Puppetter Page.close() not freeing tab memory

I’m currently using puppeteer for web scraping using one browser instance on which i connect several node apps using the websocket endpoint… I run it on 50 websites every 15 minutes, with a process queue running 40 of them concurrently. But as time goes by, the memory utilization goes crazy, and the only solution i found was to restart the browser.

Steps to reproduce

Tell us about your environment:

• Puppeteer version: tip of the tree
• Platform / OS version: Ubuntu 16.04

What steps will reproduce the problem?

This is the code i use to initiate the chrome instance

const puppeteer = require("puppeteer");

(async () => {
  const browser = await puppeteer.launch({
    headless: true,
    args: ["--remote-debugging-address=0.0.0.0", "--remote-debugging-port=1234"]
  });
  const wsendpoint = await browser.wsEndpoint();
  console.log(wsendpoint);
})();

This is the code I use to connect to a remote chrome instance

const puppeteer = require("puppeteer");
const axios = require("axios");

const extract = async () => {
  const url = process.argv[2];
  const remote_ip = process.argv[3];
  const remote_port = process.argv[4];
  const json = await axios.get(`http://${remote_ip}:${remote_port}/json/version`);
  const wsEndpoint = json.data.webSocketDebuggerUrl;
  const browser = await puppeteer.connect({ browserWSEndpoint: wsEndpoint, ignoreHTTPSErrors: true });
  const page = await browser.newPage();
  await page.goto(url, { waitUntil: "domcontentloaded", timeout: 300000 });

  // Extract the results from the page
  const links = await page.evaluate(() => {
    // Function to check if a tag is visually visible on the page
    const visible = (styles, DOMRect) => {
      if (styles.getPropertyValue("display") == "none" || styles.getPropertyValue("visibility") == "hidden" || styles.getPropertyValue("opacity") == 0) {
        return false;
      } else if (DOMRect.x < 0 || DOMRect.y < 0 || DOMRect.height == 0 || DOMRect.width == 0) {
        return false;
      }
      return true;
    };

    const pageSize = { width: document.body.scrollWidth, height: document.body.scrollHeight };
    const tags = Array.from(document.querySelectorAll("a"));
    return tags.map(tag => {
      const styles = window.getComputedStyle(tag, null);
      const DOMRect = tag.getBoundingClientRect();
      const image = tag.getElementsByTagName("img");
      return {
        url: tag.href,
        fontSize: parseInt(styles.getPropertyValue("font-size")),
        position: { x: DOMRect.x, y: DOMRect.y, left: DOMRect.left, right: DOMRect.right, top: DOMRect.top, bottom: DOMRect.bottom, width: DOMRect.width, height: DOMRect.height },
        visible: visible(styles, DOMRect),
        image: image.length > 0,
        pageSize: pageSize
      };
    });
  });
  await page.close();
  await browser.disconnect();
  return links;
};

extract()
  .then(links => {
    console.log(JSON.stringify(links));
  })
  .catch(e => {
    console.log(e.message);
    process.exit(1);
  });

What is the expected result?

I would expect the chrome instance to be steady on Memory utilization as i close every tab i open using Page.close()

What happens instead?

The chrome instance Memory utilization is growing as time goes by, And i need to restart the full chrome process for it to goes down again.

It might be a chromium related issue, but i’m not sure, so i thought i would start posting it here.

Thank you.

[grantstephens]

I found that by doing await page.goto('about:blank') freed everything up before closing the tabs

[kireerik]

I think as of version 1.3.0 opening a blank page before closing tabs is no longer needed to fee up memory.

@grantstephens and @galvez can you confirm that?