*项目aix5.3+hacmp5.4+oracle10gRAC

 总的配置过程分三步
    1.划分磁盘给做RAC的两台主机,在主机端识别磁盘阵列
    2.在两台主机之间安装HACMP,配置CONCURRENT VG
    3.在主机(aix)上配置 RAC

 

 

 

1.在DS4800 上准备空间给磁盘阵列

主机磁盘      lun(DS4800)       属性
hdisk2       ma-ocr              RAC OCR
hdisk3       ma-vote            RAC VOTE
hdisk4       ma-data01       RAC DATA
hdisk5        ma-rman          RMAN使用
hdisk6        ma-arch1        第一个节点的归档
hdisk7        ma-arch2        第二个节点的归档

 一个盘一个盘指派,然后再在主机上识别
主机识别的命令为
cfgmgr -v

2.安装配置HA

2.1 安装HACMP 软件
插入操作系统的光盘，选择从光盘打补丁
lslpp -l bos.adt.lib bos.adt.libm bos.adt.syscalls bos.net.tcp.client bos.net.tcp.server 
lslpp -l bos.rte.SRC bos.rte.libc bos.rte.libcfg bos.rte.libcur bos.rte.libpthreads bos.rte.odm 　　　　        bos.data
lslpp -l bos.rte.lvm bos.clvm.enh
这些补丁包也要安装，
特别注意，需要安装这几个软件bos.data, * rsct.compat.basic.hacmp 2.5.0.0
从AIX第三张光碟安装
插入 ha 的光碟，选择从光碟安装

挂载光盘

/usr/sbin/mount -rv cdrfs /dev/cd0 /dvd
Smitty installp
选择安装ha软件:
参照文档　hacmp_manual.pdf 选择以下的部分

   |   | * rsct.compat.basic.hacmp 2.5.0.0   第三张
  
注意：license aggree一定要选中。
 即可成功安装ＨＡＣＭＰ　软件。

2.配置HA

参考文档Hacmp_mannul.pdf

记得配置  concurrentvg,(datafile ,controlfile,redo都对应了其中的lv)

 

 

3.安装RAC ，参照 文档 CookBook_V1.10gRAC_R1_AIX5L.pdf
安装  CRS( 可以两种选择方式，本例采用的方式是 在concurrent vg上建立两个lv,分别对应 ocr，vote )
3.1 安装时间服务器
参照下面的文档
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||、、
AIX配置时间服务器及客户机实战笔记
 xntpd是关于网络时间协议的守护进程，它遵循了因特网时间服务器的通用标准。在启动 xntpd 时, xntpd 会读取 /etc/ntp.conf 配置文件来确定网络中系统时钟服务器，以 ntp 服务器的系统时间为标准，来调整本机的系统时间。

可以用 ntpq 命令来显示 xntpd 进程的内部变量。使用 ntp 时应注意，xntpd 服务器和 xntpd 客户端的时钟不能相差超过1000秒。若有大于1000秒的偏移，在客户端启动 xntpd 守护进程前，用 data 命令或 ntpdate 命令调整本机的系统时间，使偏移量在1000秒之内。然后启动 xntpd.

1. 构造一个NTP环境的基本步骤:

1.1 设置NTP 服务器(以192.168.5.2为例), 其它NTP客户服务器以此服务器的时间为准,与其进行时间同步.
1.1.1 编辑 /etc/ntp.conf 文件, 内容如下:
----------------------------
#broadcastclient
server 127.127.1.0
driftfile /etc/ntp.drift
tracefile /etc/ntp.trace
----------------------------
请注意文件中的 server 127.127.1.0 这一行, 此处的127.127.1.0 是一特殊的地址,表示NTP主服务器是与自身的系统时钟同步.

1.1.2 编辑好 /etc/ntp.conf后, 启动xntpd守护进程
# startsrc -s xntpd
也可通过调用smitty , 使 xntpd 在以后重启服务器时能自动启动.
# smitty xntpd

1.1.3 xntpd 状态查询 , 使用 #lssrc -ls xntpd
刚启动xntpd时, sys peer 为 'insane', 表明xntpd还没有完成同步, .
#lssrc -ls xntpd
Program name: --/usr/sbin/xntpd
Version: -------3
Leap indicator: 11 (Leap indicator is insane.) Sys peer: ------no peer, system is insane ...

等待 6 - 10 分钟后, sys peer 就不再是 'insane' 了.
#lssrc -ls xntpd
Program name: --/usr/sbin/xntpd
Version: -------3
Leap indicator: 00 (No leap second today.) Sys peer: ------127.127.1.0 ...

1.2. NTP客户端的设置
1.2.1 编辑 NTP 客户端上的 /etc/ntp.conf文件, 内容如下:
----------------------------
#broadcastclient
server 192.168.5.2
driftfile /etc/ntp.drift
tracefile /etc/ntp.trace
----------------------------
其中的 server 192.168.5.2 表明, 此客户端与IP地址为 192.168.5.2 的NTP服务器进行时间同步.

1.2.2 在NTP客户端启动xntpd守护进程
# startsrc -s xntpd
也可通过调用smitty , 使 xntpd 在以后重启服务器时能自动启动.
# smitty xntpd

1.2.3 查询xntpd的状态
当 system peer 不为 'insane' 时, 表明客户端已与服务器端成功地进行了同步.
# lssrc -ls xntpd
Program name: --/usr/sbin/xntpd
Version: -------3
Leap indicator: 00 (No leap second today.) Sys peer: ------192.168.5.2 ...

