Screen recording with Puppeteer fails on Ubuntu: NotReadableError when using MediaRecorder API

Other Questions Puppeteer
1

I’m having trouble with screen recording on Ubuntu using Puppeteer and the MediaRecorder API. It works fine on Windows but not on Ubuntu. Here’s what I’m trying to do:

async function captureScreen() {
  const displayStream = await navigator.mediaDevices.getDisplayMedia({
    video: { width: 1920, height: 1080 },
  });

  const recorder = new MediaRecorder(displayStream);

  recorder.ondata = (e) => {
    // Handle video data
  };
}

When I run this on Ubuntu, I get a NotReadableError when trying to get the display media. I’ve tried a bunch of things:

  • Running the browser with --use-fake-ui-for-media-stream
  • Setting --auto-select-desktop-capture-source=Entire screen
  • Turning on AudioServiceOutOfProcess
  • Turning off sandbox stuff

None of these fixed it. The browser is running in non-headless mode. I just want to record the whole screen on Ubuntu like I can on Windows. Any ideas what might be causing this or how to fix it?

2

I’ve encountered this problem before when attempting screen recording on Ubuntu. In my experience, using x11vnc in combination with a virtual X server has helped resolve the NotReadableError. I started a virtual display using Xvfb and then launched x11vnc targeting that display. This setup allowed Puppeteer to connect to the virtual display properly. Additionally, launching the browser with flags such as --no-sandbox, --disable-setuid-sandbox, and --use-fake-ui-for-media-stream ensured the environment was compatible with the MediaRecorder API. This approach helped me bypass the recording error.

3

hey jackhero, i had similar issues on ubuntu. try using xvfb-run to create a virtual display. something like:

xvfb-run --server-args=‘-screen 0 1920x1080x24’ your_puppeteer_script.js

this worked for me when i couldnt get screen capture working normally. good luck!

4

I’ve dealt with this exact issue on Ubuntu before. What finally worked for me was using FFmpeg instead of the MediaRecorder API. It’s a bit more involved, but way more reliable on Linux systems. Here’s the gist:

  1. Install FFmpeg on your Ubuntu machine
  2. In your Puppeteer script, launch a new page and navigate to the desired URL
  3. Get the page dimensions
  4. Use child_process.spawn to start FFmpeg, capturing the screen
  5. Let it record for the desired duration
  6. Stop the FFmpeg process

The FFmpeg command I used was something like:

ffmpeg -f x11grab -s 1920x1080 -i :0.0 output.mp4

This bypasses the whole MediaRecorder headache and gives you a lot more control over the output. It took some trial and error, but it’s been rock-solid for me ever since. Hope this helps!