记一次应用GI PSU而犯的低级但又容易犯的错误

事情的经过时这样的，最近在装一台自己的Oracle 11.2.0.4的数据库，应用11.2.0.4.4的GI PSU，我的命令是：

opatch auto /soft/11.2.0.4.4/19380115/ -oh /soft/product/11.2.0.4/gih/,/soft/product/11.2.0.4/dbh -ocmrf /soft/11.2.0.4.4/ocm.rsp 

但是在应用GI PSU的过程中，我却总是遇到如下错误，安装包检查不通过：

Stopping CRS... Stopped CRS successfully Error : The opatch Applicable check failed. The patch /soft/11.2.0.4.4/19380115/19121549 is not applicable to /soft/product/11.2.0.4/gih Error:Patch Applicable check failed for /soft/product/11.2.0.4/gih Starting CRS... ERROR: Prereq checkApplicable failed. Refer log file for more details. opatch auto failed. 

不论命令怎么修改，这个错误总是如影随行，其实殊不知是因为自己忽略的一些东西而导致的，后面再告诉你是什么。

第一次遇到错误时，看了一下opatch的日志opatchauto2014-11-20_23-10-24.log，结果发现日志存在一下内容：

2014-11-19 00:49:44: Status of Applicable  check  for /soft/product/11.2.0.4/gih is 12014-11-19 00:49:44: Error:Patch Applicable check failed for /soft/product/11.2.0.4/gih2014-11-19 00:49:44: Executing cmd: /bin/rpm -q sles-release2014-11-19 00:49:45: Command output:>  package sles-release is not installed >End Command output2014-11-19 00:49:45: init file = /soft/product/11.2.0.4/gih/crs/init/init.ohasd2014-11-19 00:49:45: Copying file /soft/product/11.2.0.4/gih/crs/init/init.ohasd to /etc/init.d directory2014-11-19 00:49:45: Setting init.ohasd permission in /etc/init.d directory2014-11-19 00:49:45: init file = /soft/product/11.2.0.4/gih/crs/init/ohasd2014-11-19 00:49:45: Copying file /soft/product/11.2.0.4/gih/crs/init/ohasd to /etc/init.d directory2014-11-19 00:49:45: Setting ohasd permission in /etc/init.d directory2014-11-19 00:49:45: Executing cmd: /bin/rpm -q sles-release2014-11-19 00:49:45: Command output:>  package sles-release is not installed >End Command output

说是，我的系统缺失sles-release包。这个事SuSe的版本包呀，我的系统是RedHat的呀，虽然RHEL 6.5里面的redhat-release包被替换成了redhat-release-server-6Server的名字，但是这和它有什么关系呢？

另外一个问题，GI PSU应用失败和这个错误有关系吗？暂时不知道，但是日志里面目前只有这个Error，所以不管三七二十一，既然它报错了就把这个包给补上吧。这个包还真不好找，最后还是在MOS上找到了（ID 1621417.1），其实就是没有任何内容的空的rpm安装包而已（是否可以自己打包一个？？）。

本以为安装完这个补丁后问题能够被解决，呵呵，当然还是以应用PSU失败了，只不过现在不再出现找不到sles-release的错误而已。

那么问题到底出在哪呢？

我们都知道其实在使用opatch应用PSU时，它会把每一步操作都输出到：opatchauto2014-11-20_23-10-24.report.log 日志中。
那么我们就把这些命令挨个手工执行一下试试，先从prereq的升级前检查工作开始。注意，crs_home的check使用grid用户，rac_home的check使用oracle用户。

结果发现，当执行到：/soft/product/11.2.0.4/gih/OPatch/opatch prereq CheckApplicable -ph /soft/11.2.0.4.4/19380115/19121549 -oh /soft/product/11.2.0.4/gih 时报错了，而且这正好就是opatch auto报错退出的地方。

