CentOS和Ubuntu下安装配置Greenplum数据库集群（包括安装包和源码编译安装）

http://blog.csdn.net/prettyshuang/article/details/50501904

首先说一下，无论是CentOS/RedHat还是Ubuntu都可以按源码方式、安装包方式编译安装。

1.   规划

192.168.4.93（h93）       1个主master  2个主segment、2个镜像segment

192.168.4.94（h94）       1个备master  2个主segment、2个镜像segment

都安装在/home/wslu/gp/gpsql目录。

2.   安装依赖

在h93和h94安装以下依赖：

2.1.  Ubuntu

[python] view plain copy

  1. apt-get install -y git-core
  2. apt-get install -y gcc g++
  3. apt-get install -y ccache
  4. apt-get install -y libreadline-dev
  5. apt-get install -y bison flex
  6. apt-get install -y zlib1g-dev
  7. apt-get install -y openssl libssl-dev
  8. apt-get install -y libpam-dev
  9. apt-get install -y libcurl4-dev
  10. apt-get install -y libbz2-dev
  11. apt-get install -y python-dev
  12. apt-get install -y ssh
  13. apt-get install -y libcurl4-dev
  14. Package libcurl4-dev is a virtual package provided by:
  15. libcurl4-openssl-dev 7.38.0-4+deb8u2
  16. libcurl4-nss-dev 7.38.0-4+deb8u2
  17. libcurl4-gnutls-dev 7.38.0-4+deb8u2
  18. apt-get install -y python-pip
  19. pip install lockfile
  20. pip install paramiko
  21. pip install setuptools
  22. pip install epydoc
  23. pip install psi
  24. Note: debian8 required pip install --pre psi

2.2.  Centos

yum源要按照我的总结中的设置。

[python] view plain copy

  1. yum install –y git.x86_64
  2. yum install –y gcc.x86_64 gcc-c++.x86_64
  3. yum install –y ccache.x86_64
  4. yum install readline.x86_64 readline-devel.x86_64
  5. yum install bison.x86_64 bison-devel.x86_64
  6. yum install flex.x86_64 flex-devel.x86_64
  7. yum install zlib.x86_64 zlib-devel.x86_64
  8. yum install -y openssl.x86_64 openssl-devel.x86_64
  9. yum install -y pam.x86_64 pam-devel.x86_64
  10. yum install –y libcurl.x86_64 libcurl-devel.x86_64
  11. yum install bzip2-libs.x86_64 bzip2.x86_64 bzip2-devel.x86_64
  12. yum install libssh2.x86_64 libssh2-devel.x86_64
  13. yum install python-devel.x86_64
  14. yum install -y python-pip.noarch
  15. 接着执行：
  16. pip install lockfile
  17. pip install paramiko
  18. pip install setuptools
  19. pip install epydoc
  20. pip install psi
  21. 或者执行：
  22. yum install python-lockfile.noarch
  23. yum install python-PSI.x86_64
  24. yum install python-paramiko.noarch
  25. yum install python-setuptools.noarch
  26. yum install epydoc.noarch

3.   源码安装

3.1.  克隆源码

在h93和h94执行：

[python] view plain copy

  1. $ mkdir /home/wslu/gp/greenplum
  2. $ cd /home/wslu/gp/greenplum
  3. $ git clone https://github.com/greenplum-db/gpdb.

3.2.  编译安装

在h93和h94执行：

[python] view plain copy

  1. $ cd /home/wslu/gp/greenplum
  2. $ CFLAGS+="-O2" ./configure--prefix=/home/wslu/gp/gpsql --enable-debug --enable-depend --enable-cassert
  3. $ make
  4. $ make install

安装时如果遇到某些python包(lockfile、 paramiko、PSI等)找不到，可以参考HAWQ， (源码https://github.com/apache/incubator-hawq)<hawq_src>/tools/bin/pythonSrc/ 下所有的压缩包拷贝到/home/wslu/gp/greenplum/gpMgmt/bin/pythonSrc/ext/ 中，然后再make install即可。

至此集群源码编译完成。

4.   安装包安装

下载greenplum-db-4.3.6.1-build-2-RHEL5-x86_64.zip

[python] view plain copy

  1. $ unzip greenplum-db-4.3.6.1-build-2-RHEL5-x86_64.zip

在h93和h94使用普通用户安装：

[python] view plain copy

  1. $./greenplum-db-4.3.6.1-build-2-RHEL5-x86_64.bin

安装到 /home/wslu/gp/gpsql

5.   设置操作系统参数

在h93和h94执行下述设置：

5.1.  系统级参数配置

以下配置需要在root下进行配置，所以需要能够得到root权限。

首先是防火墙的设置：

[python] view plain copy

  1. service iptables status    #查看是否开启了防火墙
  2. service iptables stop       #关闭防火墙
  3. chkconfig iptables off      #设置开机不会启动防火墙

