How to set up a second Greenplum single node instance as a replication db of the original master

[Paul Hopper]

We have been working with a single node instance of Greenplum 6.1 to replace our existing Postgre dB. It has been going well. I want to set up a second single node instance of Greenplum to be a replicated slave of the master.

 

From the documentation am I correct in thinking that this is covered in the "Dual Clusters" section of:

 

https://gpdb.docs.pivotal.io/6-1/best_practices/ha.html

 

My customers want me to look into pgpool to see if we can spread read requests across the master and slave single node GP instances.

 

We saw that pgbouncer binary is included in the GPDB rpm.

 

Is it possible to have more than one 'backup' GP cluster?

 

Cheers

 

Paul

[Peter O'Donovan]

Hi Paul,

 

There are a few different questions there, hopefully I can answer all.

It is not common to have an in sync replication setup between Greenplums like you might have with NAS storage systems. Greenplum does not have an automatic replicating system.

As per the documentation there are 2 methods used. The most common method we see is the backup and restore.

 

pgbouncer is used more for leaving connections open. In a busy system alot of time can be spend opening and closing connections to the DB, pgbouncer allows the user to have a constant connection without having to hold it from the users side.

 

You can have more than one backup cluster, but it would mean you have to restore to each cluster every time to remain in sync.

 

Does this answer everything? If no let me know.

 

Regards,

Peter