At 11.07 PM UTC +1, a database used by our platform for the purpose of user authentication went into Recovery Mode, meaning it no longer responded to requests. Under normal circumstances, a failover node takes over the responsibilities of the primary database, however this did not occur which prolonged the outage. At 11:27 PM UTC +1, the primary database went back to normal operation as the automated recovery process ended, bringing our API back up. No data was lost during the outage, which lasted a total of 20 minutes. We are adding additional monitoring for these databases, and we are implementing a more reliable failover strategy. We sincerely apologise for the inconvenience caused by the outage.
Posted Nov 14, 2018 - 00:47 CET
This incident affected: Agent Interface, Chat Widget, and Public API.