experimental/hardware_dashboard/README.md - buildbot - Git at Google

 # Hardware Dashboard - Perf Builders & Testers

 A modern, glassmorphic UI dashboard that seamlessly aggregates, visualizes, and monitors the status of Chromium performance testers and builders.

 ## Key Features Built

 1. **Starlark Triggers Mapping**:
    - Dynamically parses the `testing/buildbot/chromium.perf.star` Starlark file via Gitiles to perfectly map tester bots back to their parent CI builders.
 2. **GN Args Extraction**:
    - Fetches and parses `mb_config.pyl` via Python's AST evaluator, providing a neat togglable dropdown showing exactly what GN arguments each builder was compiled with.
 3. **Swarming API Integration**:
    - Parses `chromium.perf.json` to extract precise swarming dimensions for every tester.
    - Fetches active status and bot counts using `luci-auth` authenticated calls to Swarming's gRPC/PrPC endpoints (both `chrome-swarming` and `chromeos-swarming`).
    - Uses an `AsyncQueue` to throttle requests and avoid 429 Rate Limiting from Swarming APIs.
    - Safely de-duplicates identical bots that are tagged as both Dead and Quarantined.
 4. **Benchmarks Mapping (In-Memory Tarball Extraction)**:
    - Fetches the entire `tools/perf/core/schedule` Gitiles tarball into memory.
    - Parses the benchmark CSV files without writing them to disk to accurately determine which benchmarks run on which tester.
    - Formats benchmarks visually into clean wrap-around grid badges.
 5. **Real-time Client-Side Search**:
    - A highly responsive top-right search bar that instantly filters out the DOM.
    - Searching matches against builder names, tester names, mapped dimensions, and all benchmark titles.
 6. **One-Click Buganizer Escalation**:
    - For any tester that has Quarantined or Dead bots, a red "File Bug" button dynamically appears.
    - The button generates a massive URL payload that pre-fills a Buganizer `CUSTOMER_ISSUE` template with the bot hostnames, swarming pool, component `1735976`, auto-CC'd mailing lists, and priority statuses.
 7. **Graceful Background Refresh**:
    - Every 5 minutes, the dashboard automatically syncs all configs and APIs in the background.
    - Shows a non-intrusive floating overlay and preserves any rows you've expanded and any search filters you're actively typing.
    - Loading checklists display elapsed millisecond time durations for every API phase.

 ## Prerequisites

 - Python 3.7+
 - `luci-auth` installed and authenticated on your PATH (used for Swarming RPC).
 - Make sure to clear your browser cache if you modify `app.js` locally.

 ## Usage

 Start the backend server:

 ```bash
 # Recommended: Provide your local Chromium checkout directory for 10x faster startup
 python3 server.py --chromium-dir /usr/local/google/home/jeffyoon/Projects/chromium/src

 # Fallback: Start without directory (will fetch all files remotely via Gitiles)
 python3 server.py
 ```

 Then simply open your browser and navigate to:
 [http://localhost:8000](http://localhost:8000)

 ## Technical Architecture

 - **Backend (`server.py`)**: A lightweight `http.server.SimpleHTTPRequestHandler` with custom `/api/...` routes acting as a proxy and scraper to sidestep CORS limits.
 - **Frontend (`app.js`, `index.html`, `styles.css`)**: Vanilla JavaScript manipulating a flexbox-heavy, dynamically generated HTML table with no external dependencies (No React/Vue required).
	# Hardware Dashboard - Perf Builders & Testers

	A modern, glassmorphic UI dashboard that seamlessly aggregates, visualizes, and monitors the status of Chromium performance testers and builders.

	## Key Features Built

	1. Starlark Triggers Mapping:
	- Dynamically parses the `testing/buildbot/chromium.perf.star` Starlark file via Gitiles to perfectly map tester bots back to their parent CI builders.
	2. GN Args Extraction:
	- Fetches and parses `mb_config.pyl` via Python's AST evaluator, providing a neat togglable dropdown showing exactly what GN arguments each builder was compiled with.
	3. Swarming API Integration:
	- Parses `chromium.perf.json` to extract precise swarming dimensions for every tester.
	- Fetches active status and bot counts using `luci-auth` authenticated calls to Swarming's gRPC/PrPC endpoints (both `chrome-swarming` and `chromeos-swarming`).
	- Uses an `AsyncQueue` to throttle requests and avoid 429 Rate Limiting from Swarming APIs.
	- Safely de-duplicates identical bots that are tagged as both Dead and Quarantined.
	4. Benchmarks Mapping (In-Memory Tarball Extraction):
	- Fetches the entire `tools/perf/core/schedule` Gitiles tarball into memory.
	- Parses the benchmark CSV files without writing them to disk to accurately determine which benchmarks run on which tester.
	- Formats benchmarks visually into clean wrap-around grid badges.
	5. Real-time Client-Side Search:
	- A highly responsive top-right search bar that instantly filters out the DOM.
	- Searching matches against builder names, tester names, mapped dimensions, and all benchmark titles.
	6. One-Click Buganizer Escalation:
	- For any tester that has Quarantined or Dead bots, a red "File Bug" button dynamically appears.
	- The button generates a massive URL payload that pre-fills a Buganizer `CUSTOMER_ISSUE` template with the bot hostnames, swarming pool, component `1735976`, auto-CC'd mailing lists, and priority statuses.
	7. Graceful Background Refresh:
	- Every 5 minutes, the dashboard automatically syncs all configs and APIs in the background.
	- Shows a non-intrusive floating overlay and preserves any rows you've expanded and any search filters you're actively typing.
	- Loading checklists display elapsed millisecond time durations for every API phase.

	## Prerequisites

	- Python 3.7+
	- `luci-auth` installed and authenticated on your PATH (used for Swarming RPC).
	- Make sure to clear your browser cache if you modify `app.js` locally.

	## Usage

	Start the backend server:

	```bash
	# Recommended: Provide your local Chromium checkout directory for 10x faster startup
	python3 server.py --chromium-dir /usr/local/google/home/jeffyoon/Projects/chromium/src

	# Fallback: Start without directory (will fetch all files remotely via Gitiles)
	python3 server.py
	```

	Then simply open your browser and navigate to:
	[http://localhost:8000](http://localhost:8000)

	## Technical Architecture

	- Backend (`server.py`): A lightweight `http.server.SimpleHTTPRequestHandler` with custom `/api/...` routes acting as a proxy and scraper to sidestep CORS limits.
	- Frontend (`app.js`, `index.html`, `styles.css`): Vanilla JavaScript manipulating a flexbox-heavy, dynamically generated HTML table with no external dependencies (No React/Vue required).