其次是ssh的速度设置：

[python] view plain copy

  1. sed -i ‘s/^GSS/#&/g‘ /etc/ssh/sshd_config      #用来加速SSH连接的
  2. service sshd reload

内核处理和内存方面的设置：

[python] view plain copy

  1. # 设置内核参数, 并在启动时生效
  2. sysctl -p - >>/etc/sysctl.conf <<EOF
  3. # configurations
  4. kernel.sysrq=1
  5. kernel.core_pattern=core
  6. kernel.core_uses_pid=1
  7. kernel.msgmnb=65536
  8. kernel.msgmax=65536
  9. kernel.msgmni=2048
  10. kernel.sem=25600 3200000 10000 14200
  11. net.ipv4.tcp_syncookies=1
  12. net.ipv4.ip_forward=0
  13. net.ipv4.conf.default.accept_source_route=0
  14. net.ipv4.tcp_tw_recycle=1
  15. net.ipv4.tcp_max_syn_backlog=4096
  16. net.ipv4.conf.all.arp_filter=1
  17. net.ipv4.ip_local_port_range=1025 65535
  18. net.core.netdev_max_backlog=10000
  19. net.core.rmem_max=2097152
  20. net.core.wmem_max=2097152
  21. vm.overcommit_memory=1
  22. EOF

如果配置集群时，因为要设置的共享内存大于系统最大共享内存而报错，则可以按照以下官方推荐的设置配置：

[python] view plain copy

  1. # vi /etc/sysctl.conf
  2. kernel.shmmax = 500000000
  3. kernel.shmmni = 4096
  4. kernel.shmall = 4000000000
  5. kernel.sem = 250 512000 100 2048
  6. kernel.sysrq = 1
  7. kernel.core_uses_pid = 1
  8. kernel.msgmnb = 65536
  9. kernel.msgmax = 65536
  10. kernel.msgmni = 2048
  11. net.ipv4.tcp_syncookies = 1
  12. net.ipv4.ip_forward = 0
  13. net.ipv4.conf.default.accept_source_route = 0
  14. net.ipv4.tcp_tw_recycle = 1
  15. net.ipv4.tcp_max_syn_backlog = 4096
  16. net.ipv4.conf.all.arp_filter = 1
  17. net.ipv4.ip_local_port_range = 1025 65535
  18. net.core.netdev_max_backlog = 10000
  19. net.core.rmem_max = 2097152
  20. net.core.wmem_max = 2097152
  21. vm.overcommit_memory = 2

文件读写方面的设置：

[python] view plain copy

  1. # 设置limits
  2. cat >>/etc/security/limits.d/99-kingbasedbcloud.conf <<EOF
  3. #KingbaseDBCloud configurations
  4. * soft nofile 65536
  5. * hard nofile 65536
  6. * soft nproc 131072
  7. * hard nproc 131072
  8. EOF

5.2. 数据库级的参数设置

GUC参数设置，此设置需要按照实际机器来进行配置，常用的配置有如下几点：

[python] view plain copy

  1. work_mem=1GB
  2. shared_buffers=2GB
  3. max_connections=500
  4. max_pool_size=2000
  5. enable_mergejoin=off
  6. enable_nestloop=off
  7. max_prepared_transactions=50
  8. autovacuum=off
  9. interconnect_setup_timeout=1200

6.   demo集群

安装完成后，就可以使用demo集群了。

执行下面命令即可在本机创建包含3个segment，3个segment-mirror，1个master的集群：

[python] view plain copy

  1. $ cd /home/wslu/gp/gpsql
  2. $ source greenplum_path.sh
  3. $ gpssh-exkeys –h localhost
  4. $ cd gpAux/gpdemo
  5. $ make cluster
  6. $ source gpdemo-env.sh

如果不想用demo集群，可以直接跳过本小节。

7.   设置环境变量

[python] view plain copy

  1. $ source gpsql/greenplum_path.sh
  2. $ exportMASTER_DATA_DIRECTORY=/home/wslu/gp/gpsql/data/master/gpseg-1

8.   交换key

在h93执行：

[python] view plain copy

  1. $ gpssh-exkeys –h h94

会提示输入h94的wslu用户的密码。（用户必须和h93相同）

则h93和h94之间可以无密码登陆了。

9. 初始化集群

（1）       在h93和h94执行下述指令，以创建数据目录：

[python] view plain copy

  1. $ mkdir gpsql/data/primary gpsql/data/mirror gpsql/data/master –p

（2）       在h93创建配置文件configs/gpinitsystem_config，内容如下：

[python] view plain copy

  1. ARRAY_NAME="EMC Greenplum DW"
  2. SEG_PREFIX=gpseg
  3. PORT_BASE=40000
  4. declare -a DATA_DIRECTORY=(/home/wslu/gp/gpsql/data/primary /home/wslu/gp/gpsql/data/primary)
  5. MASTER_HOSTNAME=h93
  6. MASTER_DIRECTORY=/home/wslu/gp/gpsql/data/master
  7. MASTER_PORT=5432
  8. TRUSTED_SHELL=ssh
  9. CHECK_POINT_SEGMENTS=8
  10. ENCODING=UNICODE
  11. MIRROR_PORT_BASE=50000
  12. REPLICATION_PORT_BASE=41000
  13. MIRROR_REPLICATION_PORT_BASE=51000
  14. declare -a MIRROR_DATA_DIRECTORY=(/home/wslu/gp/gpsql/data/mirror /home/wslu/gp/gpsql/data/mirror)

