pg-xl 的基本方式安装与使用

os: centos 7.4
pgxl:pg.version ‘10.3 (Postgres-XL 10alpha2)

pgxl 是一款非常实用的横向扩展的开源软件，继承了很多pgxc的功能，在replication 和sharding 方面有着非常棒的用处。
pgxl 不严格的说是 pgxc的升级加强版。是对官方 postgresql 的版本的修改提升，为大牛点赞。

Global Transaction Monitor (GTM)
全局事务管理器，确保群集范围内的事务一致性。 GTM负责发放事务ID和快照作为其多版本并发控制的一部分。
集群可选地配置一个备用GTM，以改进可用性。此外，可以在协调器间配置代理GTM，可用于改善可扩展性，减少GTM的通信量。

GTM Standby
GTM的备节点，在pgxc,pgxl中，GTM控制所有的全局事务分配，如果出现问题，就会导致整个集群不可用，为了增加可用性，增加该备用节点。当GTM出现问题时，GTM Standby可以升级为GTM，保证集群正常工作。

GTM-Proxy
GTM需要与所有的Coordinators通信重点内容，为了降低压力，可以在每个Coordinator机器上部署一个GTM-Proxy。

Coordinator
协调员管理用户会话，并与GTM和数据节点进行交互。协调员解析，并计划查询，并给语句中的每一个组件发送下一个序列化的全局性计划。
为节省机器，通常此服务和数据节点部署在一起。

Data Node
数据节点是数据实际存储的地方。数据的分布可以由DBA来配置。为了提高可用性，可以配置数据节点的热备以便进行故障转移准备。

总结：
gtm是负责ACID的，保证分布式数据库全局事务一致性。得益于此，就算数据节点是分布的，但是你在主节点操作增删改查事务时，就如同只操作一个数据库一样简单。
Coordinator是调度的，将操作指令发送到各个数据节点。
datanodes是数据节点，分布式存储数据。

规划如下：
node1 192.168.56.101 gtm,gtm-proxy,coordinator1,coordinator2

node2 192.168.56.102 datanode
node3 192.168.56.103 datanode

在网上了解是都是把 coordinator 和 datanode放在一起，本次实验没有放在一起，运行结果完全ok。

下载

https://www.postgres-xl.org/download/

https://git.postgresql.org/gitweb/?p=postgres-xl.git;a=summary
git://git.postgresql.org/git/postgres-xl.git

安装

node1 需要安装依赖包

# yum install -y bison flex perl-ExtUtils-Embed readline-devel zlib-devel pam-devel libxml2-devel libxslt-devel openldap-devel python-devel gcc gcc-c++ openssl-devel cmake openjade docbook-style-dsssl uuid uuid-devel

关闭防火墙，selinux，其余linux性能设置请参考postgresql的设置

# systemctl stop firewalld.service 
# systemctl disable firewalld.service 
# vim /etc/selinux/config
disabled

node1 节点上创建用户

# groupadd postgres
# useradd postgres -g postgres 
# passwd postgres

# mkdir -p /usr/pgxl-10
# chown -R postgres:postgres /usr/pgxl-10

# mkdir -p /var/lib/pgxl
# mkdir -p /var/lib/pgxl/coordinator1
# mkdir -p /var/lib/pgxl/coordinator2
# mkdir -p /var/lib/pgxl/gtm
# mkdir -p /var/lib/pgxl/gtm_standby
# mkdir -p /var/lib/pgxl/gtm_proxy1
# mkdir -p /var/lib/pgxl/gtm_proxy2
# chown -R postgres:postgres /var/lib/pgxl

node1 节点 postgres 用户的环境变量

# su - postgres
$ vi ~/.bash_profile
export PGUSER=postgres
export PGHOME=/usr/pgxl-10

export LD_LIBRARY_PATH=$PGHOME/lib
export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/lib:/usr/lib:/usr/local/lib
export PATH=$PGHOME/bin:$PATH

export TEMP=/tmp
export TMPDIR=/tmp

node1 上编译安装,注意不需要安装社区版

$ cd /tmp
$ git clone git://git.postgresql.org/git/postgres-xl.git
$ cd postgres-xl
$ git branch -r
  origin/HEAD -> origin/master
  origin/XL9_5_STABLE
  origin/XL_10_STABLE
  origin/master
  origin/xl_dbt3_expt
  origin/xl_doc_update
  origin/xl_test
$ git checkout XL_10_STABLE
Branch XL_10_STABLE set up to track remote branch XL_10_STABLE from origin.
Switched to a new branch 'XL_10_STABLE'
$ git status
# On branch XL_10_STABLE
nothing to commit, working directory clean  
$ ./configure --prefix=/usr/pgxl-10 --with-perl --with-python --with-openssl --with-pam --with-ldap --with-libxml --with-libxslt
$ make world
$ make install-world

