Best Practices

info

Currently, Browserless V2 is available in production via two domains: production-sfo.browserless.io and production-lon.browserless.io

Aside from getting everything setup and running, there's a few best practices that'll help ensure your sessions are operating performantly. Below is a couple of highlights for running healthy headless scripts.

Make sure you close your session

When finished with work, or on errors, make sure you run browser.close so that other sessions may start. browserless will terminate long-running sessions via the timeout setting, but it's always a good idea to close tidly whenever you're finished.

import puppeteer from "puppeteer-core";

const TOKEN = "YOUR_API_TOKEN_HERE";

const browser = await puppeteer.connect({
  browserWSEndpoint: `wss://production-sfo.browserless.io?token=${TOKEN}`,
});

const page = await browser.newPage();

try {
  await page.goto("https://www.browserless.io/");
  await page.screenshot({ path: "./browserless.png" });
  browser.close();
} catch (error) {
  console.error({ error }, "Something happened!");
  browser.close();
}

Reduce await's as much as possible

Most of the puppeteer is async, meaning any command with await in front of it (or .then's) is going to make a round-trip from your application to browserless and back. While you can only do so much to limit this you should definitely try and do as much as possible. For instance, use page.evaluate over page.$selector as you can accomplish a lot in one evaluate versus multiple $selector calls.

DON'T DO

const $button = await page.$(".buy-now");
const buttonText = await $button.getProperty("innerText");
const clicked = await $button.click();

DO

const buttonText = await page.evaluate(() => {
  const $button = document.querySelector(".buy-now");
  const clicked = $button.click();

  return $button.innerText;
});

Use the GraphQL API to do pre-session checks

If you've been getting rejections or queued, be sure to use the GraphQL API to do pre-session checks. This allows your workers to remain up and healthy, allowing sessions to go smoothly.

# https://api.browserless.io/graphql
{
  pressure(apiToken: "YOUR_API_TOKEN_HERE") {
    running
    recentlyRejected
    queued
    isAvailable
    date
  }
}

This request will return a JSON object with the following payload:

// JSON returned
{
  "data": {
    "pressure": {
      "running": 0,
      "recentlyRejected": 0,
      "queued": 0,
      "isAvailable": true,
      "date": 1524762532204
    }
  }
}

You can use the isAvailable boolean to check to see if the instance can handle more connections. Otherwise you're free to write any sort of custom check with the provided JSON response.

Seen Improvements Yourself?

If you've seen improvements or have practices you think the community would benefit from, we'd love to hear from you!