Disaster Recovery

standby masters are good

We had a meeting last week where we were talking about Greenplum and we got on the topic of the standby masters. Which are really just sitting out there pulling transaction logs from the active master and hanging around. The consensus was that the need for the standby masters would be very few and far between and we haven’t needed them for the entire time we’ve been running GP thus far. That was last week. This week we’ve had two cases where we have needed the standby masters. One was a hardware issue with the 10G nic in the master of one cluster and the other was a table issue during recovery of another cluster. So yeah, the standby masters are needed.