1.环境
OS: CentOS release 6.4 (Final) DB: postgresql 9.3.6 pgpool服务器: pgpool 172.16.0.240 数据库主服务器:master 172.16.0.241 数据库从服务器:slave 172.16.0.242
其中主从数据库使用的流复制,并且已经配置完毕,新配置的pgpool使用postgres用户进行管理。新架构使用的主备模式外加流复制,此架构支持流复制、负载均衡、故障恢复,不支持复制和并行查询,主库可以支持读写,从库只读。
2.pgpool安装
[root@pgpool opt]# yum install -y http://yum.postgresql.org/9.3/redhat/rhel-6-x86_64/pgdg-redhat93-9.3-1.noarch.rpm [root@pgpool opt]# yum install postgresql93-devel [root@pgpool opt]# wget http://www.pgpool.net/download.php?f=pgpool-II-3.4.2.tar.gz [root@pgpool opt]# mv download.php\?f\=pgpool-II-3.4.2.tar.gz pgpool-II-3.4.2.tar.gz [root@pgpool opt]# tar -zxvf pgpool-II-3.4.2.tar.gz [root@pgpool opt]# cd pgpool-II-3.4.2 [root@pgpool pgpool-II-3.4.2]# ./configure --with-pgsql=/usr/pgsql-9.3
如果不加--with-pgsql=/usr/pgsql-9.3,可能会出现configure: error: libpq is not installed or libpq is old错误
[root@pgpool pgpool-II-3.4.2]# make [root@pgpool pgpool-II-3.4.2]# make install
3.修改pgpool配置
启用并修改配置文件pgpool.conf
[root@pgpool etc]# cp /usr/local/etc/pgpool.conf.sample-stream /usr/local/etc/pgpool.conf
修改pgpool.conf文件中的内容
# - pgpool Communication Manager Connection Settings - listen_addresses = '*' ---默认0是主库,其它是从库,backend_weight可以控制数据库读在两台机器上的分配比例 # - Backend Connection Settings - backend_hostname0 = '172.16.0.241' backend_port0 = 5432 backend_weight0 = 1 backend_data_directory0 = '/var/lib/pgsql/9.3/data' backend_flag0 = 'ALLOW_TO_FAILOVER' backend_hostname1 = '172.16.0.242' backend_port1 = 5432 backend_weight1 = 1 backend_data_directory1 = '/data1' backend_flag1 = 'ALLOW_TO_FAILOVER' backend_data_directory0 = '/var/lib/pgsql/9.3/data' # - Authentication - ---开启pgpool的hba认证 enable_pool_hba = on #------------------------------------------------------------------------------ # MASTER/SLAVE MODE #------------------------------------------------------------------------------ sr_check_user = 'postgres' sr_check_password = '123456' #------------------------------------------------------------------------------ # HEALTH CHECK #------------------------------------------------------------------------------ health_check_period = 1 health_check_user = 'postgres' health_check_password = '123456' #------------------------------------------------------------------------------ # FAILOVER AND FAILBACK #------------------------------------------------------------------------------ ---用来在主库失败时,把只读的从库切为主库 failover_command = '/usr/local/bin/failover_stream.sh %d %H /tmp/trigger_file0'
启用并修改配置文件pcp.conf
[root@pgpool etc]# cp /usr/local/etc/pcp.conf.sample /usr/local/etc/pcp.conf
此文件是管理端口,可以暂时不用配置。
启用并修改配置文件pool_hba.conf
[root@pgpool etc]# cp /usr/local/etc/pool_hba.conf.sample /usr/local/etc/pool_hba.conf ---添加一行: host all all 172.16.0.0/24 md5 ---删除一行 host all all 127.0.0.1/32 trust
对pgpool的访问策略,要设置为md5的方式。
启用配置文件pool_passwd
[root@pgpool etc]# pg_md5 -m -p -u postgres pool_passwd password:
密码填的是123456,该文件会在执行文件后自动生成。从远程连接pgpool的时候,需要使用该密码来访问,pgpool使用该用户密码来访问后面的真正的数据库。
添加主备脚本切换脚本failover_stream.sh
[root@pgpool bin]# vi /usr/local/bin/failover_stream.sh # Failover command for streaming replication. # This script assumes that DB node 0 is primary, and 1 is standby. # # If standby goes down, do nothing. If primary goes down, create a # trigger file so that standby takes over primary node. # # Arguments: $1: failed node id. $2: new master hostname. $3: path to # trigger file. failed_node=$1 new_master=$2 trigger_file=$3 # Do nothing if standby goes down. if [ $failed_node = 1 ]; then exit 0; fi # Create the trigger file. /usr/bin/ssh -T $new_master /bin/touch $trigger_file exit 0; [root@pgpool bin]# chmod 700 failover_stream.sh
对应的从库slave的recovery.conf配置:
