There could be thousands (exaggerated) reasons why a Oracle cloning process could go all bad. I’m not an application DBA, however, have enough experience with the architecture, technology as I interact with it everyday as a part of my job.
Few months back, I started doing something what a DBA should do, cloning. My prior attempts were mostly at home, using virtual machines and test instances and they were NOT as mission critical as what we do at work.
So, after the storage device was revamped with new partition structures I was asked to do a cloning for the production instance. Let me explain how the application was deployed prior the storage restructuring
- We had the database tier on mount point /u05
- Application on /u06 mount point
So, I recreated the same mount points and started the cloning process for dbTier and the process got terminated at 2% and the log files shown me an error that I was not familiar with.
“ouicli.pl INSTE8_APPLY 1”
Google searches fetched me hundreds of results for “ouicli.pl INSTE8_APPLY”, however the error codes were mostly for 255 or “-1” and apparently I didn’t have any clue what was going wrong.
So I unzipped the tar ball for database tier once again, and the cloning process got aborted at 2%, and I was getting nervous as I was expected to make the instance online by early morning 7AM
Most of the reference materials were explaining about non-existent Oracle inventory locations, and I confirmed that it was not the case from my part (Obviously, I was overlooking at this constrain itself!)
After half a dozen times tasting failure, finally I tried to see what was written inside the oraInst.loc file
oraprod@erp-prod:/home/oraprod>cd $ORACLE_HOME oraprod@erp-prod:/u05/oraprod/PROD/db/tech_st/10.2.0>cat oraInst.loc inventory_loc=/u01/oraprod/PROD/db/tech_st/10.2.0/admin/oui/PROD_erp-prod/oraInventory
and I realized that inventory location was wrongly pointing towards an non-existing mount point and physical location!
I modified the oraInst.loc content with the correct mount point
and the cloning process went ahead without giving another errors.
We had an instance that was running from last 6 years, which was only once cloned from a cold backup during the storage device change, and somehow the inventory location remain unchanged with the repositories.
I hope this finding could help few newbies like me out there