Alright, let's break down how to tackle getting that 85k-member Facebook group list—since Chrome's struggling with the scale, we need smarter, more scalable approaches. Here's what I've seen work for similar large-scale tasks:

1. Switch to Headless Browser Automation (Playwright/Puppeteer) #

Chrome's full GUI mode chews through memory when loading tens of thousands of dynamic entries. Headless browsers (no visual window) are way more resource-efficient and built for this kind of automation.

• Playwright or Puppeteer let you automate incremental scrolling, wait for new members to load, and extract data without Chrome's GUI overhead.
• Avoid over-reliance on XPath—CSS selectors are faster and more reliable for dynamic Facebook content.
• Add 1-2 second delays between scrolls to avoid triggering Facebook's anti-scraping filters, and implement retry logic for failed requests.
• For example, with Playwright, you can write a loop that scrolls to the bottom, waits for the new member batch to load via waitForSelector, then extracts and saves the data before repeating.

2. Use the Facebook Graph API (Legitimate, Scalable Option) #

Scraping directly violates Facebook's Terms of Service, so if you can get proper access, the Graph API is the safest and most efficient route:

• You'll need a Facebook Developer account, a registered app, and the groups_member permission (submit for review if it's not for personal use).
• Use the endpoint /{group-id}/members with pagination: set limit=1000 per request, and use the after cursor from each response to fetch the next batch. This is designed for large datasets and won't crash like a browser.

3. Optimize Your UiPath Workflow for Scale #

If you want to stick with UiPath, here are tweaks to fix the Chrome crash issue:

• Enable headless Chrome: Add the argument --headless=new to your UiPath Chrome browser activity. This cuts memory usage drastically.
• Replace infinite scrolling with batch processing: Look for the "See More" button (or the scroll trigger that loads new members) and click it programmatically. Process each batch of members, save them to a CSV/Excel file, then move to the next batch—this prevents Chrome from loading all 85k members at once.
• Add memory cleanup: Every 5k-10k members, close and reopen the browser, then resume from where you left off (using your saved CSV as a progress marker).

4. Distributed Scraping (Last Resort, Use With Caution) #

If scraping is your only option (and you understand the TOS risks), split the workload across multiple instances:

• Tools like Playwright Cluster or Puppeteer Cluster let you run parallel scraping tasks, each handling a subset of members. This spreads memory load across processes/machines, avoiding single-browser crashes.
• Rotate user agents and use proxies to avoid getting blocked—Facebook detects aggressive scraping quickly.

Critical Reminders #

• Always review Facebook's Terms of Service before proceeding. Unauthorized scraping can lead to account bans or legal consequences.
• Save extracted data incrementally (append to a file after each batch) instead of storing everything in memory—this prevents your tool from running out of RAM.

内容的提问来源于stack exchange，提问作者Robert Fischer