注意：configs目录是我自己创建的、便于保存自定义配置文件的目录。该步骤的目的是创建一个初始化时要用的配置文件，并没有路径的要求。

（3）       在h93创建配置文件configs/hostfile_gpinitsystem，内容如下：

[python] view plain copy

  1. h93
  2. h94

注意：configs目录是我自己创建的、便于保存自定义配置文件的目录。该步骤的目的是创建一个初始化时要用的配置文件，并没有路径的要求。

（4）       在h93执行下述指令初始化集群：

[python] view plain copy

  1. [[email protected] gpsql]$ gpinitsystem -c configs/gpinitsystem_config -h configs/hostfile_gpinitsystem –a
  2. 20160114:14:30:03:005980 gpinitsystem:h93:wslu-[INFO]:-Checking configuration parameters, please wait...
  3. 20160114:14:30:03:005980 gpinitsystem:h93:wslu-[INFO]:-Reading Greenplum configuration file configs/gpinitsystem_config
  4. 20160114:14:30:03:005980 gpinitsystem:h93:wslu-[INFO]:-Locale has not been set in configs/gpinitsystem_config, will set to default value
  5. 20160114:14:30:03:005980 gpinitsystem:h93:wslu-[INFO]:-Locale set to en_US.utf8
  6. 20160114:14:30:03:005980 gpinitsystem:h93:wslu-[INFO]:-No DATABASE_NAME set, will exit following template1 updates
  7. 20160114:14:30:03:005980 gpinitsystem:h93:wslu-[INFO]:-MASTER_MAX_CONNECT not set, will set to default value 250
  8. 20160114:14:30:03:005980 gpinitsystem:h93:wslu-[INFO]:-Checking configuration parameters, Completed
  9. 20160114:14:30:04:005980 gpinitsystem:h93:wslu-[INFO]:-Commencing multi-home checks, please wait...
  10. ..
  11. 20160114:14:30:05:005980 gpinitsystem:h93:wslu-[INFO]:-Configuring build for standard array
  12. 20160114:14:30:05:005980 gpinitsystem:h93:wslu-[INFO]:-Commencing multi-home checks, Completed
  13. 20160114:14:30:05:005980 gpinitsystem:h93:wslu-[INFO]:-Building primary segment instance array, please wait...
  14. ....
  15. 20160114:14:30:08:005980 gpinitsystem:h93:wslu-[INFO]:-Building group mirror array type , please wait...
  16. ....
  17. 20160114:14:30:12:005980 gpinitsystem:h93:wslu-[INFO]:-Checking Master host
  18. 20160114:14:30:12:005980 gpinitsystem:h93:wslu-[INFO]:-Checking new segment hosts, please wait...
  19. ........
  20. 20160114:14:30:28:005980 gpinitsystem:h93:wslu-[INFO]:-Checking new segment hosts, Completed
  21. 20160114:14:30:28:005980 gpinitsystem:h93:wslu-[INFO]:-Greenplum Database Creation Parameters
  22. 20160114:14:30:28:005980 gpinitsystem:h93:wslu-[INFO]:---------------------------------------
  23. 20160114:14:30:28:005980 gpinitsystem:h93:wslu-[INFO]:-Master Configuration
  24. 20160114:14:30:28:005980 gpinitsystem:h93:wslu-[INFO]:---------------------------------------
  25. 20160114:14:30:28:005980 gpinitsystem:h93:wslu-[INFO]:-Master instance name       = EMC Greenplum DW
  26. 20160114:14:30:28:005980 gpinitsystem:h93:wslu-[INFO]:-Master hostname            = h93
  27. 20160114:14:30:28:005980 gpinitsystem:h93:wslu-[INFO]:-Master port                = 5432
  28. 20160114:14:30:28:005980 gpinitsystem:h93:wslu-[INFO]:-Master instance dir        = /home/wslu/gp/gpsql/data/master/gpseg-1
  29. 20160114:14:30:28:005980 gpinitsystem:h93:wslu-[INFO]:-Master LOCALE              = en_US.utf8
  30. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Greenplum segment prefix   = gpseg
  31. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Master Database            =
  32. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Master connections         = 250
  33. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Master buffers             = 128000kB
  34. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Segment connections        = 750
  35. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Segment buffers            = 128000kB
  36. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Checkpoint segments        = 8
  37. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Encoding                   = UNICODE
  38. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Postgres param file        = Off
  39. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Initdb to be used          = /home/wslu/gp/gpsql/bin/initdb
  40. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-GP_LIBRARY_PATH is         = /home/wslu/gp/gpsql/lib
  41. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Ulimit check               = Passed
  42. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Array host connect type    = Single hostname per node
  43. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Master IP address [1]      = ::1
  44. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Master IP address [2]      = 192.168.4.93
  45. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Master IP address [3]      = fe80::225:90ff:fe3b:86c2
  46. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Standby Master             = Not Configured
  47. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Primary segment #          = 2
  48. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Total Database segments    = 4
  49. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Trusted shell              = ssh
  50. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Number segment hosts       = 2
  51. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Mirror port base           = 50000
  52. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Replicaton port base       = 41000
  53. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Mirror replicaton port base= 51000
  54. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Mirror segment #           = 2
  55. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Mirroring config           = ON
  56. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Mirroring type             = Group
  57. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:----------------------------------------
  58. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Greenplum Primary Segment Configuration
