[Pgpool-general] [pgpool-general] load balancing for dummies

[Tatsuo Ishii]

> Hi,
> 
> I'm looking at the load balancing and seems to me like is there a way
> to broke the sync of a replication using load balancing if a do
> something stupid like:
> 
> """
> create table t1(col1 date);
> 
> create table t_int(col1 int);
> insert into t_int select generate_series(1, 10);
> 
> create or replace function f1() returns integer as $$
>     insert into t1 values(current_date);
>     select 1;
> $$ language sql volatile;
> 
> select * from t_int where col1 = f1();
> """
> 
> the insert executed as a side effect in the function f1() could be
> sent to an slave instead of the master, something we can make to solve
> this?

The short answer is, don't do that.

Long answer:

f1() has two problems:

1) It has a side effect

2) It uses "stable" function (note that CURRENT_DATE() is a stable
   function, *not* a volatile function).

If f1() has a side effect and does use IMMUTABLE function only, using
/*REPLICATION*/ comment should solve the problem.
--
Tatsuo Ishii
SRA OSS, Inc. Japan