This content has been marked as final. Show 4 replies
Either approach will work - how big are theses databases and how much down downtime can you afford ?1 person found this helpful
Starting with 126.96.36.199, "out-of-place" upgrades are recommended
Important Changes to Oracle Database Patch Sets Starting With 188.8.131.52 [ID 1189783.1]
Thanks - since the first one I am moving over is a test database, I just created the instance on the new server, shutdown the source test database then copied sp & control files & datafile directory structure over with data files, and decided to try DBUA rather than startup upgrade. Right now it's about 1/2 way through the upgrade so we'll see how it goes.
For the production database I am going to first try my cloning technique then do a startup upgrade or use DBUA. What I'm thinking about doing is copying my last (user managed) online backup over (after creating the instance & directories), doing a startup no mount, rebuild the controlfile to reflect any changes, recover with backup controlfile & apply latest redo, open with resetlogs then shutdown and startup upgrade (or DBUA).
If that works well that's probably the way I'll do the actual production roll over this way, but leave it in recover mode until I shutdown the old database and apply the last redolog, then upgrade it.
I suppose that given these databases are relatively small I could also do the upgrade first then just drop the schemas with tables in them and do a schema level impdp.
If you need to open your copied over database on the new server, then you will need to copy/clone the 184.108.40.206 software also (since you cannot open a 220.127.116.11 database using 18.104.22.168 binaries, unless you upgrade first). Also DBUA is highly recommended approach.
I think I ran into this situation awhile ago when I inadvertently cloned a 22.214.171.124 database to my 126.96.36.199 install (from an online backup) on my workstation as a test without thinking I was going to a different version.
I think I remember it allowing me to startup mount, create control file and apply redo, but when I tried to open with resetlogs I was getting errors, so I shut it down and if I recall I had to to a startup upgrade and eventually got it opened up.
Pretty sure this would not be an Oracle recommended approach though..