$ cd contrib
$ make 
$ make install

node1、node2、node3配置ssh相互免密登录

过程略

node1、node2、node3同步下时间

# ntpdate asia.pool.ntp.org

node1节点上 gtm

$ which initgtm
/usr/pgxl-10/bin/initgtm

初始化 gtm

$ initgtm -Z gtm -D /var/lib/pgxl/gtm

The files belonging to this GTM system will be owned by user "postgres".
This user must also own the server process.


fixing permissions on existing directory /var/lib/pgxl/gtm ... ok
creating configuration files ... ok
creating control file ... ok

Success.
You can now start the GTM server using:

    gtm -D /var/lib/pgxl/gtm
or
    gtm_ctl -Z gtm -D /var/lib/pgxl/gtm -l logfile start

gtm.conf 的详细内容

$ vi /var/lib/pgxl/gtm/gtm.conf

nodename = 'node1_gtm'
listen_addresses = '*'
port = 6666
startup = ACT
log_file = 'gtm.log'
log_min_messages = NOTICE

node1节点上 gtm_proxy

由于node1上计划安装两个 coordinator，所以也需要安装两个 gtm_proxy

$ which initgtm
/usr/pgxl-10/bin/initgtm

初始化第一个 gtm_proxy

$ initgtm -Z gtm_proxy -D /var/lib/pgxl/gtm_proxy1
The files belonging to this GTM system will be owned by user "postgres".
This user must also own the server process.


fixing permissions on existing directory /var/lib/pgxl/gtm_proxy1 ... ok
creating configuration files ... ok

Success.
You can now start the GTM proxy server using:

    gtm_proxy -D /var/lib/pgxl/gtm_proxy1
or
    gtm_ctl -Z gtm_proxy -D /var/lib/pgxl/gtm_proxy1 -l logfile start

第一个 gtm_proxy 参数文件

$ vi /var/lib/pgxl/gtm_proxy1/gtm_proxy.conf

nodename = 'node1_gtm_proxy1'
listen_addresses = '*'
port = 6668
gtm_host = 'node1'
gtm_port = 6666
log_file = 'gtm_proxy.log'
log_min_messages = NOTICE

初始化第二个 gtm_proxy

$ initgtm -Z gtm_proxy -D /var/lib/pgxl/gtm_proxy2
The files belonging to this GTM system will be owned by user "postgres".
This user must also own the server process.


fixing permissions on existing directory /var/lib/pgxl/gtm_proxy2 ... ok
creating configuration files ... ok

Success.
You can now start the GTM proxy server using:

    gtm_proxy -D /var/lib/pgxl/gtm_proxy2
or
    gtm_ctl -Z gtm_proxy -D /var/lib/pgxl/gtm_proxy2 -l logfile start

第二个 gtm_proxy 参数文件

$ vi /var/lib/pgxl/gtm_proxy2/gtm_proxy.conf

nodename = 'node1_gtm_proxy2'
listen_addresses = '*'
port = 6669
gtm_host = 'node1'
gtm_port = 6666
log_file = 'gtm_proxy.log'
log_min_messages = NOTICE

node1节点上 coordinator

两个 coordinator

初始化第一个 coordinator

$ initdb -D /var/lib/pgxl/coordinator1 --nodename coordinator1 -E UTF8 --locale=C -U postgres -W

Success.

Success. You can now start the database server of the Postgres-XL coordinator using:

    pg_ctl -D /var/lib/pgxl/coordinator1 -l logfile start -Z coordinator

or
 You can now start the database server of the Postgres-XL datanode using:

    pg_ctl -D /var/lib/pgxl/coordinator1 -l logfile start -Z datanode

第一个 coordinator 参数文件

$ vi /var/lib/pgxl/coordinator1/pg_hba.conf

host    all             all             192.168.56.101/32         trust
host    all             all             192.168.56.102/32         trust
host    all             all             192.168.56.103/32         trust

$ vi /var/lib/pgxl/coordinator1/postgresql.conf

listen_addresses = '*'
port = 5432
max_connections = 100

pooler_port = 6670
max_pool_size = 100

gtm_host = 'node1'
gtm_port = 6668
pgxc_node_name = 'coordinator1'

初始化第二个 coordinator

$ initdb -D /var/lib/pgxl/coordinator2 --nodename coordinator2 -E UTF8 --locale=C -U postgres -W

Success.

