TL;DR

Triggering the regeneration of the blog in GitHub Pages through the API, for when the automatic way fails.

As I elaborated multiple times so far, this blog is hosted in GitHub Pages.

The workflow is usually quite smooth: you push something to the repository, GitHub notices the change and rebuilds the site. This allows using e.g. Jekyll to use Markdown instead of plain HTML.

Alas, things can go wrong from time to time, as it happened on July 13th:

Considering that busypub for me attempts publishing at 07:30 AM in Europe/Rome (i.e. 05:30 AM UTC at this time of the year), it’s easy to see that the blog was totally hit by the incident.

Some DuckDuckGoing around led me to this answer in StackOverflow: use the API! So by all means take a look at the answer for some instructions, and then you should be able to leverage the following script:

Semi-local version here.

The credentials you get are supposed to end up in a file located at ~/.github/ghp-rebuild, in the form of <username>:<token>, like polettix:90238102938deadbeef902320. Otherwise, set environment variable GHP_CREDENTIALS to a string with the same format.

The repository against which you want to trigger a rebuild must have a form like <username>/<reponame>, like polettix/ETOOBUSY for this very blog. Either set it through environment variable GHP_FQREPO (a.k.a. Fully Qualified Repo) or make sure that the origin value in your _config.yml file is set to something like https://github.com/<username>/<reponame> (and have teepee installed!).

Now you’re done: if GitHub Pages misses its beat, just call it and wait for the rebuild to happen!