If mirroring is enabled, Greenplum Database will automatically failover to a mirror segment when a primary segment goes down. As long as one segment instance is alive per portion of data, it will not typically be apparent to users of the system that a segment instance is down. If a transaction is in progress when a fault occurs, the in progress transaction will roll back, and then automatically restart on the reconfigured set of segments.
If the entire Greenplum Database system becomes unoperational due to a segment failure (for example if mirroring is not enabled or there are not enough segments alive to serve all the portions of user data), users will see errors when trying to connect to a database. The errors returned to the client program may give some indication of the failure. For example:
ERROR: All segment databases are unavailable
Enabling Alerts and Notifications
Greenplum Database administrators can enable email and/or SNMP alerts to be notified on system events such as segment failures.