Success. You can now start the database server of the Postgres-XL coordinator using:

    pg_ctl -D /var/lib/pgxl/coordinator2 -l logfile start -Z coordinator

or
 You can now start the database server of the Postgres-XL datanode using:

    pg_ctl -D /var/lib/pgxl/coordinator2 -l logfile start -Z datanode

第二个 coordinator 参数文件

$ vi /var/lib/pgxl/coordinator2/pg_hba.conf

host    all             all             192.168.56.101/32         trust
host    all             all             192.168.56.102/32         trust
host    all             all             192.168.56.103/32         trust

$ vi /var/lib/pgxl/coordinator2/postgresql.conf

listen_addresses = '*'
port = 5433
max_connections = 100

pooler_port = 6671
max_pool_size = 100

gtm_host = 'node1'
gtm_port = 6669
pgxc_node_name = 'coordinator2'

node2节点上 datanode

安装 pgxl，可参考node1节点上的安装过程

$ initdb -D /var/lib/pgxl/data --nodename datanode1 -E UTF8 --locale=C -U postgres -W
Success. You can now start the database server of the Postgres-XL coordinator using:

    pg_ctl -D /var/lib/pgxl/data -l logfile start -Z coordinator

or
 You can now start the database server of the Postgres-XL datanode using:

    pg_ctl -D /var/lib/pgxl/data -l logfile start -Z datanode

$ vi /var/lib/pgxl/data/pg_hba.conf

host    all             all             192.168.56.101/32         trust
host    all             all             192.168.56.102/32         trust
host    all             all             192.168.56.103/32         trust


$ vi /var/lib/pgxl/data/postgresql.conf

listen_addresses = '*'
port = 5432
max_connections = 100

pooler_port = 6667
max_pool_size = 100

gtm_host = 'node1'
gtm_port = 6668
pgxc_node_name = 'datanode1'

node3节点上 datanode

安装 pgxl，可参考node1节点上的安装过程

$ initdb -D /var/lib/pgxl/data --nodename datanode2 -E UTF8 --locale=C -U postgres -W
Success. You can now start the database server of the Postgres-XL coordinator using:

    pg_ctl -D /var/lib/pgxl/data -l logfile start -Z coordinator

or
 You can now start the database server of the Postgres-XL datanode using:

    pg_ctl -D /var/lib/pgxl/data -l logfile start -Z datanode

$ vi /var/lib/pgxl/data/pg_hba.conf

host    all             all             192.168.56.101/32         trust
host    all             all             192.168.56.102/32         trust
host    all             all             192.168.56.103/32         trust

$ vi /var/lib/pgxl/data/postgresql.conf    

listen_addresses = '*'
port = 5432
max_connections = 100

pooler_port = 6667
max_pool_size = 100

gtm_host = 'node1'
gtm_port = 6669
pgxc_node_name = 'datanode2'

pg-xl集群启动

启动顺序是 GTM=>GTM Standby=>GTM-Proxy=>Datanodes=>Coordinators

node1 节点启动 gtm

$ gtm_ctl -Z gtm -D /var/lib/pgxl/gtm start

对应的关闭

$ gtm_ctl -Z gtm -D /var/lib/pgxl/gtm stop

node1 节点启动 gtm-proxy

$ gtm_ctl -Z gtm_proxy -D /var/lib/pgxl/gtm_proxy1 start
$ gtm_ctl -Z gtm_proxy -D /var/lib/pgxl/gtm_proxy2 start

对应的关闭

$ gtm_ctl -Z gtm_proxy -D /var/lib/pgxl/gtm_proxy1 stop
$ gtm_ctl -Z gtm_proxy -D /var/lib/pgxl/gtm_proxy2 stop

node2 节点启动 datanode

$ pg_ctl -D /var/lib/pgxl/data start -Z datanode

对应的关闭

$ pg_ctl stop -m fast -D /var/lib/pgxl/data -Z datanode

node3 节点启动 datanode

$ pg_ctl -D /var/lib/pgxl/data start -Z datanode

对应的关闭

$ pg_ctl stop -m fast -D /var/lib/pgxl/data -Z datanode

node1 节点启动 coordinator

$ pg_ctl -D /var/lib/pgxl/coordinator1 start -Z coordinator
$ pg_ctl -D /var/lib/pgxl/coordinator2 start -Z coordinator

对应的关闭

$ pg_ctl stop -m fast -D /var/lib/pgxl/coordinator1 -Z coordinator
$ pg_ctl stop -m fast -D /var/lib/pgxl/coordinator2 -Z coordinator

pg-xl配置

node1 节点上登录 coordinator1

$ psql -p 5432
psql (PGXL 10alpha2, based on PG 10.3 (Postgres-XL 10alpha2))
Type "help" for help.

