- 论坛徽章:
- 0
|
Sequoia for EnterpriseDB Postgres Plus AS 8.3 数据库集群简明安装指南
字体: 小 中 大 | 打印 发表于: 2008-10-31 17:43 作者: scott.siu 来源: EnterpriseDB中文社区(www.chinaedb.org)
Sequoia下载地址
理论上这个Sequoia可以为所有提供JDBC驱动的数据库实现集群，是C-JDBC项目的一个很好的扩展
https://forge.continuent.org/frs/?group_id=6
环境：
Red Hat Enterprise Linux 5 Update 2
EnterpriseDB Postgres Plus Advanced Server 8.3
Sequoia 2.10.10
两台服务器
controller1：192.168.100.21
controller2：192.168.100.22
gogogogogogogogogogogogogogogogogogogogogogo
一、安装前准备(在两台服务器中进行以下操作)
1. JDK安装
2. 设置NTP时间服务，使两台服务器保持时间同步
3. 安装EDB及Sequoia(将到/opt/sequoia-2.10.10xxxxx目录)
4. 设置EDB网络连接关系
修改 /opt/PostgresPlus/8.3AS/data/pg_hba.conf
找到：
host all all 127.0.0.1/32 md5
在这一行之前加入：
host all all 192.168.100.21/32 trust
host all all 192.168.100.22/32 trust
5. 设置环境变量
修改 /etc/bashrc
加入:
JAVA_HOME=/usr/java/jdk1.5.0_13
EDB_BIN=/opt/PostgresPlus/8.3AS/dbserver/bin
SEQUOIA_HOME=/opt/sequoia-2.10.10-bin
PATH=$PATH [点击图片可在新窗口打开] EDB_BIN
export PATH JAVA_HOME SEQUOIA_HOME
6. reboot
7. 初始化EDB，在两台controller上建立集群数据库及其RecoveryLog库
# edb-psql edb enterprisedb
输入密码后执行以下SQL语句
CREATE ROLE "edbCluster" LOGIN ENCRYPTED PASSWORD 'md5fe5cf0e29db4880fbb6e05cff4d17677'
NOINHERIT CREATEDB
VALID UNTIL 'infinity';
\c edbCluster;
CREATE DATABASE "edbCluster"
WITH OWNER = "edbCluster"
ENCODING = 'UTF8';
CREATE SCHEMA "edbCluster"
AUTHORIZATION "edbCluster";
CREATE DATABASE "edbCluster_RecLog"
WITH OWNER = "edbCluster"
ENCODING = 'UTF8';
\q;
二、配置Sequoia
1. 复制/opt/PostgresPlus/8.3AS/connectors/jdbc/edb-jdbc14.jar到/opt/sequoia-2.10.10-bin/drivers
2. 将/opt/sequoia-2.10.10-bin/lib下的sequoia-backupers.jar及jgroups-core.jar改名为*.bak
3. 将我修改好的
[点击图片可在新窗口打开] jgroups-core.jar
(2008-10-31 17:43:54, Size: 1.17 MB, Downloads: 0)
<附件:jgroups-core.jar>及
[点击图片可在新窗口打开] sequoia-backupers.jar
(2008-10-31 17:43:54, Size: 82.8 KB, Downloads: 0)
<附件:sequoia-backup.jar>到/opt/sequoia-2.10.10-bin/lib下
4. 复制/opt/sequoia-2.10.10-bin/config/virtualdatabase/hsqldb-raidb1-distribution-1.xml
为/opt/sequoia-2.10.10-bin/config/virtualdatabase/edb-raidb1-distribution.xml
5. 修改/opt/sequoia-2.10.10-bin/config/virtualdatabase/edb-raidb1-distribution.xml
---------修改VDB名
找到：
<VirtualDatabase name="myDB">
改为：
<VirtualDatabase name="edbCluster">
---------指定数据库Backuper，dumpServer为本机IP [两台Controller各自修改]
找到：
<Backuper backuperName="Octopus"
className="org.continuent.sequoia.controller.backup.backupers.OctopusBackuper"
options="zip=true"/>
改为：
[controller1]：
<Backuper backuperName="EdbBackuper1"
className="org.continuent.sequoia.controller.backup.backupers.EnterpriseDBBinaryBackuper"
options="dumpServer=192.168.100.21"/>
[controller2]：
<Backuper backuperName="EdbBackuper2"
className="org.continuent.sequoia.controller.backup.backupers.EnterpriseDBBinaryBackuper"
options="dumpServer=192.168.100.22"/>
---------指定Controller的Admin用户名及密码，及连接的User用户名及密码
找到：
<AuthenticationManager>
<Admin>
<User username="admin" password=""/>
