This incident is resolved. We now restart any request that runs too long, moved to servers that don’t run out of capacity, and added alerts so we hear about slowdowns before you do. Sorry for the disruption over the weekend.

We restarted the affected servers and service is back. We’re watching error rates and response times.

Errors increased and most of the site became unavailable. One of our servers ran out of capacity after a stuck process, and the same thing spread to the others.

Superset is responding slowly, and some requests on the web and mobile apps are failing. We’re looking into it.