那么应该就是和这个地方有关了吧？我们先来看一下报错内容：

Patch 19121549:Copy Action: Source File "/soft/11.2.0.4.4/19380115/19121549/files/bin/appvipcfg.pl" does not exists or is not readable'oracle.crs, 11.2.0.4.0': Cannot copy file from 'appvipcfg.pl' to '/soft/product/11.2.0.4/gih/bin/appvipcfg.pl'Copy Action: Source File "/soft/11.2.0.4.4/19380115/19121549/files/bin/oclumon.bin" does not exists or is not readable'oracle.crs, 11.2.0.4.0': Cannot copy file from 'oclumon.bin' to '/soft/product/11.2.0.4/gih/bin/oclumon.bin'Copy Action: Source File "/soft/11.2.0.4.4/19380115/19121549/files/bin/ologgerd" does not exists or is not readable'oracle.crs, 11.2.0.4.0': Cannot copy file from 'ologgerd' to '/soft/product/11.2.0.4/gih/bin/ologgerd'Copy Action: Source File "/soft/11.2.0.4.4/19380115/19121549/files/bin/osysmond.bin" does not exists or is not readable'oracle.crs, 11.2.0.4.0': Cannot copy file from 'osysmond.bin' to '/soft/product/11.2.0.4/gih/bin/osysmond.bin'Copy Action: Source File "/soft/11.2.0.4.4/19380115/19121549/files/crs/demo/coldfailover/act_db.pl" does not exists or is not readable'oracle.crs, 11.2.0.4.0': Cannot copy file from 'act_db.pl' to '/soft/product/11.2.0.4/gih/crs/demo/coldfailover/act_db.pl'Copy Action: Source File "/soft/11.2.0.4.4/19380115/19121549/files/crs/demo/coldfailover/act_listener.pl" does not exists or is not readable'oracle.crs, 11.2.0.4.0': Cannot copy file from 'act_listener.pl' to '/soft/product/11.2.0.4/gih/crs/demo/coldfailover/act_listener.pl'Copy Action: Source File "/soft/11.2.0.4.4/19380115/19121549/files/crs/demo/coldfailover/act_resgroup.pl" does not exists or is not readable'oracle.crs, 11.2.0.4.0': Cannot copy file from 'act_resgroup.pl' to '/soft/product/11.2.0.4/gih/crs/demo/coldfailover/act_resgroup.pl'Copy Action: Source File "/soft/11.2.0.4.4/19380115/19121549/files/crs/demo/demoActionScript" does not exists or is not readable'oracle.crs, 11.2.0.4.0': Cannot copy file from 'demoActionScript' to '/soft/product/11.2.0.4/gih/crs/demo/demoActionScript'Copy Action: Source File "/soft/11.2.0.4.4/19380115/19121549/files/crs/install/tfa_setup.sh" does not exists or is not readable'oracle.crs, 11.2.0.4.0': Cannot copy file from 'tfa_setup.sh' to '/soft/product/11.2.0.4/gih/crs/install/tfa_setup.sh'Copy Action: Directory is not writeable: "/soft/product/11.2.0.4/gih"'oracle.crs, 11.2.0.4.0': Cannot copy file from 'libsrvm11.so' to '/soft/product/11.2.0.4/gih/oui/lib/linux/libsrvm11.so'

这个错误应就是权限问题，试了一下果然是权限问题，原来是因为grid用户没有权限访问这些文件：

[grid@MHAD2-11g ~]$ cp /soft/11.2.0.4.4/19380115/19121549/files/bin/appvipcfg.pl /tmpcp: cannot open `/soft/11.2.0.4.4/19380115/19121549/files/bin/appvipcfg.pl' for reading: Permission denied[grid@MHAD2-11g ~]$ l /soft/11.2.0.4.4/19380115/19121549/files/bin/appvipcfg.pl-rwxr-x--- 1 root root 9051 Oct  6 18:27 /soft/11.2.0.4.4/19380115/19121549/files/bin/appvipcfg.pl[grid@MHAD2-11g ~]$ 

这简直就是低级错误啊，原来因为opatch auto是需要使用root用户执行的，结果我就直接使用root用户区解压PSU的压缩包，然后生成的文件用户和组自然就是root用户的咯，如下：

[root@MHAD2 11.2.0.4.4]# ls -l /soft/11.2.0.4.4/|grep -v ziptotal 6108644drwxr-xr-x 5 root   root           4096 Oct 11 10:04 19380115-rw-r--r-- 1 root   root            621 Nov 20 23:09 ocm.rsp-rw-rw-r-- 1 root   root           2186 Oct 15 02:54 PatchSearch.xml[root@MHAD2 11.2.0.4.4]# 

虽然，opatch auto命令必须使用root用户执行，但是其实还是需要通过root用户su到grid和oracle用户来执行相关的检查工作，自然就会出现权限问题咯。

