[pgpool-hackers: 458] Re: Multi statement query

[Bruce Momjian]

On Thu, Feb 13, 2014 at 07:32:30AM +0900, Tatsuo Ishii wrote:
> > On Tue, Feb 11, 2014 at 07:18:16PM +0900, Tatsuo Ishii wrote:
> >> Hi pgpool hackers,
> >> 
> >> Pgpool-II has a long standing restriction regarding "multi-statement
> >> query". A multi-statement query is a series of queries separated by
> >> ";". An example is "BEGIN;SELECT 1;SELECT 2;END". Currently pgpool-II
> >> sends such a query to primary node only (in streaming replication
> >> mode) or sends to certain node by just looking at the first query
> >> ("BEGIN" in the example).
> >> 
> >> 1) Analyze each query in a multi-statement query to find a query which
> >>    needs to be sent to all DB nodes. "BEGIN" or "END" are such query
> >>    in the example above.
> >> 
> >> 2) If such a query is found, sends to all DB nodes.
> >> 
> >> 3) Otherwise sends to the primary node.
> >> 
> >> This seems to work except following case:
> >> 
> >> SET SQL_inheritance TO off; INSERT INTO t1...;
> >> 
> >> "SET" needs be sent to all nodes but "INSERT" must only be sent to
> >> primary node. It seems it's impossible to satify those conditions at a
> >> same time. Probably it's users fault to create sunch a query...
> > 
> > Are you saying that "SET SQL_inheritance TO off; INSERT INTO t1...;"
> > comes to the server as a single string?
> 
> Yes, exactly.

I must be missing something.  Why does a BEGIN/END block or SET need to
be sent to all nodes?  I understand INSERT only should be sent to the
master.

-- 
  Bruce Momjian  <bruce at momjian.us>        http://momjian.us
  EnterpriseDB                             http://enterprisedb.com

  + Everyone has their own god. +