS	test-api-key	Comma-separated merchant API keys

CI artifacts

The CI load-test job uploads all results to the load-test-results artifact (retained 30 days):

test-results/load/
├── webhook-actions-raw.json      # k6 raw JSON output
├── webhook-actions-summary.json  # handleSummary output
├── user-balance-raw.json
├── user-balance-summary.json
├── campaigns-list-raw.json
├── campaigns-list-summary.json
├── full-suite-raw.json
└── full-suite-summary.json

Bottleneck analysis

See BOTTLENECKS.md for identified performance bottlenecks and remediation recommendations for each endpoint.