Jekyll requires deploys to publish new content, and deploys were down for 76 hours. So, partial outage.

My point is if uptime is your goal you should go with a proven solution. Relying on an unproven custom setup to be 100% mistake free smacks a bit of young engineer hubris.