Install patchset 10.2.0.4 with pysical Standby (data guard)
I searched some time to find the documentation which tell how to install a patch on a data guard configuration with a pysical standby.
You need this tow pieces of infomation:
- http://download.oracle.com/docs/cd/B19306_01/server.102/b14239/rollup.htm#i1029537
- Metalink Doc ID: 278641.1
My short summary for 10.2.0.3 to 10.2.0.4 upgrade:
1. primary: shutdown immedaite
2. primary: lsnrctl stop
3. standby: recover managed standby database cancel;
4. standby: shutdown immediate
5. standby: lsnrctl stop
6. primary: opatch apply
7. standby: opatch apply
8. standby: starup mount
9. standby: recover managed standby database nodelay disconnect;
10. primary: startup upgrade
11. primary: check log shipping, v$archive_dest.status
12. primary: alter system archive log current;
13. primary: select dest_id, status from v$archive_dest;
14. primary:
# disconnect & reconnet
SPOOL upgrade_info.log
@?/rdbms/admin/utlu102i.sql
SPOOL OFF
SPOOL patch.log
@?/rdbms/admin/catupgrd.sql
SPOOL OFF
SELECT COMP_NAME, VERSION, STATUS FROM SYS.DBA_REGISTRY;
15. alter system archive log current;
16. primary: shutdown / startup
17. standby: select max(sequence#) from v$log_history; — check recovery works
I don’t think, that it is good advice, because it bypass all runInstaller and upgrade assistant provided by Oracle, so increasing a chances of mistake (esp. if you have 4 - 6 databases on the server). It is good for senior DBA (but he don’t need this article) but not as a normal advice for average DBA (it is always better to use standard ways to do things and not do them on low level manually). Or, at least, it should describe both methods.
Comment by Alex R — January 19, 2010 @ 10:05 pm
To read the doucmentation is always mandatory - even for a junior dba - so it’s not risky at all.
Every change must be testet on a integration system first anyway - so you will find out if you miss something in advance.
Comment by admin — February 12, 2010 @ 3:14 pm