Two unrelated changes deployed close together caused problems at the same time. One was a database configuration update and the other was a small TypeScript-related change in our front-end code that ended up breaking how the app loaded in the browser.

For about 15 minutes, people trying to log in or load the Buffer web app saw it fail to load. Anyone who already had the app open could keep working since the issue was on the front-end loading side, not the API. Separately, for about 30 minutes there was a delay in posts going out (up to 30 minutes late) and a delay in incoming comments showing up in Buffer for engagement. No data was lost in either case - just delayed.

We reverted the changes, and everything came back to normal quickly. We are now investing into improving our CI to catch issues like this earlier in the future.