Error generating PDFs: Protocol error (Page.printToPDF): Target closed.

• Puppeteer version: puppeteer@1.5.0-next.1528948338383
• Chromium 66.0.3359.181 Built on Ubuntu , running on Ubuntu 17.10
• Node.js version: v9.11.2
• Fails on VirtualBox VM with plentiful resources (4cores, >8GB mem free)

Also an issue inside docker container with puppeteer 1.0.0, Ubuntu 18.04 and node v8.

What steps will reproduce the problem?

(async () => {
    const browser = await puppeteer.launch({
        executablePath: '/usr/bin/chromium-browser',
        args: ['--headless', '--disable-gpu', '--full-memory-crash-report', '--unlimited-storage',
               '--no-sandbox', '--disable-setuid-sandbox', '--disable-dev-shm-usage'],
    });
    const page = await browser.newPage();
    await page.goto(url, {waitUntil: 'networkidle0'});
    await page.pdf({path: outfile, format: 'A4'});
    await browser.close();
})().catch(error => {
    console.error(error);
    process.exit(1);
});

What is the expected result?

PDF file saved to outfile, exit code 0

What happens instead?

chromium-browser goes 100% CPU at page.pdf() and after 35s:

{ Error: Protocol error (Page.printToPDF): Target closed.
at Promise (/home/jon/dev/web-exporter/node_modules/puppeteer/lib/Connection.js:202:56)
at new Promise (<anonymous>)
at CDPSession.send (/home/jon/dev/web-exporter/node_modules/puppeteer/lib/Connection.js:201:12)
at Page.pdf (/home/jon/dev/web-exporter/node_modules/puppeteer/lib/Page.js:851:39)
at /home/jon/dev/web-exporter/web_exporter.js:40:20
at <anonymous> message: 'Protocol error (Page.printToPDF): Target closed.' }

Exit code of 1.

I’ve tried various different args for chromium-browser and page.pdf. This application has been working generating 100s of PDFs for 6 months. In the last few days however would fail with the output above on specific pages.

The page being loaded is a react application with a 10s of small images (50-200 pixels square) embedded. The issue reproducibley fails on small number of ‘bad’ pages. Unfortunately the URL is not public.

I can open the url without issue in chrome and the same chromium-browser (non-headless), also the print to pdf works as expected.

Any help in tracking down this issue much appreciated.

Jon