CentOS下源码安装部署PostgreSQL-XL

摘要: 本文主要介绍了postgresql一个分布式集群方案postgresql-xl的源码安装部署配置情况

环境
三台服务器，ip分别 10.0.1.12，10.0.1.13，10.0.1.14

hostname分别为gtm，dbnode01，dbnode02,其中gtm为master
即：物理节点，三台，分别为gtm,dbnode01,dbnode02

逻辑节点，七个，分别为gtm，coordinator01，datanode11，datanode12，
coordinator02，datanode21，datanode22

下载源码
http://sourceforge.net/projects/postgres-xl/files/Releases/Version_9.2rc/postgres-xl-v9.2-src.tar.gz/download 目前的版本是基于postgresql9.2版本修改的，9.5版本也有，但是还没有正式发布。

解压编译安装
tar xvf postgres-xl-v9.2-src.tar.gz
./configure
make&make install

增加postgres用户
useradd postgres
passwd postgres

创建数据目录
sudo mkdir /pgdata
chown -R postgres:postgres /pgdata

初始化节点
GTM节点（gtm）
initgtm -Z gtm -D /pgdata/
coordinator&datanode节点（dbnode01）
initdb -D /pgdata/coordinator --nodename coordinator01
initdb -D /pgdata/datanode11 --nodename datanode11
initdb -D /pgdata/datanode12 --nodename datanode12
coordinator&datanode节点（dbnode02）

initdb -D /pgdata/coordinator --nodename coordinator02
initdb -D /pgdata/datanode21 --nodename datanode21
initdb -D /pgdata/datanode22 --nodename datanode22
7. 配置文件修改
每个物理节点hosts文件修改

vim /etc/hosts
10.0.1.13 gtm

10.0.1.14 dbnode01

10.0.1.15 dbnode02

master节点（gtm）
vim /pgdata/gtm.conf

nodename = 'gtm'
listen_addresses = '*'
port = 10000
startup = ACT

coordinator01&datanode11/datanode12节点（dbnode01）
coordinator01

vim /pgdata/coordinator/postgresql.conf

listen_addresses = '0.0.0.0'
port = 10010
shared_buffers = 1000M
pooler_port = 11010
gtm_host = 'gtm'
gtm_port = 10000
pgxc_node_name = 'coordinator01'

datanode11

vim /pgdata/datanode11/postgresql.conf

listen_addresses = '0.0.0.0'
port = 10011
pooler_port = 11011
shared_buffers = 1000M
gtm_host = 'gtm'
gtm_port = 10000
pgxc_node_name = 'datanode11'

datanode12

vim /pgdata/datanode12/postgresql.conf

listen_addresses = '0.0.0.0'
port = 10012
pooler_port = 11012
shared_buffers = 1000M
gtm_host = 'gtm'
gtm_port = 10000
pgxc_node_name = 'datanode12'

三个节点的pg_hba.conf文件都增加

host   all              all             10.0.1.13/32          trust
host   all              all             10.0.1.14/32          trust
host   all              all             10.0.1.15/32          trust

coordinator02&datanode21/datanode22节点（dbnode02）
coordinator02

vim /pgdata/coordinator/postgresql.conf

listen_addresses = '0.0.0.0'
port = 10020
pooler_port = 11020
shared_buffers = 1000M
gtm_host = 'gtm'
gtm_port = 10000
pgxc_node_name = 'coordinator02'

datanode21

vim /pgdata/datanode11/postgresql.conf

listen_addresses = '0.0.0.0'
port = 10021
pooler_port = 11021
shared_buffers = 1000M
gtm_host = 'gtm'
gtm_port = 10000
pgxc_node_name = 'datanode21'

datanode22

vim /pgdata/datanode12/postgresql.conf

listen_addresses = '0.0.0.0'
port = 10022
pooler_port = 11022
shared_buffers = 1000M
gtm_host = 'gtm'
gtm_port = 10000
pgxc_node_name = 'datanode22'

三个节点的pg_hba.conf文件都增加

host   all              all             10.0.1.13/32          trust
host   all              all             10.0.1.14/32          trust
host   all              all             10.0.1.15/32          trust

