For about 4 minutes starting at 06:39am UTC a significant number of API-requests were rejected.
The root cause was an automatic database fail-over that some servers didn't pick up correctly, which meant they kept querying the old database which caused the 500 error codes.
At 04:25am UTC an automatic database failover was triggered. At 08:55am UTC we identified the cause of the issue and deployed a fix. At 09:06am UTC the fix was deployed successfully and we kept monitoring the results. At 09:10am UTC the issue was resolved.