skills.rest
victorlou
victorlouCOMMUNITY
Explore all skills by this author →
GitHub

load-testing

Identifies Databricks App maximum sustainable QPS via Locust load tests on POST /invocations.

4|Updated May 9, 2026
Search Tags:#load testing#locust#sse streaming#databricks apps#performance benchmarking#qps#m2m oauth
One-click install
npx skills add https://github.com/victorlou/housing-assistant --skill load-testing-victorlou
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: load-testing
Source: https://github.com/victorlou/housing-assistant/tree/main/app/app-templates/.claude/skills/load-testing
Command: npx skills add https://github.com/victorlou/housing-assistant --skill load-testing-victorlou

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires locust, urllib3, requests.

What problem does it solve?

Load testing Databricks Apps to determine the maximum sustainable throughput and latency profile, so you can confidently size infrastructure and validate performance before production traffic.

Core Features & Use Cases

  • Provision repeatable Locust-based load tests: Drive concurrent streaming requests against /invocations and capture QPS, latency, TTFT, and failure rate.
  • Use M2M OAuth for long-running tests: Authenticate with service principal credentials to avoid token expiry mid-test.
  • Optionally isolate infrastructure vs. LLM latency: Mock agent streaming behavior with configurable chunk timing for capacity planning.
  • Generate an interactive results dashboard: Produce an HTML dashboard with KPI cards and ramp/latency charts to identify the saturation point.

Quick Start

Tell the assistant to “set up and run a load test for my Databricks App URL https://my-app.aws.databricksapps.com using medium compute, ramp up to 300 users, and generate a dashboard.”

Frequently Asked Questions about load-testing

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find the maximum QPS my Databricks App can handle?▼

You can find the maximum QPS by running controlled concurrent streaming load tests against the POST /invocations endpoint using Locust-driven ramp progression to identify the saturation point.

How does load testing work for SSE streaming responses?▼

Load testing SSE streaming responses involves driving concurrent streaming requests to capture QPS, latency, Time To First Token (TTFT), and failure rate while optionally mocking LLM chunk timing to isolate infrastructure capacity.

Do I need M2M OAuth to run long-running load tests against Databricks Apps?▼

Yes, you need M2M OAuth client_credentials token handling for long-running tests to authenticate with service principal credentials and avoid token expiry mid-test.

What's the best way to isolate infrastructure latency from LLM latency during capacity planning?▼

The best way to isolate infrastructure latency is to mock agent streaming behavior with configurable chunk timing, allowing you to measure pure infrastructure capacity without actual LLM processing delays.

Can I generate an interactive dashboard for throughput and latency analysis after a load test?▼

Yes, you can generate an interactive HTML results dashboard featuring KPI cards and ramp/latency charts to visualize throughput and latency analysis, with results persisted by run and configuration.

Why does my load test fail when token expires before the Databricks App ramp up completes?▼

Tokens expire mid-test when standard authentication is used, so you should use M2M OAuth client_credentials token handling to maintain authenticated concurrent streaming requests throughout the ramp progression.

Related Skills

Explore More Software Engineering Skills→
AnanyaDBJAnanyaDBJ
6

load-testing

Load test Databricks Apps to determine maximum QPS and generate interactive dashboards.

Community
Advanced
rm2thaddeusrm2thaddeus
4

dashboard-creator

Generate interactive HTML dashboards with KPI cards and charts.

Community
Intermediate
bcheng004bcheng004

load-testing

Run Locust-based load tests to measure Databricks App QPS limits.

Community
Advanced
keqingli1129keqingli1129

load-testing

Run Locust-based streaming load tests on Databricks Apps to measure QPS, TTFT, and latency.

Community
Advanced