</Admin>
<VirtualUsers>
<VirtualLogin vLogin="user" vPassword=""/>
</VirtualUsers>
</AuthenticationManager>
改为：
<AuthenticationManager>
<Admin>
<User username="admin" password="admin"/>
</Admin>
<VirtualUsers>
<VirtualLogin vLogin="user" vPassword="user"/>
</VirtualUsers>
</AuthenticationManager>
---------指定本地Backend数据库，name及url为本地的数据库参数 [两台Controller各自修改]
找到：
<DatabaseBackend name="localhost1" driver="org.hsqldb.jdbcDriver" url="jdbc:hsqldb:hsql://localhost:9001" connecti>
<ConnectionManager vLogin="user" rLogin="TEST" rPassword="">
<VariablePoolConnectionManager initPoolSize="10" minPoolSize="5" maxPoolSize="50" idleTimeout="30" waitTimeout="10"/>
</ConnectionManager>
</DatabaseBackend>
改为：
[controller1]：
<DatabaseBackend name="edb1" driver="com.edb.Driver" url="jdbc:edb://192.168.100.21:5444/edbCluster" connecti>
<ConnectionManager vLogin="user" rLogin="edbCluster" rPassword="edbCluster">
<VariablePoolConnectionManager initPoolSize="10" minPoolSize="5" maxPoolSize="50" idleTimeout="30" waitTimeout="10"/>
</ConnectionManager>
</DatabaseBackend>
[controller2]：
<DatabaseBackend name="edb2" driver="com.edb.Driver" url="jdbc:edb://192.168.100.22:5444/edbCluster" connecti>
<ConnectionManager vLogin="user" rLogin="edbCluster" rPassword="edbCluster">
<VariablePoolConnectionManager initPoolSize="10" minPoolSize="5" maxPoolSize="50" idleTimeout="30" waitTimeout="10"/>
</ConnectionManager>
</DatabaseBackend>
---------注释以下内容
找到：
<DatabaseBackend name="localhost2" driver="org.hsqldb.jdbcDriver" url="jdbc:hsqldb:hsql://localhost:9002" connecti>
<ConnectionManager vLogin="user" rLogin="TEST" rPassword="">
<VariablePoolConnectionManager initPoolSize="10" minPoolSize="5" maxPoolSize="50" idleTimeout="30" waitTimeout="10"/>
</ConnectionManager>
</DatabaseBackend>
改为：
<!--
<DatabaseBackend name="localhost2" driver="org.hsqldb.jdbcDriver" url="jdbc:hsqldb:hsql://localhost:9002" connecti>
<ConnectionManager vLogin="user" rLogin="TEST" rPassword="">
<VariablePoolConnectionManager initPoolSize="10" minPoolSize="5" maxPoolSize="50" idleTimeout="30" waitTimeout="10"/>
</ConnectionManager>
</DatabaseBackend>
-->
---------各Controller的RecoveryLog所存放的位置 [两台Controller各自修改]
找到：
<RecoveryLog driver="org.hsqldb.jdbcDriver"
url="jdbc:hsqldb:hsql://localhost:9003" login="TEST" password="">
改为：
[controller1]：
<RecoveryLog driver="com.edb.Driver"
url="jdbc:edb://192.168.100.21:5444/edbCluster_RecLog" login="edbCluster" password="edbCluster">
[controller2]：
<RecoveryLog driver="com.edb.Driver"
url="jdbc:edb://192.168.100.22:5444/edbCluster_RecLog" login="edbCluster" password="edbCluster">
6. 复制/opt/sequoia-2.10.10-bin/config/controller/controller-distributed-1.xml
为/opt/sequoia-2.10.10-bin/config/controller/controller-edb.xml
7. 修改/opt/sequoia-2.10.10-bin/config/controller/controller-edb.xml
---------修改监听的网络地址及端口
找到：
<Controller ipAddress="127.0.0.1" port="25322">
