There was an observed increase in failures when attempting to join the room from the EU region between 1:20 PM IST and 2:00 PM IST on November 9, 2023.
This issue arose due to the primary cluster in the EU region being under maintenance, resulting in the temporary rerouting of traffic to a backup cluster.
Although originally planned as a zero-downtime activity, an oversight in our standard operating procedure led to an unplanned 30-minute downtime.
The critical oversight involved the configuration of the DNS record for a key endpoint.