8. 启动各个节点

gtm节点（gtm）

gtm_ctl -Z gtm start -D /pgdata/

coordinator01、datanode11、datanode12（dbnode01）

pg_ctl -Z datanode start -D /pgdata/datanode11
pg_ctl -Z datanode start -D /pgdata/datanode12
pg_ctl -Z coordinator start -D /pgdata/coordinator

coordinator02、datanode21、datanode22（dbnode02）

pg_ctl -Z datanode start -D /pgdata/datanode21
pg_ctl -Z datanode start -D /pgdata/datanode22
pg_ctl -Z coordinator start -D /pgdata/coordinator

启动完成后节点进程



9. 添加节点信息到数据库

dbnode01，coordinator01节点

psql -p 10010

ALTER NODE coordinator01 WITH (TYPE=coordinator,host=dbnode01,PORT = 10010);
CREATE NODE coordinator02 WITH (TYPE=coordinator,host=dbnode02,PORT = 10020);
CREATE NODE datanode11 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10011);
CREATE NODE datanode12 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10012);
CREATE NODE datanode21 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10021);
CREATE NODE datanode22 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10022);
SELECT pgxc_pool_reload();

dbnode01，datanode11节点

psql -p 10011

CREATE NODE coordinator01 WITH (TYPE=coordinator,host=dbnode01,PORT = 10010);
CREATE NODE coordinator02 WITH (TYPE=coordinator,host=dbnode02,PORT = 10020);
ALTER NODE datanode11 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10011);
CREATE NODE datanode12 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10012);
CREATE NODE datanode21 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10021);
CREATE NODE datanode22 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10022);
SELECT pgxc_pool_reload();

dbnode01，datanode12节点

psql -p 10012

CREATE NODE coordinator01 WITH (TYPE=coordinator,host=dbnode01,PORT = 10010);
CREATE NODE coordinator02 WITH (TYPE=coordinator,host=dbnode02,PORT = 10020);
ALTER NODE datanode11 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10011);
CREATE NODE datanode12 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10012);
CREATE NODE datanode21 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10021);
CREATE NODE datanode22 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10022);
SELECT pgxc_pool_reload();

dbnode02，coordinator02节点

psql -p 10020

CREATE NODE coordinator01 WITH (TYPE=coordinator,host=dbnode01,PORT = 10010);
ALTER NODE coordinator02 WITH (TYPE=coordinator,host=dbnode02,PORT = 10020);
CREATE NODE datanode11 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10011);
CREATE NODE datanode12 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10012);
CREATE NODE datanode21 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10021);
CREATE NODE datanode22 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10022);
SELECT pgxc_pool_reload();

dbnode02，datanode21节点

psql -p 10021

CREATE NODE coordinator01 WITH (TYPE=coordinator,host=dbnode01,PORT = 10010);
CREATE NODE coordinator02 WITH (TYPE=coordinator,host=dbnode02,PORT = 10020);
CREATE NODE datanode11 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10011);
CREATE NODE datanode12 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10012);
ALTER NODE datanode21 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10021);
CREATE NODE datanode22 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10022);
SELECT pgxc_pool_reload();

dbnode02，datanode22节点

psql -p 10022

CREATE NODE coordinator01 WITH (TYPE=coordinator,host=dbnode01,PORT = 10010);
CREATE NODE coordinator02 WITH (TYPE=coordinator,host=dbnode02,PORT = 10020);
CREATE NODE datanode11 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10011);
CREATE NODE datanode12 WITH (TYPE=datanode,host =dbnode01 ,PORT = 10012);
CREATE NODE datanode21 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10021);
ALTER NODE datanode22 WITH (TYPE=datanode,host =dbnode02 ,PORT = 10022);
SELECT pgxc_pool_reload();

10. 查看节点加载情况

select * from pgxc_node;



11. 简单测试验证

postgresql-xl只有coordinator有写入权限，datanode只有只读权限。

psql -p 10010

create table t1(f1 int);
create table t2(f1 int);
insert into t1 values(1);
insert into t2 values(2);

分别连接coordinator01，coordinator02，查看表和记录情况

minmin[postgres@postgres]> \dt



minmin[postgres@postgres]> select * from t1;



如果两个C节点都可以正常显示，那说明配置成功。

12. 完