于是就把19380115目录改成了grid:oinstall用户组的属性，也可以删掉再使用grid用户重新解压文件。

[root@MHAD2 11.2.0.4.4]# chown -R grid:oinstall 19380115

然后，高高兴兴的重新执行opatch auto命令，这次应该没问题了吧。

但是，我还是太天真了，opatch又和我开了个玩笑，再次报错。好吧，有了之前的经验，这次就乖乖地去手工执行这些命令了。

结果，还是：/soft/product/11.2.0.4/gih/OPatch/opatch prereq CheckApplicable -ph /soft/11.2.0.4.4/19380115/19121549 -oh /soft/product/11.2.0.4/gih 报错：

Patch 19121549:Copy Action: Destination File "/soft/product/11.2.0.4/gih/crs/install/tfa_setup.sh" is not writeable.'oracle.crs, 11.2.0.4.0': Cannot copy file from 'tfa_setup.sh' to '/soft/product/11.2.0.4/gih/crs/install/tfa_setup.sh'Copy Action: Directory is not writeable: "/soft/product/11.2.0.4/gih"'oracle.crs, 11.2.0.4.0': Cannot copy file from 'libsrvm11.so' to '/soft/product/11.2.0.4/gih/oui/lib/linux/libsrvm11.so'

而且，还是权限的问题，哭了都。

第一个文件tfa_setup.sh没有权限是因为，这个文件是使用root用户执行opatch auto时生成的，是root用户的，所以当grid用户再次执行opatch要往这个文件中写东西时自然就没权限了。所以这个文件使用root用户直接删掉或者修改grid:oinstall的用户属组即可。

[root@MHAD2 11.2.0.4]# chown grid:oinstall /soft/product/11.2.0.4/gih/crs/install/tfa_setup.sh

那么第二错误又是怎么回事？grid用户无法写GI HOME?

原来GI安装完后执行root.sh脚本后会把GI HOME的属性用户改成root用户的：

[root@MHAD2 11.2.0.4]# ls -ltrtotal 8drwxr-xr-x 76 oracle oinstall 4096 Nov 20 23:11 dbh<span style="color:#ff0000;">drwxr-x--- 68 root   oinstall 4096 Nov 20 23:14 gih</span>

呵呵，原来这又是我犯的第二个低级错误，好吧先临时把其修改为grid用户的目录：

[root@MHAD2 11.2.0.4]# chown grid gih/[root@MHAD2 11.2.0.4]# ls -ltr total 8drwxr-xr-x 76 oracle oinstall 4096 Nov 20 23:11 dbhdrwxr-x--- 68 grid   oinstall 4096 Nov 20 23:14 gih

这次终于可以看到久违的成功信息了：

Stopping CRS...Stopped CRS successfullypatch /soft/11.2.0.4.4/19380115/19121551  apply successful for home  /soft/product/11.2.0.4/gih patch /soft/11.2.0.4.4/19380115/19121549  apply successful for home  /soft/product/11.2.0.4/gih patch /soft/11.2.0.4.4/19380115/19121552  apply successful for home  /soft/product/11.2.0.4/gih 

不过，只是部分成功，因为，我装个单机的数据库，木有RAC，所以不能同时写GI HOME和DB HOME：

Stopping RAC /soft/product/11.2.0.4/dbh ...Failed to stop resources from  database home /soft/product/11.2.0.4/dbhERROR: Refer log file for more details.

正确的写法是： 

opatch auto /soft/11.2.0.4.4/19380115/ -och /soft/product/11.2.0.4/gih/ -oh /soft/product/11.2.0.4/dbh -ocmrf /soft/11.2.0.4.4/ocm.rsp 

指成功应用了GI的PSU，现在只需要手工启动HAS然后，单独为DB HOME应用PSU即可：

[root@MHAD2 11.2.0.4.4]# crsctl start hasCRS-4123: Oracle High Availability Services has been started.[root@MHAD2 11.2.0.4.4]# opatch auto /soft/11.2.0.4.4/19380115/ -oh /soft/product/11.2.0.4/dbh -ocmrf /soft/11.2.0.4.4/ocm.rsp Stopping RAC /soft/product/11.2.0.4/dbh ...Stopped RAC /soft/product/11.2.0.4/dbh successfullypatch /soft/11.2.0.4.4/19380115/19121551  apply successful for home  /soft/product/11.2.0.4/dbh patch /soft/11.2.0.4.4/19380115/19121549/custom/server/19121549  apply successful for home  /soft/product/11.2.0.4/dbh Starting RAC /soft/product/11.2.0.4/dbh ...Started RAC /soft/product/11.2.0.4/dbh successfully

至此，问题是处理完了，总结一下就是：权限！权限！还是权限！