The error causing the build failure is ye olde Rate Limit error from our calls to GraphQL, which, were supposed to have been fixed with the introduction of the throttling plugin.
I was unable to reproduce the same error locally. So, re-ran the action in GitHub with debugging enabled, to see if any new information gets logged when it errors out. However, the action completed successfully!
Looking back through action history, the only thing I can think of that might lead to this being a re-occurring issue is when multiple build actions are kicked off in a very short amount of time?? For example, on Aug 1st there were 4 failures in a row. But, there were multiple triggering commits from @tsadler1988 (2nd commit 2 minutes from the 1st) and @nysenthil (3x commits each 1 min apart).
Unfortunately, the older failures from 4 months back have expired logs so I can't see that Rate Limit errors were the issue with those or not.
With this as my working theory I manually re-ran the last 4 failed builds all together to see if I could reproduce it that way, and it worked! 2 of the builds that were both simultaneously making GraphQL calls hit the Rate Limit error and failed. If you want to try to reproduce this yourself, re-run the Publish To Website actions for:
The remaining builds failed due to code conflicts and never made it to the GraphQL phase.
Now I am reading up on GitHub actions, to see if there is any way to set them up to have multiple triggers queue and run the workflow sequentially rather than simultaneously.