改为：
<Controller ipAddress="192.168.100.21" port="25322">
---------修改VDB所用的配置文件及名字
找到：
<VirtualDatabase c virtualDatabaseName="myDB" autoEnableBackends="true"/>
改为：
<VirtualDatabase c virtualDatabaseName="edbCluster" autoEnableBackends="true"/>
8. 修改/opt/sequoia-2.10.10-bin/config/sequencer.xml [两台Controller各自修改]
找到：
<UDP mcast_port="45566"
改为：
[controller1]：
<UDP bind_addr="192.168.100.21"
mcast_port="45566"
[controller2]：
<UDP bind_addr="192.168.100.22"
mcast_port="45566"
三、起动Sequoia
1. 在contoller1中运行
# /opt/sequoia-2.10.10-bin/bin/controller.sh -f /opt/sequoia-2.10.10-bin/config/controller/controller-edb.xml
2. 在contoller2中运行
# /opt/sequoia-2.10.10-bin/bin/controller.sh -f /opt/sequoia-2.10.10-bin/config/controller/controller-edb.xml
------这两步没有Error证明系统可以正常启动了
四、初始化Sequoia
controller1：
# /opt/sequoia-2.10.10-bin/bin/console.sh
localhost:1090 > admin edbCluster
Virtual database Administrator Login > admin
Virtual database Administrator Password > admin
Ready to administrate virtual database edbCluster
edbCluster(admin) > initialize edb1
Virtual Database edbCluster has been successfully initialized from backend edb1
edbCluster(admin) > backup edb1 initial_dump EdbBackuper1 /tmp
Backend login for backup process > enterprisedb
Backend password for backup process > edb
Backup backend edb1 in dump file initial_dump
13:50:02,943 INFO controller.RequestManager.edbCluster Starting backup of backend edb1
13:50:03,046 INFO backup.backupers.NativeCommandExec Starting execution of "pg_dump -h 192.168.100.21 -p 5444 -U enterprisedb --format=c -f /tmp/initial_dump edbCluster"
13:50:03,668 INFO backup.backupers.NativeCommandExec Command "pg_dump -h 192.168.100.21 -p 5444 -U enterprisedb --format=c -f /tmp/initial_dump edbCluster" logged 0 errors and terminated with exitcode 0
13:50:03,764 INFO controller.RequestManager.edbCluster Backup has completed.
edbCluster(admin) > enable edb1
Enabling backend edb1 from its last known checkpoint
13:52:03,810 INFO DatabaseBackend.edbCluster.edb1 Detected backend as: EnterpriseDB
13:52:29,263 INFO controller.RequestManager.edbCluster Setting new virtual database schema.
13:52:29,370 INFO controller.recoverylog.RecoverThread Starting recovery
13:52:29,375 INFO controller.RequestManager.edbCluster Suspending activity for edbCluster
13:52:29,514 INFO controller.RequestManager.edbCluster All activity is suspended for edbCluster
13:52:29,555 INFO controller.recoverylog.RecoverThread Recovery completed
13:52:50,817 INFO controller.RequestManager.edbCluster Resuming activity for edbCluster
13:52:50,974 INFO controller.RequestManager.edbCluster All activity is now resumed for edbCluster
13:52:50,975 INFO controller.recoverylog.RecoverThread Database backend edb1 is now enabled
edbCluster(admin) > transfer dump initial_dump 192.168.100.22:1090
Transferring dump initial_dump to controller 192.168.100.22:1090...
