From 1e916ee8f66dff44c108bd7642bef998a6caee29 Mon Sep 17 00:00:00 2001
From: User yamaguti <yamaguti at pgfoundry.org>
Date: Fri, 18 May 2007 02:45:05 +0000
Subject: [PATCH] added an explanation about /*REPLICATION*/ .

---
 README | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/README b/README
index 55ffbb8..e3e173d 100644
--- a/README
+++ b/README
@@ -23,8 +23,8 @@ pgpool version 3.2 README
 
    pgpool could be used as a replication server. This allows real-time
    backuping of the database to avoid disk failures. pgpool sends
-   exactly same query to each PostgreSQL servers to accomplish
-   replication. So pgpool can be regarded as a "synchronous
+   exactly the same updating queries to each PostgreSQL servers to
+   accomplish replication. So pgpool can be regarded as a "synchronous
    replication server".  Currently pgpool supports up to 2 PostgreSQL
    servers.
 
@@ -45,6 +45,17 @@ pgpool version 3.2 README
    the live server to the dead server. Once the DB contents match you
    start the two postmaster then restart pgpool.
 
+   While pgpool is in the replication mode, SELECT queries are sent to
+   the master node only. If the load balancing is enabled, SELECT's
+   are sent to randomly chosen node. If you need to replicate a SELECT
+   query which has a side effect of updaing some data, append the
+   following comment at the beginning of the query.
+
+      /*REPLICATION*/ SELCT ...
+
+   Note that queries that begin with either "SELECT nextval" or
+   "SELECT setval" are automatically replicated.
+
 1.2 Avoiding deadlocks
 
    pgpool could send a query to the "master" server then send to
-- 
2.39.5