1.3. NTP客户端查询NTP服务器
ntp客户端使用 ntpdate 命令来确认是否可用指定的 ntp 服务器进行时间同步。
命令的结果会显示客户端与服务器的时间偏移。
#ntpdate -d 192.168.5.2
...
18 Jan 15:49:57 ntpdate[98462]: adjust time server 192.168.5.2 offset -0.000868
2 sec

若时间间隔大于1000秒，使用ntpdate 进行手工调整, 如：
#date
Thu Jan 18 15:52:00 BEIST 2007

#ntpdate 192.168.5.2
18 Jan 15:49:57 ntpdate[23832]: step time server 9.185.43.189 offset 23.40260
7 sec

#date
Thu Jan 18 15:52:00 BEIST 2007
完成上述操作后 ntp 客户端与 ntp 服务器系统时钟完成同步，ntp 客户端得到了新的系统时间。

查看同步的时间间隔（poll）
# xntpdc -c peers
    remote           local    st poll reach   delay offset disp
=======================================================================
*192.168.5.2     192.168.7.1    4 64   377 0.00230 -0.000426 0.00026
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

3.2.在concurrent vg 上建立lv,并修改lv 属主为 oracle:dba ，权限为 660

3.3 安装 CRS
    解压缩 patch 6718715 执行 rootpre.sh，在两个节点都要执行
    安装过程中曾经出现错误
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Expecting the CRS daemons to be up within 600 seconds.
Failure at final check of Oracle CRS stack.
10
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
按照metalink 文档    725878.1  ，

通过下面的步骤来验证是否是 init.d的问题
This can be verified by adding a simple entry in /etc/inittab:
test:2:once:/usr/bin/echo "HELLO TEST" > /tmp/test.log
run "init q" as root user. If the init is working, then there should be a file /tmp/test.log generated.

通过下面步骤来解决 init.d起不来的问题

1. Starting the script install_assist (AIX GUI utility Installation Assistance)
2. Updating for example the date, then exit install_assist properly
3. Reboot the system
After that daemon process in /etc/inittab started, CRS installation completed.

     3.4 安装DB

附录; metalink相关的内容

TIP:           Click help for a detailed explanation of this page.
        书签         转到末尾

主题:         Root.sh failed at Failure at final check of Oracle CRS stack 10
          文档 ID:         725878.1         类型:         PROBLEM
          上次修订日期:         16-JAN-2009         状态:         PUBLISHED

In this Document
  Symptoms
  Changes
  Cause
  Solution
  References

Applies to:
Oracle Server - Enterprise Edition - Version: 10.2.0.1
This problem can occur on any platform.
Symptoms

2 node RAC, performing CRS 10.2.0.1 installation, failure at root.sh, CRS stack not started

WARNING: directory '/oracle' is not owned by root
Checking to see if Oracle CRS stack is already configured
Setting the permissions on OCR backup directory
Setting up NS directories
Oracle Cluster Registry configuration upgraded successfully
WARNING: directory '/oracle' is not owned by root
Successfully accumulated necessary OCR keys.
Using ports: CSS=49895 CRS=49896 EVMC=49898 and EVMR=49897.
node <nodenumber>: <nodename> <private interconnect name> <hostname>
node 1: ndb1 ndb1-rac ndb1
node 2: ndb2 ndb2-rac ndb2
Creating OCR keys for user 'root', privgrp 'system'..
Operation successful.
Now formatting voting device: /dev/rhdisk3
Format of 1 voting devices complete.
Startup will be queued to init within 30 seconds.
Adding daemons to inittab
Adding daemons to inittab
Expecting the CRS daemons to be up within 600 seconds.
Failure at final check of Oracle CRS stack.
10

Checking "ps -ef | grep init" during this 600 seconds period, see there is no crs related init processes, eg: init.crsd/init.evmd/init.cssd running. When no processes are detected, then this case apply.
Changes
New installation
Cause

This particular case is caused by the OS init system does not working.

" Failure at final check of Oracle CRS stack.
10"
means CRS daemon did not startup during 600 seconds period.

In root.sh script, it will add CRS related entry in /etc/inittab, run "init q" and expect those 3 CRS related daemon processes to start. With init system problem, none of these daemon processes spawned, this caused CRS process startup failure as they rely on init daemon process to start first.

This can be verified by adding a simple entry in /etc/inittab:
test:2:once:/usr/bin/echo "HELLO TEST" > /tmp/test.log
run "init q" as root user. If the init is working, then there should be a file /tmp/test.log generated.
Solution

Please consult with system administrator for init issue.

e.g. here the solution reference only valid for AIX platform:
1. Starting the script install_assist (AIX GUI utility Installation Assistance)
2. Updating for example the date, then exit install_assist properly
3. Reboot the system
After that daemon process in /etc/inittab started, CRS installation completed.

For other platforms, please consult your system admin or vendor for its solution.
References
Note 761259.1 - How to Check the Clusterware Processes via the ps Command
Keywords
INSTALLATION~FAILS ; RAC ; INIT ; PROCESS~STARTUP ;
Help us improve our service. Please email us your comments for this document. .