13:58:25,689 INFO controller.backup.BackupManager Using provided dump-server address: 192.168.100.21
13:58:25,699 INFO controller.backup.BackupManager Dump server started @ 0.0.0.0/0.0.0.0:46809
13:58:25,963 INFO controller.backup.BackupManager Client connected to dump server from /192.168.100.22:38963
13:58:26,047 INFO controller.backup.BackupManager Dump server terminated.
Done
edbCluster(admin) > expert on
Expert mode on
edbCluster(admin) > restore log initial_dump 192.168.100.21:1090
A restore log command must be applied to a remote controller
edbCluster(admin) > restore log initial_dump 192.168.100.22:1090
14:01:09,632 INFO controller.RequestManager.edbCluster Suspending activity for edbCluster
14:01:09,880 INFO controller.RequestManager.edbCluster All activity is suspended for edbCluster
14:01:09,932 INFO controller.virtualdatabase.edbCluster Storing checkpoint now-192.168.100.21:25322-20081021140109631+0800
14:01:09,937 WARN controller.virtualdatabase.edbCluster No information has been found for remote controller Member(address=/192.168.100.21:32787, uid=edbCluster)
14:01:10,014 WARN sequoia.controller.recoverylog Checkpoint now-192.168.100.21:25322-20081021140109631+0800 was stored
14:01:10,260 INFO controller.RequestManager.edbCluster Resuming activity for edbCluster
14:01:10,390 INFO controller.RequestManager.edbCluster All activity is now resumed for edbCluster
Done
controller2：
# /opt/sequoia-2.10.10-bin/bin/console.sh
localhost:1090 > admin edbCluster
Virtual database Administrator Login > admin
Virtual database Administrator Password > admin
Ready to administrate virtual database edbCluster
edbCluster(admin) > show dumps
+--------------+----------------------------+--------------------------+------+------------------+---------+--------+
| Name | Checkpoint | Format | Path | Date | Backend | Tables |
+--------------+----------------------------+--------------------------+------+------------------+---------+--------+
| initial_dump | Initial_empty_recovery_log | EnterpriseDB Binary Dump | /tmp | 10/21/08 1:50 PM | edb1 | * |
+--------------+----------------------------+--------------------------+------+------------------+---------+--------+
edbCluster(admin) > restore backend edb2 initial_dump
Backend login for restore process > enterprisedb
Backend password for restore process > edb
Restoring backend edb2 with dump initial_dump
edbCluster(admin) > enable edb2
Enabling backend edb2 from its last known checkpoint
五、测试
controller2:
edbCluster(admin) > quit
localhost:1090 > sql client jdbc:sequoia://192.168.100.21:25322,192.168.100.22:25322/edbCluster
Login > user
Password > user
jdbc:sequoia://192.168.100.21:25322,192.168.100.22:25322/edbCluster (user) > create table test(i integer);
Affected rows: 0
Query executed in 7 s 737 ms .
jdbc:sequoia://192.168.100.21:25322,192.168.100.22:25322/edbCluster (user) > select * from test;
Query executed in 5 s 237 ms .
jdbc:sequoia://192.168.100.21:25322,192.168.100.22:25322/edbCluster (user) > insert into test values(1);
Affected rows: 1
Query executed in 5 s 801 ms .
jdbc:sequoia://192.168.100.21:25322,192.168.100.22:25322/edbCluster (user) > insert into test values(2);
Affected rows: 1
Query executed in 0 s 325 ms .
jdbc:sequoia://192.168.100.21:25322,192.168.100.22:25322/edbCluster (user) > select * from test;
+---+
| i |
+---+
| 1 |
| 2 |
+---+
Query executed in 0 s 114 ms .
此时可以通过Postgres Studio查看两台服务器上的edbCluster数据库，我们所做的create table、insert等操作，在两个数据库中同时生效，由此证明并行数据库已经生效
到些集群中的数据实时同步及负载均衡查询已经实现
哪一但集群中的一个数据库backend宕机，后应该如何回复？待续……
[ 本帖最后由 scottsiu 于 2008-11-5 12:46 编辑 ] |
|
免费注册
发表于 2008-11-03 23:55
rg/jgroups/ChannelException。不知道怎么出来，请高手指点，急需，谢谢。