[root@slave data]# cat recovery.conf standby_mode = 'on' primary_conninfo = 'host=172.16.0.241 port=5432 user=rep_user' trigger_file = '/tmp/trigger_file0'
该trigger_file的参数要和pgpool.conf文件中failover_command对应的文件保持一致。
4.设置主机互信
修改hosts文件
对每个主机的/etc/hosts文件添加以下内容,
172.16.0.240 pgpool 172.16.0.241 master 172.16.0.242 slave
pgpool主机生成公钥和私钥
[root@pgpool bin]# su - postgres -bash-4.1$ ssh-keygen -t rsa Generating public/private rsa key pair. Enter file in which to save the key (/var/lib/pgsql/.ssh/id_rsa): Created directory '/var/lib/pgsql/.ssh'. Enter passphrase (empty for no passphrase): Enter same passphrase again: Your identification has been saved in /var/lib/pgsql/.ssh/id_rsa. Your public key has been saved in /var/lib/pgsql/.ssh/id_rsa.pub. The key fingerprint is: a0:93:d4:b5:ed:26:d0:94:a5:e7:99:95:6b:d6:18:af postgres@pgpool The key's randomart image is: +--[ RSA 2048]----+ | oo. | | . +.+ . | | . + + o + | | . o o + + * | | + S * = o | | . o o . | | E | | | | | +-----------------+
输入命令后一直回车即可。在master和slave节点同样执行上面的操作。
同步公钥
在pgpool节点上执行
-bash-4.1$ ssh-copy-id -i ~/.ssh/id_rsa.pub postgres@172.16.0.241 -bash-4.1$ ssh-copy-id -i ~/.ssh/id_rsa.pub postgres@172.16.0.242
在master节点上执行
-bash-4.1$ ssh-copy-id -i ~/.ssh/id_rsa.pub postgres@172.16.0.241 -bash-4.1$ ssh-copy-id -i ~/.ssh/id_rsa.pub postgres@172.16.0.242
在slave节点上执行
-bash-4.1$ ssh-copy-id -i ~/.ssh/id_rsa.pub postgres@172.16.0.241 -bash-4.1$ ssh-copy-id -i ~/.ssh/id_rsa.pub postgres@172.16.0.242
上面的操作执行后,在每个机器检查一下,是否可以用ssh username@nodename的方式,对其他节点进行免密码的ssh访问,第一次访问的时候会记录known_hosts,否则ssh时候会有是否信任节点的提示。
5.启动pgpool
使用postgres用户启动pgpool服务的话,需要修改对应目录的权限,否则pgpool会启动失败
[root@pgpool etc]# mkdir /var/run/pgpool [root@pgpool run]# chown -R postgres:postgres /var/run/pgpool [root@pgpool local]# chown -R postgres:postgres /usr/local/etc [root@pgpool local]# chown -R postgres:postgres /usr/local/bin
检验主从数据库切换的脚本是否可以正常运行,在pgpool服务器执行命令,检测在slave服务器是否生成了trigger_file0文件
shell脚本在pgpool执行
-bash-4.1$ /usr/local/bin/failover_stream.sh 0 slave /tmp/trigger_file0
切换到slave主机检查
[root@slave ~]# ls /tmp ssh-rangWC1783 trigger_file0 yum.log
启动pgpool
-bash-4.1$ pgpool -n -d > /tmp/pgpool.log 2>&1 & [1] 10474
6.检测pgpool功能
任一节点(如slave节点)连接pgpool
[root@slave ~]# psql -h 172.16.0.240 -p 9999 -d test -U postgres Password for user postgres: psql (9.3.6) Type "help" for help. test=# select * from t4; id ---- 1 (1 row) test=# show pool_nodes; node_id | hostname | port | status | lb_weight | role ---------+--------------+------+--------+-----------+--------- 0 | 172.16.0.241 | 5432 | 2 | 0.500000 | primary 1 | 172.16.0.242 | 5432 | 2 | 0.500000 | standby (2 rows)
可以看到此时241数据库是主节点,242是从节点。
关闭主要节点,可以关闭服务也可以直接杀进程模拟数据库崩溃
[root@master data]# service postgresql-9.3 stop Stopping postgresql-9.3 service: [ OK ] ---master关闭后,从节点的连接中断后又成功连接上 server closed the connection unexpectedly This probably means the server terminated abnormally before or while processing the request. The connection to the server was lost. Attempting reset: Succeeded. test=# show pool_nodes; node_id | hostname | port | status | lb_weight | role ---------+--------------+------+--------+-----------+--------- 0 | 172.16.0.241 | 5432 | 3 | 0.500000 | standby 1 | 172.16.0.242 | 5432 | 2 | 0.500000 | primary test=# insert into t4 values (6); INSERT 0 1 test=# select * from t4; id ---- 1 6 (2 rows) test=#
pgpool发生主从切换后,242节点变为主节点,241节点关闭(status为3)。
上面查询show pool_nodes中得status字段含义
Status 由数字 [0 - 3]来表示。 0 - 该状态仅仅用于初始化,PCP从不显示它。 1 - 节点已启动,还没有连接。 2 - 节点已启动,连接被缓冲。 3 - 节点已关闭。
可以看到主备已经切换完成。
7.其它注意事项
- 若pgpool上的主从切换脚本忘记写,或者没能正常执行,show pool_nodes命令会显示两个节点都是standby,而集群整体是只读的,此时可以关闭并启动节点的数据库服务,然后重启pgpool。
- 测试trigger文件的时候,注意要及时删除生成的trigger 文件,否则会破坏主从架构,导致后面的实验失败。