We’re planning scheduled maintenance tomorrow starting at 01:00 GMT lasting no more than one hour. During this maintenance window our data center staff will reconfigure our primary switch in a virtual stack. This change requires that the switches be restarted. We’ll also use this opportunity to restart our PostgreSQL cluster to enable more autovacuum workers.
Thanks for your patience as we work to improve Bitbucket’s reliability.
In preparing for this scheduled maintenance window, we made a backup of the switch config over tftp, which caused the switch to become unresponsive. We did a hard reset which restored service, and we’re now postponing the maintenance window until we understand the problem in full. Our sincerest apologies for this unplanned downtime.