Debugging silent failures of GitHub Pages

I just set up GitHub Pages for the first time for a repository, in what’s admittedly a nontrivial configuration (private but soon-to-be-public repo, doing most of the site generation with Sphinx and pointing Jekyll just at the outputs). But after pushing all of it to master, the result is silence – the page still 404’s, no errors show up in the repo settings, no status anywhere indicates anything.

All of the troubleshooting suggestions on the site are about handling Jekyll errors which surface, but when nothing surfaces, how does one go about debugging? Is there any way to see what Jekyll is trying to do during the build, what files it thinks it’s serving, what its _site directory looks like?

Refinement of the question: I’m about 90% sure at this point that in prod, Jekyll is running, but it’s not seeing the correct _config.yml file. As I’ve discovered, Jekyll handles being pointed at a nonexistent _config.yml by silently pretending that it sees a default file there and building whatever it would under those circumstances, which looks like what’s happening in my case right now. 

So perhaps another way to refine this is: Is there any way to know exactly what command GHP is running to build Jekyll (and in what directory it’s running it)? If you select to serve from master, or master/docs, how does that translate into what GHP will actually attempt to do?

(Running bundle exec jekyll build -V or bundle exec jekyll serve -V from my repo root directory on localhost works fine)

Hi - I’m also getting a 404. And I’ve trippple checked everything, and everything looks fine enough locally. Maybe there’s a problem with Github Pages now that Github actions has been launched?


Expected Website:

Hey so I just got it to work with what this post said to do:

You need to login to github as the user that owns the repo and make a commit. That will kick everything off.