  59. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:----------------------------------------
  60. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-h93      /home/wslu/gp/gpsql/data/primary/gpseg0        40000          2          0       41000
  61. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-h93      /home/wslu/gp/gpsql/data/primary/gpseg1        40001          3          1       41001
  62. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-h94      /home/wslu/gp/gpsql/data/primary/gpseg2        40000          4          2       41000
  63. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-h94      /home/wslu/gp/gpsql/data/primary/gpseg3        40001          5          3       41001
  64. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:---------------------------------------
  65. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-Greenplum Mirror Segment Configuration
  66. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:---------------------------------------
  67. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-h94      /home/wslu/gp/gpsql/data/mirror/gpseg0          50000          6          0       51000
  68. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-h94      /home/wslu/gp/gpsql/data/mirror/gpseg1          50001          7          1       51001
  69. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-h93      /home/wslu/gp/gpsql/data/mirror/gpseg2          50000          8          2       51000
  70. 20160114:14:30:29:005980 gpinitsystem:h93:wslu-[INFO]:-h93      /home/wslu/gp/gpsql/data/mirror/gpseg3          50001          9          3       51001
  71. Continue with Greenplum creation Yy/Nn>
  72. y
  73. 20160114:14:30:32:005980 gpinitsystem:h93:wslu-[INFO]:-Building the Master instance database, please wait...
  74. 20160114:14:31:08:005980 gpinitsystem:h93:wslu-[INFO]:-Starting the Master in admin mode
  75. 20160114:14:32:01:005980 gpinitsystem:h93:wslu-[INFO]:-Commencing parallel build of primary segment instances
  76. 20160114:14:32:01:005980 gpinitsystem:h93:wslu-[INFO]:-Spawning parallel processes    batch [1], please wait...
  77. ....
  78. 20160114:14:32:02:005980 gpinitsystem:h93:wslu-[INFO]:-Waiting for parallel processes batch [1], please wait...
  79. ...........................................................
  80. 20160114:14:33:01:005980 gpinitsystem:h93:wslu-[INFO]:------------------------------------------------
  81. 20160114:14:33:01:005980 gpinitsystem:h93:wslu-[INFO]:-Parallel process exit status
  82. 20160114:14:33:01:005980 gpinitsystem:h93:wslu-[INFO]:------------------------------------------------
  83. 20160114:14:33:01:005980 gpinitsystem:h93:wslu-[INFO]:-Total processes marked as completed           = 4
  84. 20160114:14:33:01:005980 gpinitsystem:h93:wslu-[INFO]:-Total processes marked as killed              = 0
  85. 20160114:14:33:01:005980 gpinitsystem:h93:wslu-[INFO]:-Total processes marked as failed              = 0
  86. 20160114:14:33:01:005980 gpinitsystem:h93:wslu-[INFO]:------------------------------------------------
  87. 20160114:14:33:01:005980 gpinitsystem:h93:wslu-[INFO]:-Commencing parallel build of mirror segment instances
  88. 20160114:14:33:01:005980 gpinitsystem:h93:wslu-[INFO]:-Spawning parallel processes    batch [1], please wait...
  89. ....
  90. 20160114:14:33:01:005980 gpinitsystem:h93:wslu-[INFO]:-Waiting for parallel processes batch [1], please wait...
  91. .........................................
  92. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:------------------------------------------------
  93. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:-Parallel process exit status
  94. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:------------------------------------------------
  95. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:-Total processes marked as completed           = 4
  96. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:-Total processes marked as killed              = 0
  97. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:-Total processes marked as failed              = 0
  98. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:------------------------------------------------
  99. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:-Deleting distributed backout files
  100. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:-Removing back out file
  101. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:-No errors generated from parallel processes
  102. 20160114:14:33:43:005980 gpinitsystem:h93:wslu-[INFO]:-Restarting the Greenplum instance in production mode
  103. 20160114:14:33:43:001932 gpstop:h93:wslu-[INFO]:-Starting gpstop with args: -a -i -m -d /home/wslu/gp/gpsql/data/master/gpseg-1
  104. 20160114:14:33:43:001932 gpstop:h93:wslu-[INFO]:-Gathering information and validating the environment...
  105. 20160114:14:33:43:001932 gpstop:h93:wslu-[INFO]:-Obtaining Greenplum Master catalog information
  106. 20160114:14:33:43:001932 gpstop:h93:wslu-[INFO]:-Obtaining Segment details from master...
  107. 20160114:14:33:43:001932 gpstop:h93:wslu-[INFO]:-Greenplum Version: ‘postgres (Greenplum Database) 4.3.99.00 build dev‘
  108. 20160114:14:33:43:001932 gpstop:h93:wslu-[INFO]:-There are 0 connections to the database
  109. 20160114:14:33:43:001932 gpstop:h93:wslu-[INFO]:-Commencing Master instance shutdown with mode=‘immediate‘
  110. 20160114:14:33:43:001932 gpstop:h93:wslu-[INFO]:-Master host=h93
  111. 20160114:14:33:43:001932 gpstop:h93:wslu-[INFO]:-Commencing Master instance shutdown with mode=immediate
  112. 20160114:14:33:43:001932 gpstop:h93:wslu-[INFO]:-Master segment instance directory=/home/wslu/gp/gpsql/data/master/gpseg-1
