A series of overlapping updates flooded a specific write queue rendering it 'blocked' until the writes completed, meanwhile a failover caused all of these writes to retry all at once when the new instance came back online.
Posted Feb 03, 2021 - 08:30 PST
Database cluster health has returned to normal. We will continue to monitor this incident. Will post a more in-depth look at this when we have a more complete idea of the root cause.
Posted Feb 03, 2021 - 07:45 PST
Our database appears to have failed over two times back to back and is taking some time to fully warm a new node for the replica set
Posted Feb 03, 2021 - 07:23 PST
We have seen increased error rates in responding to requests to identify users, loading the Chameleon Builder etc
Posted Feb 03, 2021 - 07:19 PST
This incident affected: User profile API, Sidebar API, and Data API.