search

hapostgres / pg_auto_failover

Postgres extension and service for automated failover and high-availability
Other
1.09k stars 114 forks source link

Monitor node high availability #877

Closed qiuwenhuifx closed 2 years ago

qiuwenhuifx commented 2 years ago

The current monitor node is a single node，if monitor node went down ,the cluster is in a non-highly available state， Will zk or etcd be used to monitor the status of the cluster in the future?

Eulerizeit commented 2 years ago

I hope not.

Currently in the minimal configuration, 1 Primary, 1 Mirror, 1 monitor if any one goes down the system is degraded but still available. The solution is to bring back up the missing component.

Adding etcd or zookeeper just increases the complexity of the system. It effectively requires you to run a database in parallel to a database for a database to work.

DimCitus commented 2 years ago

If the monitor node is down, the only impact is that no failover is possible. But if the monitor node is the only node down, no failover is needed. See also https://github.com/citusdata/pg_auto_failover/issues/867. Closing for duplicate.