  113. 20160114:14:33:44:001932 gpstop:h93:wslu-[INFO]:-Attempting forceful termination of any leftover master process
  114. 20160114:14:33:44:001932 gpstop:h93:wslu-[INFO]:-Terminating processes for segment /home/wslu/gp/gpsql/data/master/gpseg-1
  115. 20160114:14:33:45:002019 gpstart:h93:wslu-[INFO]:-Starting gpstart with args: -a -d /home/wslu/gp/gpsql/data/master/gpseg-1
  116. 20160114:14:33:45:002019 gpstart:h93:wslu-[INFO]:-Gathering information and validating the environment...
  117. 20160114:14:33:45:002019 gpstart:h93:wslu-[INFO]:-Greenplum Binary Version: ‘postgres (Greenplum Database) 4.3.99.00 build dev‘
  118. 20160114:14:33:45:002019 gpstart:h93:wslu-[INFO]:-Greenplum Catalog Version: ‘300701081‘
  119. 20160114:14:33:45:002019 gpstart:h93:wslu-[INFO]:-Starting Master instance in admin mode
  120. 20160114:14:33:46:002019 gpstart:h93:wslu-[INFO]:-Obtaining Greenplum Master catalog information
  121. 20160114:14:33:46:002019 gpstart:h93:wslu-[INFO]:-Obtaining Segment details from master...
  122. 20160114:14:33:46:002019 gpstart:h93:wslu-[INFO]:-Setting new master era
  123. 20160114:14:33:46:002019 gpstart:h93:wslu-[INFO]:-Master Started...
  124. 20160114:14:33:46:002019 gpstart:h93:wslu-[INFO]:-Shutting down master
  125. 20160114:14:33:47:002019 gpstart:h93:wslu-[INFO]:-Commencing parallel primary and mirror segment instance startup, please wait...
  126. ........
  127. 20160114:14:33:55:002019 gpstart:h93:wslu-[INFO]:-Process results...
  128. 20160114:14:33:55:002019 gpstart:h93:wslu-[INFO]:-----------------------------------------------------
  129. 20160114:14:33:55:002019 gpstart:h93:wslu-[INFO]:-   Successful segment starts                                            = 8
  130. 20160114:14:33:55:002019 gpstart:h93:wslu-[INFO]:-   Failed segment starts                                                = 0
  131. 20160114:14:33:55:002019 gpstart:h93:wslu-[INFO]:-   Skipped segment starts (segments are marked down in configuration)   = 0
  132. 20160114:14:33:55:002019 gpstart:h93:wslu-[INFO]:-----------------------------------------------------
  133. 20160114:14:33:55:002019 gpstart:h93:wslu-[INFO]:-
  134. 20160114:14:33:55:002019 gpstart:h93:wslu-[INFO]:-Successfully started 8 of 8 segment instances
  135. 20160114:14:33:55:002019 gpstart:h93:wslu-[INFO]:-----------------------------------------------------
  136. 20160114:14:33:55:002019 gpstart:h93:wslu-[INFO]:-Starting Master instance h93 directory /home/wslu/gp/gpsql/data/master/gpseg-1
  137. 20160114:14:33:56:002019 gpstart:h93:wslu-[INFO]:-Command pg_ctl reports Master h93 instance active
  138. 20160114:14:33:56:002019 gpstart:h93:wslu-[INFO]:-No standby master configured.  skipping...
  139. 20160114:14:33:56:002019 gpstart:h93:wslu-[INFO]:-Database successfully started
  140. 20160114:14:33:59:005980 gpinitsystem:h93:wslu-[INFO]:-Completed restart of Greenplum instance in production mode
  141. 20160114:14:33:59:005980 gpinitsystem:h93:wslu-[INFO]:-Loading gp_toolkit...
  142. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-Scanning utility log file for any warning messages
  143. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-Log file scan check passed
  144. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-Greenplum Database instance successfully created
  145. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-------------------------------------------------------
  146. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-To complete the environment configuration, please
  147. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-update wslu .bashrc file with the following
  148. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-1. Ensure that the greenplum_path.sh file is sourced
  149. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-2. Add "export MASTER_DATA_DIRECTORY=/home/wslu/gp/gpsql/data/master/gpseg-1"
  150. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-   to access the Greenplum scripts for this instance:
  151. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-   or, use -d /home/wslu/gp/gpsql/data/master/gpseg-1 option for the Greenplum scripts
  152. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-   Example gpstate -d /home/wslu/gp/gpsql/data/master/gpseg-1
  153. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-Script log file = /home/wslu/gpAdminLogs/gpinitsystem_20160114.log
  154. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-To remove instance, run gpdeletesystem utility
  155. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-To initialize a Standby Master Segment for this Greenplum instance
  156. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-Review options for gpinitstandby
  157. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-------------------------------------------------------
  158. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-The Master /home/wslu/gp/gpsql/data/master/gpseg-1/pg_hba.conf post gpinitsystem
  159. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-has been configured to allow all hosts within this new
  160. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-array to intercommunicate. Any hosts external to this
  161. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-new array must be explicitly added to this file
  162. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-Refer to the Greenplum Admin support guide which is
  163. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-located in the /home/wslu/gp/gpsql/docs directory
  164. 20160114:14:34:02:005980 gpinitsystem:h93:wslu-[INFO]:-------------------------------------------------------
  165. [[email protected] gpsql]$

