Vmware+Linux+10G RAC 安装中的问题

# ocfs2console

Configure Nodes --&gt Add --&gt 输入NODE名和IP --&gt OK --&gt Apply
出现如下错误：
    o2cb_ctl: Unable to access cluster service while creating node
        Could not add node node1

解决办法:

将/etc/ocfs2/下的不正确的cluster.conf文件删掉，重新用ocfs2console 配置

------------------------------------

ocfs2的一个问题

在10gR2的RAC下，node1主机上已经做了ocfs2的格式化，mount上来也没有问题，但node2 mount的时候就会报下面的错误：
ocfs2_hb_ctl: OCFS2 directory corrupted while reading uuid
mount.ocfs2: Error when attempting to run /sbin/ocfs2_hb_ctl: "Operation not permitted"

经过检查发现，这个问题是由于在为虚拟机创建磁盘的时候没有选择“allocate all disk space now”，导致ocfs在加入第二个节点时出错，关掉虚拟机删除该磁盘，重新创建一个立即分配空间的盘再格式化，两个节点都能正常mount了。

在正式的生产环境下，这个错误出现的概率应该是很小。但如今虚拟机使用的越来越广泛，有时候出现这种错误难免让人摸不着头脑。特别是在磁盘活动较频繁，或磁盘大小增长快速时，更有可能出现这种问题，因此建议ocfs这类盘还是使用预先分配所有空间为好。
-------------------

# rpm -Uvh compat-libstdc++-7.3-2.96.128.i386.rpm
warning: compat-libstdc++-7.3-2.96.128.i386.rpm: Header V3 DSA signature: NOKEY, key ID db42a60e
Preparing...
########################################### [100%]

file /usr/lib/libstdc++-3-libc6.2-2-2.10.0.so from install of compat-libstdc++-7.3-2.96.128 conflicts with file from package compat-libstdc++-296-2.96-138

由于系统中已经存在旧版本的补丁包，故加--force选项，进行强制安装。
[root@ahwss-bb2 app]# rpm -Uvh compat-libstdc++-7.3-2.96.128.i386.rpm --force
warning: compat-libstdc++-7.3-2.96.128.i386.rpm: Header V3 DSA signature: NOKEY, key ID db42a60e
Preparing...
########################################### [100%]

1:compat-libstdc++
########################################### [100%]
-------------------

在安装之前，通常要检查当前的环境，命令如下：

[oracle@rac1 cluvfy]$ /data/rhel5/clusterware/cluvfy/runcluvfy.sh stage -pre crsinst -n rac1,rac2 -verbose

在这个过程中，系统可能会报错，比如说什么包需要安装啊，swap区过小啊，这些东西统统不用管他们，跳过即可。下面开始安装clusterware：
安装时出现 中文乱码。

　　提示：Cannot convert string "-isas-song ti-medium-r-normal--*-140-*-*-c-*-gb2312.1980-0" to type FontStruct

　　解决方法： 用英文安装 export LC_ALL=en_US

 

1，以oracle用户进入到clusterware目录，以不进行系统环境检查的机制运行runInstaller：

切记runInstaller后的-ignoreSysPreReqs关键字是必不可少的，因为默认RedHat-5是不支持clusterware的安装的，我们必须过了这个山头才能安营扎寨。

# more /etc/redhat-release
Red Hat Enterprise Linux Server release 5 (Tikanga)
因为默认RHEL AS5不支持oracle 10.2,0.1,所以要修改一些参数。编辑/etc/redhat-release文件，将现在的版本信息（Red Hat Enterprise Linux Server release 5 (Tikanga）)，替换为（Red Hat Enterprise Linux Server release 4 (Tikanga）)，这样才可以通过oracle的安装条件检查。
注：网上说到另外一种方法如下：
通过修改安装文件oraparam.ini,添加redhat-5（二者选一即可）
# vi /10201_database_linux32/database/install/oraparam.ini
### #[Certified Versions]

再添加
[Linux-redhat-5.0-optional]TEMP_SPACE=80
SWAP_SPACE=150
MIN_DISPLAY_COLORS=256
按这种方法修改后，不能通过oracle的安装条件检查。
-------------------------------------

在metalink上是一个bug,
This assumption is based on an Internet standard defined in RFC 1918.

[url]http://www.faqs.org/rfcs/rfc1918.html[/url]

This issue is being tracked in BUG:4437727

Per BUG:4437727, cluvfy makes an incorrect assumption based on RFC 1918 that any IP address that begins with any of the following octets is private and hence may not be fit for being used as a VIP:

172.16.x.x  through 172.31.x.x
192.168.x.x
10.x.x.x

However, this assumption does not take into account that it is possible to use these IPs as Public IP's on an internal network  (or intranet).   Therefore, it is very common to use IP addresses in these ranges as Public IP's and as Virtual IP(s).

Solution
The solution to the error above that is given when running 'cluvfy' is to simply ignore it if you intend to use an IP in one of the above ranges for your VIP. The installation and configuration can continue with no corrective action necessary.

One result of this, as noted in the problem section, is that the silent VIPCA will fail at the end of the 10gR2 CRS installation.   This is because VIPCA is running in silent mode and is trying to notify that the IPs that were provided may not be fit to be used as VIP(s). To correct this, you can manually execute the VIPCA after the CRS installation is complete.  VIPCA needs to be executed from the CRS_HOME/bin directory as the 'root'  user  (on Unix/Linux)  or as a Local Administrator (on Windows):

$ cd $ORA_CRS_HOME/bin

$ ./vipca

Follow the prompts for VIPCA to select the appropriate interface for the public network, and assign the VIPs for each node when prompted.  Manually running VIPCA with the same IPs should complete successfully.
--------------------------------------------------------------

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/8797129/viewspace-694106/，如需转载，请注明出处，否则将追究法律责任。

转载于:http://blog.itpub.net/8797129/viewspace-694106/