-
Notifications
You must be signed in to change notification settings - Fork 2
Home
BiSoft Information Technologies edited this page Oct 1, 2022
·
6 revisions
PostgreSQL BFM is a HA service that has automatic failover and switchover properties. A typical PostgreSQL Cluster has a master server and standby servers of the master server. In case of a failure on the master server, one of the standby servers should take the role of the master server by a failover mechanism. When the ex-master PostgreSQL service is online, it will rejoin to cluster as a standby server. In the BFM logic, the master server has a VIP (Virtual IP) and during the failover process, this VIP is transferred to the new master server.
VIP - Virtual IP
- Is active only on the master server
- All the clients are connected via this VIP
- Is transferred to the ex-standby new-master server during the failover or switchover process
BFM Instance
- Regularly checks all nodes in the cluster
- In case of any service failure, it coordinates any necessary operations on the database servers
- There should be only one active BFM instance, but there can be more than one BFM instance.
Minipg
- It performs any operating system level actions sent by the BFM such as
-
- Service start-stop operations
-
- Rewind operations
-
- VIP operations
- Runs a separate service on all PostgreSQL servers.