（5）       查看目录结构：

[python] view plain copy

  1. [[email protected] gpsql]$ ls data
  2. master  mirror  primary
  3. [[email protected] gpsql]$ ls data/master/
  4. gpseg-1
  5. [[email protected] gpsql]$ ls data/mirror/
  6. gpseg2  gpseg3
  7. [[email protected] gpsql]$ ls data/primary/
  8. gpseg0  gpseg1
  9. [[email protected] gpsql]$
  10. [[email protected] gpsql]$ ls data/
  11. master  mirror  primary
  12. [[email protected] gpsql]$ ls data/master/
  13. [[email protected] gpsql]$ ls data/primary/
  14. gpseg2  gpseg3
  15. [[email protected] gpsql]$ ls data/mirror/
  16. gpseg0  gpseg1
  17. [[email protected] gpsql]$

（6）       在h94初始化备master（主备master必须在不同主机，如果要配置单机多节点，则不能配置备master。这是因为目前主备master必须在相同目录，所以必然不同主机，如果端口不是5432，那么需要指定PGPORT）：

[python] view plain copy

  1. [[email protected] gpsql]$ PGPORT=5432 PGDATABASE=postgres gpinitstandby -s h94
  2. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-Validating environment and parameters for standby initialization...
  3. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-Checking for filespace directory /home/wslu/gp/gpsql/data/master/gpseg-1 on h94
  4. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:------------------------------------------------------
  5. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-Greenplum standby master initialization parameters
  6. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:------------------------------------------------------
  7. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-Greenplum master hostname               = h93
  8. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-Greenplum master data directory         = /home/wslu/gp/gpsql/data/master/gpseg-1
  9. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-Greenplum master port                   = 5432
  10. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-Greenplum standby master hostname       = h94
  11. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-Greenplum standby master port           = 5432
  12. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-Greenplum standby master data directory = /home/wslu/gp/gpsql/data/master/gpseg-1
  13. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-Greenplum update system catalog         = On
  14. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:------------------------------------------------------
  15. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:- Filespace locations
  16. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:------------------------------------------------------
  17. 20160114:14:40:47:003933 gpinitstandby:h93:wslu-[INFO]:-pg_system -> /home/wslu/gp/gpsql/data/master/gpseg-1
  18. Do you want to continue with standby master initialization? Yy|Nn (default=N):
  19. > y
  20. 20160114:14:40:53:003933 gpinitstandby:h93:wslu-[INFO]:-Syncing Greenplum Database extensions to standby
  21. 20160114:14:40:53:003933 gpinitstandby:h93:wslu-[INFO]:-The packages on h94 are consistent.
  22. 20160114:14:40:53:003933 gpinitstandby:h93:wslu-[INFO]:-Adding standby master to catalog...
  23. 20160114:14:40:53:003933 gpinitstandby:h93:wslu-[INFO]:-Database catalog updated successfully.
  24. 20160114:14:40:54:003933 gpinitstandby:h93:wslu-[INFO]:-Updating pg_hba.conf file...
  25. 20160114:14:41:00:003933 gpinitstandby:h93:wslu-[INFO]:-pg_hba.conf files updated successfully.
  26. 20160114:14:41:09:003933 gpinitstandby:h93:wslu-[INFO]:-Updating filespace flat files...
  27. 20160114:14:41:09:003933 gpinitstandby:h93:wslu-[INFO]:-Filespace flat file updated successfully.
  28. 20160114:14:41:10:003933 gpinitstandby:h93:wslu-[INFO]:-Starting standby master
  29. 20160114:14:41:10:003933 gpinitstandby:h93:wslu-[INFO]:-Checking if standby master is running on host: h94  in directory: /home/wslu/gp/gpsql/data/master/gpseg-1
  30. 20160114:14:41:11:003933 gpinitstandby:h93:wslu-[INFO]:-Cleaning up pg_hba.conf backup files...
  31. 20160114:14:41:17:003933 gpinitstandby:h93:wslu-[INFO]:-Backup files of pg_hba.conf cleaned up successfully.
  32. 20160114:14:41:17:003933 gpinitstandby:h93:wslu-[INFO]:-Successfully created standby master on h94

