puppeteer: page.evaluate() is leaking memory

Steps to reproduce

Tell us about your environment:

• Puppeteer version: 1.12.2
• Platform / OS version: Ubuntu 18.04.2
• URLs (if applicable): https://www.gofundme.com/cross-country-bike-trip
• Node.js version: 10.15.1

What steps will reproduce the problem?

I am using Puppeteer to scrape data from tens of thousands of URLs which are listed in a file. After a few hundred URLs, my program always crashes with this error:

Error: Failed to scrape URL https://www.gofundme.com/cross-country-bike-trip: TimeoutError: Navigation Timeout Exceeded: 30000ms exceeded at Promise.then (/var/www/scraper/node_modules/puppeteer/lib/LifecycleWatcher.js:143:21) -- ASYNC -- at Frame. (/var/www/scraper/node_modules/puppeteer/lib/helper.js:108:27) at Page.goto (/var/www/scraper/node_modules/puppeteer/lib/Page.js:662:49) at Page. (/var/www/scraper/node_modules/puppeteer/lib/helper.js:109:23) at scanURLs (/var/www/scraper/middleware/scan-urls.js:30:33) at process._tickCallback (internal/process/next_tick.js:68:7) at scanURLs (/var/www/scraper/middleware/scan-urls.js:205:13)

If I restart the program, it always picks up where it left off and scrapes the URL just fine… so I know it is not this particular URL that is the issue.

What is the expected result?

Since it scraped hundreds of URLs just fine, and there doesn’t appear to be anything special about this URL, and it scrapes this URL just fine when I restart the program, I would expect it to scrape this URL (and all other URLs) without any issues.

What happens instead?

The program crashes with the above error.

My code

It is very long, but I will include what I believe are the relevant parts here:

index.js:

require('dotenv').config()
const fs = require('fs')
const error = require('./middleware/error')
const mongoose = require('mongoose')
const puppeteer = require('puppeteer')
const Marker = require('./models/marker')
const scanSubmaps = require('./middleware/scan-submaps');

(async () => {

  // connect to the database
  await mongoose.connect(
    'mongodb://' + process.env.MONGO_USER_DEV + ':' + process.env.MONGO_PW_DEV + '@' + process.env.MONGO_IP_DEV + ':' + process.env.MONGO_PORT_DEV + '/' + process.env.MONGO_DB_DEV,
    {
      useNewUrlParser: true
    }
  )

  // initiate page
  let page = null
  const browser = await puppeteer.launch({
    headless: true,
    args: ['--full-memory-crash-report']
  })
  page = await browser.newPage()
  await page.setViewport({ width: 1366, height: 768 })
  await page.setRequestInterception(true)

  // don't download unnecessary resources
  page.on('request', (req) => {
    if(req.resourceType() === 'image' || req.resourceType() === 'stylesheet' || req.resourceType() === 'font'){
      req.abort()
    } else {
      req.continue()
    }
  })
  
  // get markers from database
  let markers = await Promise.all([
    Marker.findOne({ name: 'submap' })
      .select('value')
      .exec(),
    Marker.findOne({ name: 'url' })
      .select('value')
      .exec()
  ])
  let submapMarker = markers[0]
  let urlMarker = markers[1]

  // if no submap marker
  if (submapMarker == null) {

    // set submap marker to 0
    submapMarker = 0

  // else, set to its value
  } else {
    submapMarker = submapMarker.value
  }

  // if no URL marker
  if (urlMarker == null) {

    // set URL marker to 0
    urlMarker = 0

  // else, set to its value
  } else {
    urlMarker = urlMarker.value
  }

  // get list of submaps from sitemap
  let sitemap = fs.readFileSync('./sitemaps/sitemap.xml')
  let submapList = String(sitemap).match(/<loc>([^<]+)<\/loc>/g)

  // scan submaps (starting at submap marker)
  await scanSubmaps(page, submapList, submapMarker, urlMarker)

  // delete all markers
  await Marker.deleteMany({})
    .exec()

})().catch(err => {
  await mongoose.disconnect()
  if (typeof page !== 'undefined') {
    await page.close()
  }
  if (typeof browser !== 'undefined') {
    await browser.close()
  }
  error(err)
})

scan-submaps.js:

const Marker = require('../models/marker')
const fs = require('fs')
const scanURLs = require('./scan-urls')

module.exports = async function scanSubmaps(page, submapList, submapMarker, urlMarker) {

  // slice submapList
  submapList = submapList.slice(submapMarker)

  // for each submap
  for (const item of submapList) {

    // get submap name
    let submapName = item.match(/sitemap[0-9a-z_]+\.xml/)

    // get list of URLs from submap
    let submapContents = fs.readFileSync('./sitemaps/' + submapName)
    let urlList = String(submapContents).match(/<url>(.+?)<\/url>/g)

    // scan URLs (starting at URL marker)
    await scanURLs(page, urlList, urlMarker, submapMarker)

    // increment submap marker
    submapMarker++
    await Marker.findOneAndUpdate({ name: 'submap' }, { value: submapMarker }, { upsert: true })
      .exec()

    // reset URL marker
    urlMarker = 0
    await Marker.findOneAndUpdate({ name: 'url' }, { value: 0 })
      .exec()
  }

  // scan complete
  return 'Submaps scanned successfully.'
}

scan-urls.js:

const Entry = require('../models/entry')
const Marker = require('../models/marker')

module.exports = async function scanURLs(page, urlList, urlMarker, submapMarker) {

  // slice urlList
  urlList = urlList.slice(urlMarker)
  
  // for each URL
  for (const item of urlList) {

    // get URL
    let url = item.match(/<loc>([^<]+)<\/loc>/)[1]

    try {

      // load URL
      let response = await page.goto(url)

      // if redirect, skip URL
      let redirects = response.request().redirectChain()
      if (redirects.length > 0) {
        // skip URL
      } else {

        // scrape page content
        let scraped = await page.evaluate(() => {

          // SCRAPE A BUNCH OF DATA

          // return scraped data
          return {
            data1: data1,
            data2: data2,
            data3: data3,
            etc: etc
          }
        })

        let entry = {
          url: url,
          data1: scraped.data1,
          data2: scraped.data2,
          data3: scraped.data3,
          etc: scraped.etc
        }

        // add/update database entry
        await Entry.findOneAndUpdate(
          { url: url },
          entry,
          { upsert: true }
        )
          .exec()
      }

      // increment URL marker
      urlMarker++
      await Marker.findOneAndUpdate({ name: 'url' }, { value: urlMarker }, { upsert: true })
        .exec()

    // log URL if error
    } catch (err) {
      throw new Error('Failed to scrape URL ' + url + ': ' + err.stack)
    }
  }

  // scan complete
  return 'URLs scraped successfully.'
}