[pgpool-general: 7647] Re: Can PGPool manage multiple Postgres Clusters on every node?
ishii at sraoss.co.jp
Mon Jul 26 10:56:57 JST 2021
> we are setting up a 3 nodes clusters similar to this one:
> we are using CentOS 7.9 physical nodes with PGPool 4.2 and PostgreSQL 12.
> We need to have on every node 4 PostgreSQL instances (1 for prod, 1 for
> pre-production, 1 for test and 1 for dev), all of them must be in streaming
> replication with the other 2 nodes.
> Can PGPool manage such configuration? From the documentation we didn't find
> any reference.
> We are thinking to create 4 different pgpool instances, each one with its
> set of files and tcp port, is this a supported configuration?
The short answer is no.
Pgpool-II assumes that there's only one streaming replication primary
server exists among PostgreSQL instances. That means:
Suppose you have following PostgreSQL instances:
"A" group for "prod": A1, A2, A3
Pgpool-II can manage A1, A2 and A3, assuming that one of A1, A2,
A3 is the primary server and rest of it are standby servers.
"B" group for "pre-production": B1, B2, B3
Pgpool-II can manage B1, B2 and B3, assuming that one of B1, B2,
B3 is the primary server and rest of it are standby servers.
*BUT* a Pgpool-II instance cannot manage A group and B group altogeter
at the same time.
So it seems you have 4 groups of PostgreSQL instances (prod,
pre-production, test and dev), you can manage each group by assigning
4 Pgpool-II instances to each. In this case each Pgpool-II instance is
completely separated and will not communicate each other. I am not
sure what you can achieve in this configuration.
SRA OSS, Inc. Japan
More information about the pgpool-general