Oracle Enterprise Manager Cloud Control 13.5 安装失败后重新安装

如果由于各种原因导致的安装过程中断需要进行重新安装的

①如果是虚拟机，首推利用快照进行还原进行重新安装

②如果不是虚拟机或者没有给虚拟机打快照，就需要进行卸载，并对数据库已经建立了的配置进行清空

卸载方法可参考oracle官方文档：https://docs.oracle.com/en/enterprise-manager/cloud-control/enterprise-manager-cloud-control/13.5/emadv/deinstalling-enterprise-manager-single-and-multi-oms-environments.html#

其中重要的内容如下：

1、Copy the deinstallation script from the Oracle home of the OMS host to a temporary or stage location.

cp <ORACLE_HOME>/sysman/install/EMDeinstall.pl <temporary_location>

For example,

cp /u01/software/em13c/oraclehome/sysman/install/EMDeinstall.pl /u01/tmp_deinstall

 

注意：此处的ORACLE_HOME实际为OMS_HOME  如我的目录为：/u01/app/em13c/middleware/sysman/install/EMDeinstall.pl

2、Deinstall the OMS.

<ORACLE_HOME>/perl/bin/perl <temporary_location>/EMDeinstall.pl -mwHome <ORACLE_HOME> -stageLoc <temporary_location>

You will be prompted for database credentials (SYS and SYSMAN) and the WebLogic Domain credentials. Enter the credentials and proceed with the deinstallation.

For example,

/u01/software/em13c/oraclehome/perl/bin/perl /u01/tmp_deinstall/EMDeinstall.pl -mwHome /u01/software/em13c/oraclehome -stageLoc /u01/tmp_deinstall

 

严格按照该语句执行，会自动卸载掉已装的OMS

3、将OMS配置所在数据库进行清空

如果不清空进行安装可能会有如下报错

 The referenced database doesn’t contain a valid Management Repository

 

 

 该提示原因是：Doc ID 2183332.1

This issue occurs when a previous upgrade attempt had failed during the configuration phase, with partial repository updates, and the repository was restored to a pre-upgrade state.

The /gc_inst directory was deleted, but the EM 13c system software binaries had been retained from the previous software only install.

也就是说之前oms孙然卸载了，但是前一次安装的信息仍然存在数据库中

Doc中给的方法如下：

The upgrade to 13c should be initiated from scratch to match the state of the restored EM 12.1.0.5 repository:

1) Deinstall the 13c OMS and remove the 13c installation directories.

2) Install an EM 13c Cloud Control system with the software-only method.

3) Upgrade the OMS to 13c using: <OMS Home>/sysman/install/ConfigureGC.sh

 

我按照这种方法操作后，仍然报该错误，后在别处查资料发现需要在OMS所配置的数据库

1. Drop sysman related schemas.   ---因为OMS会建sysman相关用户，所以需要删除上一次建立的sysman相关schema

DROP USER SYSMAN CASCADE;
DROP USER SYSMAN_OPSS CASCADE;
DROP USER SYSMAN_MDS CASCADE;
DROP USER SYSMAN_APM CASCADE;
DROP USER SYSMAN_RO CASCADE;
DROP USER SYSMAN_BIPLATFORM CASCADE;
DROP USER SYSMAN_STB CASCADE;

2. Remove Synonyms related to sysman accounts :   ---删除sysman相关SYNONYM

DECLARE
CURSOR l_syn_csr IS
SELECT 'DROP ' ||
CASE owner
WHEN 'PUBLIC'
THEN 'PUBLIC SYNONYM '
ELSE 'SYNONYM ' || owner || '.'
END ||
synonym_name AS cmd
FROM
dba_synonyms
WHERE
table_owner IN (
'SYSMAN',
'SYSMAN_MDS',
'MGMT_VIEW',
'SYSMAN_BIP',
'SYSMAN_APM',
'BIP',
'SYSMAN_OPSS',
'SYSMAN_RO'
);
BEGIN
FOR l_syn_rec IN l_syn_csr LOOP
BEGIN
EXECUTE IMMEDIATE l_syn_rec.cmd;
EXCEPTION
WHEN OTHERS THEN
dbms_output.put_line( '===> ' || l_syn_rec.cmd );
dbms_output.put_line( sqlerrm );
END;
END LOOP;
END;
/

3. Removing remaining Objects and tablespaces :   ---删除相关表空间

DROP USER mgmt_view CASCADE;
DROP TABLESPACE mgmt_ecm_depot_ts INCLUDING CONTENTS AND DATAFILES CASCADE CONSTRAINTS;
DROP TABLESPACE mgmt_tablespace INCLUDING CONTENTS AND DATAFILES CASCADE CONSTRAINTS;
DROP TABLESPACE mgmt_ad4j_ts INCLUDING CONTENTS AND DATAFILES CASCADE CONSTRAINTS;

4. As proper database cleaning using RepManager dropall didn't happen, so we have to clean up the registry details :   ---删除相关schema registry信息

DELETE
FROM
schema_version_registry
WHERE
(comp_name,owner) IN (
('Authorization Policy Manager','SYSMAN_APM'),
('Metadata Services','SYSMAN_MDS'),
('Oracle Platform Security Services','SYSMAN_OPSS')
);
commit;

5. Retry with the installation, it should complete successfully.

进行上述操作时可能会遇到 表或视图不存在的情况，无需理会

删除完成后再次安装就不会再有报错了