Autonomous primary-mirror synchronized reset

Computer readable non-transitory storage medium for storing instructions for controlling supervisory processes of computers on primary and mirror segment nodes of a database to reset segments, comprising instructions for:

• detecting an event that requires a reset;
• sending exclusively by a primary segment of a primary-mirror pair of segments a reset request to the mirror segment of said pair upon said detecting of said event;
• and
• the mirror segment initiating a reset of itself solely in response to said reset request and informing the primary segment upon completing reset.

A method of handling faults on a replicated primary-mirror pair of primary and mirror segments of a distributed database, the method comprising:

• detecting by the primary segment of said primary-mirror pair a reset event;
• upon detecting said reset event, stopping all primary processes on said primary segment and sending exclusively by said primary segment a reset request to the mirror segment;
• checking by the primary segment whether the mirror segment was reset in response to said reset request;
• and
• upon determining by the primary segment that the mirror segment was reset, restarting said primary processes, otherwise notifying a fault tolerant service of said distributed database of a fault.