Degraded Performance for Production

We noticed low success rates and slow responses around 14:19 UTC. Starting investigation.

Source of problem identified, traffic returning to normal. Monitoring the situation.

Traffic has remained steady for the last hour and a half. Closing this incident.

Earlier fix seems not to have addressed the problem. Seeing low success rates and slow responses again. Investigating.

Still monitoring cause of alert. Traffic returned to normal since about 20:05 UTC.

While we have not seen any recurrences of the slowness, we're going to leave this incident in the monitoring state overnight in case anything unexpected happens.