[Pgcluster-general] PGCluster Features and Use Cases

[At.Mitani]

Hi Mark,

> Q1. Could a selected DB participate in replication as a read-only
> slave but be excluded from load balancing? This would give DB that is
> intended purely for hot failover.

Yes,
You can set 0 to <Max_Connect> in the "pglb.conf" for target cluster db.

> Q2. Can PGCluster be configured to operate in Reliable Mode for one
> slave, and Normal Mode for a different slave?

Sorry, currentry, there is no difference between reliable mode and normal mode. All clusters were replicated with reliable mode. 

> Q3. Can PGCluster be configured to treat one subset of tables in a
> cluster as multi-master mode, and a different subset within that same
> cluster as read-only slaves?  This would allow multiple clusters to
> operate in multi-master mode on the first subset of tables, but
> restrict the second (mutually distinct) subset of tables to have a
> single master.

When you set partial replication setup to <Not_Replicate_Info> in "cluster.conf", it works as your want.
Partial replication should be specified each table.

> Q4. Does PBCluster automatically propagate schema changes?

PGCluster replicate the schema change query.
(i.e. ALTER...)

> Q5. What is the impact of Normal and Reliable Mode on write
> performance? I'm looking for some anecdotal or testimonial evidence
> here.

same answer as Q2.
there is no difference now.

Regards,
-- 
At.Mitani <mitani at sraw.co.jp>