At 07:30 on 2021-09-27, we observed an elevated number of failures in requests to our presence service for a 6 minute period, which may have resulted in some clients appearing to go offline and then rejoin. An infrastructure provider notified us of downgraded performance in a service we use to maintain presence state then quickly resolved the issue before we were able to perform mitigation steps. The Issue was resolved at 07:36 UTC on 2021-09-27.
To prevent a similar issue from occurring in the future we are working with our infrastructure provider to determine how this storage module can be better distributed to better protect us from failures in a single node or region.