（7）       此时，h94的data/master目录就不为空了：

[python] view plain copy

  1. $ [[email protected] gpsql]$ ls data/master/
  2. gpseg-1

10. 测试

[python] view plain copy

  1. [[email protected] gpsql]$ psql -p 5432 postgres
  2. psql (8.3devel)
  3. Type "help" for help.
  4. postgres=#
  5. postgres=#
  6. postgres=# \db
  7. List of tablespaces
  8. Name    | Owner | Filespae Name
  9. ------------+-------+---------------
  10. pg_default | wslu  | pg_system
  11. pg_global  | wslu  | pg_system
  12. (2 rows)

至此，集群就初始化完成了。

集群初始化完成后，默认会启动集群。

---------------------------------------------------------华丽的分割线-------------------------------------------------------------

以下为Greenplum集群部分常用指令，以及关于配置单机多节点的一点说明。

11. 启动集群

以后，想手动启动集群，执行：

[python] view plain copy

  1. $ gpstart –a

这里说一下，每次想使用集群的任何指令，必须执行：

[python] view plain copy

  1. $ source greenplum-path.sh
  2. $ exportMASTER_DATA_DIRECTORY=/home/wslu/gp/gpsql/data/master/gpseg-1

下文不再赘述。

12. 停止集群

[python] view plain copy

  1. $ gpstop –a

13. 重启集群

[python] view plain copy

  1. $ gpstop –a –r

14. 查看集群状态

[python] view plain copy

  1. $ gpstate –m | -e

15. reload配置文件

在不停止集群情况下，若配置文件修改，reload配置文件：

[python] view plain copy

  1. $ gpstop –u

16. 维护模式下启动master

仅仅启动master来执行维护管理任务，不会影响segment中的数据。例如，在维护模式下你可以仅连接master实例的数据库并且编辑系统表设置。

1.运行gpstart使用-m选项：

[python] view plain copy

  1. $ gpstart –m

2.维护模式下连接master来维护系统表。例如：

[python] view plain copy

  1. $ PGOPTIONS=‘-c gp_session_role=utility‘psql template1

3.完成管理任务后，使master关闭工具模式。然后，重启进入正常模式：

[python] view plain copy

  1. $ gpstop -m

17. 访问数据库

可以使用psql连接集群：

[python] view plain copy

  1. [[email protected] gpsql]$ psql -p 5432 postgres
  2. psql (8.3devel)
  3. Type "help" for help.
  4. postgres=#

基本上常见命令与PostgreSQL的相同。可参考PostgreSQL官方手册和Greenplum官方手册。

18. GUC参数配置

使用gpconfig设置guc参数

[python] view plain copy

  1. $ gpconfig -c gp_vmem_protect_limit -v4096MB

gpconfig可以设置master和所有segment的guc参数，也可以使用--masteronly参数只设置master的参数。设置完guc参数后需要根据guc参数类型决定重启集群或reload配置文件。

显示guc参数

[python] view plain copy

  1. $ psql –c ‘showstatement_mem;’ 或 gpconfig –show statement_mem
  2. $ psql –c ‘show all;’ 或 gpconfig –l

19. 附录：关于配置单机多节点的说明

如果要配置单机多节点，比如只想在h93配置2个主segment、2个镜像segment、1个master，只需要把hostfile_config中的h94删掉，然后在h93删除data/primary，data/mirror，data/master目录下的内容，重新初始化即可。

时间： 2024-10-22 04:58:12

CentOS和Ubuntu下安装配置Greenplum数据库集群（包括安装包和源码编译安装）的相关文章

Centos 6.4下 MySQL配置主从服务(集群)