postgres=# 
postgres=# select * from pgxc_node;
 node_name    | node_type | node_port | node_host | nodeis_primary | nodeis_preferred |  node_id   
--------------+-----------+-----------+-----------+----------------+------------------+------------
 coordinator1 | C         |      5432 | localhost | f              | f                | 1148549230
(1 row)

postgres=# alter node coordinator1 with (type=coordinator,host='node1', port=5432);
postgres=# create node coordinator2 with (type=coordinator,host='node1', port=5433);

postgres=# create node datanode1 with (type=datanode, host='node2',port=5432,primary,preferred);
postgres=# create node datanode2 with (type=datanode, host='node3',port=5432);

postgres=# select pgxc_pool_reload();
postgres=# select * from pgxc_node;
  node_name   | node_type | node_port | node_host | nodeis_primary | nodeis_preferred |   node_id   
--------------+-----------+-----------+-----------+----------------+------------------+-------------
 coordinator1 | C         |      5432 | node1     | f              | f                |  1148549230
 coordinator2 | C         |      5433 | node1     | f              | f                | -2089598990
 datanode1    | D         |      5432 | node2     | t              | t                |   888802358
 datanode2    | D         |      5432 | node3     | f              | f                |  -905831925
(4 rows)

依次操作 node1 节点的 coordinator2，node2 节点的 datanode，node3 节点的datanode

验证

只能通过node1 节点的 coordinator1、coordinator1 去操作。node2，node3节点的数据都是只读的。

目前的coordinator配置如下，也可以往前翻看看记录。
node1 coordinator1 5432
node1 coordinator2 5433

$ psql -p 5432
psql (PGXL 10alpha2, based on PG 10.3 (Postgres-XL 10alpha2))
Type "help" for help.

postgres=#
postgres=# create table tmp_t0(c0 varchar(100),c1 varchar(100));
peiybdb=# insert into tmp_t0(c0,c1) SELECT id::varchar,md5(id::varchar) FROM generate_series(1,10000) as id;
INSERT 0 10000
peiybdb=# d+ tmp_t0
                                          Table "public.tmp_t0"
 Column |          Type          | Collation | Nullable | Default | Storage  | Stats target | Description 
--------+------------------------+-----------+----------+---------+----------+--------------+-------------
 c0     | character varying(100) |           |          |         | extended |              | 
 c1     | character varying(100) |           |          |         | extended |              | 
Distribute By: HASH(c0)
Location Nodes: ALL DATANODES

到 node2、node3的数据节点查看对应的数据

node2 查询结果

peiybdb=# select count(1) from tmp_t0;
 count 
-------
  5081
(1 row)

node3 查询结果

peiybdb=# select count(1) from tmp_t0;
 count 
-------
  4919
(1 row)

如果在node2，node3上插入数据，会收到报错提示。

peiybdb=# insert into tmp_t0(c0,c1) SELECT id::varchar,md5(id::varchar) FROM generate_series(1,10000) as id;
ERROR:  cannot execute INSERT in a read-only transaction

下面是简单的建表分析

CREATE TABLE table_name(...)
DISTRIBUTE BY 
HASH(col)|MODULO(col)|ROUNDROBIN|REPLICATION
TO NODE(nodename1,nodename2...)

可以看到，如果DISTRIBUTE BY 后面有如下选项：
REPLICATION，则是复制模式，其余则是分片模式，
HASH 指的是按照指定列的哈希值分布数据，
MODULO 指的是按照指定列的取摩运算分布数据，
ROUNDROBIN 指的是按照轮询的方式分布数据

TO NODE指定了数据分布的节点范围，如果没有指定则默认所有数据节点参与数据分布。如果没有指定分布模式，即使用普通的CREATE TABLE语句，PGXL会默认采用分片模式将数据分布到所有数据节点。

参考：
https://www.postgres-xl.org/
https://www.postgres-xl.org/overview/
https://www.postgres-xl.org/download/
https://www.postgres-xl.org/documentation/

https://git.postgresql.org/gitweb/?p=postgres-xl.git;a=summary

https://www.2ndquadrant.com/en/resources/postgres-xl/

https://www.postgresql.org/download

相关阅读:
MapReduce原理
 用redis构建分布式锁
 Python中类的特殊变量
 Python之元类
 python之WSGI与Guincorn
一种消息和任务队列——beanstalkd
LRU 算法
 extern、static、restrict、volatile 关键字
 bigtable原理
 Go的微服务库kite
原文地址：https://www.cnblogs.com/ctypyb2002/p/9792937.html