tools

Load Test Dashboard

This replaces the one-off conversion workflow with a small local web app.

For the end-to-end load-test workflow, named test cases, and queue benchmark usage, see:

• tests/loadtest/README.md

Start the dashboard

python3 tests/loadtest/tools/dashboard.py

That starts a local server, opens the dashboard in your browser, and scans tests/loadtest/ for:

• k6 summary.json files
• raw k6 NDJSON exports
• CSV files produced by the old minute-bucket flow

Optional flags

python3 tests/loadtest/tools/dashboard.py --no-open
python3 tests/loadtest/tools/dashboard.py --port 9000
python3 tests/loadtest/tools/dashboard.py --root tests/loadtest

Queue drain benchmark

For temporary queue-drain benchmarking, generate a k6-shaped summary JSON from Asynq queue stats:

go run ./tests/loadtest/tools/queue_benchmark.go \
  -queue-prefix new:transaction_ \
  -wait \
  -out tests/loadtest/queue-summary.json

Or target one specific queue:

go run ./tests/loadtest/tools/queue_benchmark.go \
  -queues new:transaction_18 \
  -out tests/loadtest/queue-summary.json

If you are not loading the full app config, pass Redis directly:

go run ./tests/loadtest/tools/queue_benchmark.go \
  -redis-dsn "$BLNK_REDIS_DNS" \
  -queue-prefix new:transaction_ \
  -wait \
  -out tests/loadtest/queue-summary.json

Useful flags:

• -config <path>: load Blnk config from a specific file before reading env vars
• -interval 1s: polling cadence
• -timeout 15m: maximum drain benchmark duration
• -wait: wait for backlog to appear before starting measurement

What the UI supports

• Discovered load-test runs from disk
• Drag-and-drop upload of JSON, NDJSON, or CSV load-test files
• KPI cards for throughput, latency, success, Apdex, duration, and VUs
• Threshold and check summaries
• Simple trend charts when minute-level data exists