The incident is now resolved.

Users continue to report they can access the application again and the issue is resolved. A hard refresh (Cmd+Shift+R on Mac, Ctrl+Shift+F5 on Windows) is helping to speed up the resolution of the issue on your device.

Users are reporting they can access the application again.

We are deploying the fix to remove the Frontend caching. The loading of the frontend might be a little bit slower than usual as a result. We're monitoring.

We purged the cached file at our CDN at 20:42 UTC and shortened its cache lifetime. The bad Javascript asset is still being served to affected customers.

We are now removing front-end caching from the application entirely and re-purging at the CDN. This eliminates the mechanism that caused the failure.

Login failures are caused by a stale JavaScript chunk cached at our CDN edge after a recent deploy. The chunk was cached with a success response, so affected browsers keep loading a broken bundle and cannot complete login. This is not an outage of our application or database — every other service is running normally.

Some users are reporting an inability to access a dashboard. The issue doesn't seem to be widespread. We're investigating.