What's the cause of this?

[christerdk]

I'm running my watches once a day, during the night. Every morning I see these kind of errors. It's (seemingly) completely random, which URL that fails like this. All pages go through the Chrome Playwright, however, without having set up any manual steps. So all in all pretty basic.

What can I do to fix this?

Docker Desktop
Windows 11, Ram 128GB
Change Detection v0.50.27

[dgtlmoon]

Not enough information i think, any proxies? what URL? how many concurrent checks?

can you atleast paste the full error?

does the machine go to idle then wakeup?

[christerdk]

• No proxies.
• Here's an example URL: https://www.steinberg.net/
• The error for that watch is: Exception: Page.goto: Target page, context or browser has been closed Call log: navigating to "https://www.steinberg.net/", waiting until "load"
• Not sure if there's an even more detailed log entry in the system.
• No idle / wakeup
• I dialed the workers down to 1 a while back to see if that would fix it.

[dgtlmoon]

hmmmm should work for sure, do you use the sockpuppetbrowser or selenium?

[christerdk]

I haven't set anything up apart from the Chrome Playwright container as part of the changedetection docker compose file. Curious to know how sockpuppetbrowser or selenium would fit into the setup?

Also, I think I was a bit too quick to answer your question about proxy. Some of my containers, including changedetection, are exposed to the outside behind nginx, so they have a nginx_customnetwork config section and a 10.x.x.x address as well. But it didn't cross my mind during my previous response. Currently the playwright is being referred to with - PLAYWRIGHT_DRIVER_URL=ws://playwright-chrome:3000

[dgtlmoon]

hmm its basically saying the chrome couldnt fetch it or something happened to the browser

attach the playwright/chrome container logs i think would be best (as .txt attach)

[christerdk]

I'd be happy to supply the logs, but I'd prefer doing it in private because of the URLs etc. How can I send it to you? Here are a few examples I see in the log:

Log for a failed page:

2025-11-08 02:08:50 navigating to "[REDACTED URL]", waiting until "load"
2025-11-08 02:08:50
2025-11-08 02:08:50 2025-11-07 19:08:50.630 | ERROR | changedetectionio.html_tools:extract_title:604 - Title extraction received unsupported data type: <class 'NoneType'>

A different URL:

2025-11-08 02:16:02 navigating to "[REDACTED URL]", waiting until "load"
2025-11-08 02:16:02
2025-11-08 02:16:02 2025-11-07 19:16:02.520 | ERROR | changedetectionio.html_tools:extract_title:604 - Title extraction received unsupported data type: <class 'NoneType'>
2025-11-08 02:21:02 2025-11-07 19:21:02.739 | ERROR | changedetectionio.async_update_worker:async_update_worker:292 - Worker 0 exception processing watch UUID: 8ee3fa19-e3f0-48d7-8d73-c83409f0da19
2025-11-08 02:21:02 2025-11-07 19:21:02.739 | ERROR | changedetectionio.async_update_worker:async_update_worker:293 - Page.goto: Target page, context or browser has been closed

For both I guess changedetection gets invalid data because something fails in the Chrome Playwrigt. Is this soup.find?

If I change back to Basic Fast plantext, after running a few times, things are looking better, but I do see a few of these:

Warning, no filters were found, no change detection ran - Did the page change layout? update your Visual Filter if necessary.

But there are filters definted and the layout hasn't changed.