我们前面两篇文章都分别介绍了Mysql的安装及配置.备份及还原,今天我们继续前面的环境介绍一下Centos 6.4下MySQL配置主从实现数据同步,也可以理解为双机热备,来提高服务器的高可用性.对于Mysql的服务器配置就不多介绍了,如果有不明白的可以参阅前面的MySQL文章介绍: 配置MySQL主服务器的my.cnf文件 我们首先打开MySQL主服务器的配置文件 vim /etc/my.cnf 默认配置 我们需要添加以下内容到MySQL服务主配置文件内: #设置服务器ID,为1的表示为主服务器

Ubuntu下用hadoop2.4搭建集群（伪分布式）

要真正的学习hadoop,就必须要使用集群,但是对于普通开发者来说,没有大规模的集群用来测试,所以只能使用伪分布式了.下面介绍如何搭建一个伪分布式集群. 为了节省时间和篇幅,前面一些步骤不再叙述.本文是在基于单机模式的前提下进行得搭建.若不会搭建单机模式,请看我的前一篇文章.Ubuntu下用hadoop2.4搭建集群(单机模式) 第一步 配置hdfs-site.xml /usr/local/hadoop/etc/hadoop/hdfs-site.xml用来配置集群中每台主机都可用,指定主机上作为

配置mysql数据库集群

1.先在所有数据库服务器上安装mysql-cluster包 管理机:192.168.1.6 数据节点:192.168.1.31   192.168.1.32 sql节点:192.168.1.33    192.168.1.111 2.在管理节点上配置  [[email protected] ~]# vim /etc/config.ini [ndbd default]    //数据节点公共配置 noofreplicas=2   //数据备份数 datamemory=300M  //查询缓存 in

&lt;04&gt;VIM编辑器的使用和源码编译安装

<04>VIM编辑器的使用和源码编译安装 知识点:一. vim文本编辑器二.网络配置 三.文件查找 (locate find whereis which)四.解压/打包(zip gzip bzip2 tar)五.安装(rpm yum) 一. vim文本编辑器1.命令模式():即vim刚进去的时候,或者按一下ESC的时候的状态 yy 复制一行: 3yy:复制三行小p 粘贴到光标的下一行大P 粘贴到光标的上一行dd 删除一行: 3dd:删除三行u 撤销上一次操作,若想多次撤销,当然多按几下ushi

yum的更多用法和源码编译安装apache

一.yum的更多用法 上篇文章介绍了yum的基础用法及yum源的基本配置,除此之外yum还有更多的用法,以下将继续介绍yum的使用方法. 1.yum-config-manager工具,该工具可以自动生成yum的repo文件,而不需要手动创建编写,使用方法如下: 用法:yum-config-manager [options] [section] -add-repo:添加仓库 # 添加一个阿里云的yum仓库 yum-config-manager -add-repo=http://mirrors.al

CentOS 6.5 下 CDH 5.2.1 集群安装（一）

集群节点数量3 个 192.168.1.170 cdh-master 192.168.1.171 cdh-slave-1 192.168.1.171 cdh-slave-2 一.安装CentOS6.5 (64位)并搭建基本环境,包括: (1)添加sudo 权限 (2)修改主机名.网关.静态IP地址.DNS (3)关闭SELINUX.防火墙 可参考文章 (4)修改系统时区.配置ntp服务(首先可选宿主机器配置ntp server,参考文章) 二.克隆为三台机器,分别修改静态IP 地址.MAC.主机

Redis的安装配置及简单集群部署

最近针对中铁一局项目,跟事业部讨论之后需要我们的KF平台能够接入一些开源的数据库,于是这两天研究了一下Redis的原理. 1. Redis的数据存储原理及简述 1.1Redis简述 Redis是一个基于内存且支持持久化的key-value的NoSQL数据库,其中每个key和value都是使用对象表示的,具有以以下特征:多样数据类型.持久化.主从同步.它支持存储的value类型包括string(字符串).list(链表).hash(哈希).set(集合)和zset(有序集合).这些数据类型都支持p

一步一步安装配置Ceph分布式存储集群

Ceph可以说是当今最流行的分布式存储系统了,本文记录一下安装和配置Ceph的详细步骤. 提前配置工作 从第一个集群节点开始的,然后逐渐加入其它的节点.对于Ceph,我们加入的第一个节点应该是Monitor,我们设置为Monitor1. 我使用的操作系统是Ubuntu Server 19.10,目前没有对应的包地址,于是就是用了bionic,应该是18.04的.还有地址也换成了中科大的镜像http://mirrors.ustc.edu.cn/ceph/,而不是默认的https://downloa

Greenplum数据库集群

首选操作系统 Red Hat Enterprise Linux (RHEL)是首选操作系统.应该使用最新的受支持的主版本,当前是RHEL 6. 我使用的系统版本:centos7.6 文件系统 XFS是Greenplum数据库数据目录的最佳实践文件系统.XFS应该用下列选项挂载: ##官方推荐rw,noatime,inode64 我的实际配置: cat /etc/fstab /dev/mapper/centos-home /home xfs nodev,noatime,nobarrier,inod