Mirror Segment Fail Over

[SeihaHoy]

Currently I am testing cloudberry on Virtual Machines with a Coordinator Node and 2 Segment nodes with each has 2 primary segments and 2 mirror segments. I applied default mirror configuration to the database. After that I simulated one segment node fails. I can see that the whole database stop working for a few minutes, then the database activates the mirror segments as the primary segments. What affects the fail over performance of switching between down segments to mirror segments? Thanks in advance!