elastic/perf8

Your Tool For Python Performance Tracking

perf8

THIS IS ALPHA QUALITY, UNSUPPORTED, RUN AT YOUR OWN RISKS

Your Tool For Python Performance Tracking

perf8 is a curated list of tools to track the performance of your Python app.

The project is pluggable, and ships with a few tools:

• cprofile - a cProfile to Dot graph generator
• pyspy - a py-spy speedscope generator
• memray - a memory flamegraph generator
• psutil - a psutil integration
• asyncstats - stats on the asyncio eventloop usage (for async apps)

Installation

Use pip:

pip install perf8

If you use the cprofile plugin, you will need to install Graphviz to get the dot utility. See. https://graphviz.org/download/

Usage

Running the perf8 command against your Python module:

perf8 --all -c /my/python/script.py --option1

Will generate a self-contained HTML report, making it suitable for running it in automation and produce performance artifacts.

You can pick specific plugins. Run perf --help and use the ones you want.

Async applications

Running the asyncstats plugin requires to provide your application event loop.

In order to do this, you need to instrument your application to give perf8 the loop to watch. You can use the enable and disable coroutines:

import perf8

async def my_app():
     await perf8.enable(my_loop)
     try:
         # my code
         await run_app()
     finally:
         await perf8.disable()

To avoid running this code in production you can use the PERF8 environment variable to detect if perf8 is calling your app:

import os

if 'PERF8' in os.environ:
    import perf8

    async with perf8.measure():
        await run_app()
 else:
     await run_app()

Running and publishing using Github

You can use Github Actions and Github Pages to run and publish a performance report. In your project, add a .github/wokflows/perf.yml file with this content:

---
name: Perf report demo
on:
  workflow_dispatch:

permissions:
  contents: read
  pages: write
  id-token: write

jobs:
  build:
    runs-on: ubuntu-latest
    strategy:
      matrix:
        python-version:
          - '3.10'
      fail-fast: false
    name: Performance report using Python ${{ matrix.python-version }}
    steps:
      - uses: actions/checkout@v3

      - name: Setup Python
        uses: actions/setup-python@v4
        with:
          python-version: ${{ matrix.python-version }}

      - name: Install dot
        run: sudo apt-get install graphviz -y

      - name: Install perf8
        run: pip3 install perf8

      - name: Run Perf test
        run: perf8 --all

      - name: Setup GitHub Pages
        id: pages
        uses: actions/configure-pages@v1

      - name: Upload pages artifact
        uses: actions/upload-pages-artifact@v1
        with:
          path: /home/runner/work/perf8/perf8/perf8-report

      - name: Deploy to GitHub Pages
        id: deployment
        uses: actions/deploy-pages@v1

And extend perf8 --all so it runs your app. You will also need to go in the settings of your project and the Pages page to switch to GitHub Actions for the Build And Deployement source.

When the action runs, it will update https://<organization>.github.io/<project>

Screencast