Thursday, November 22, 2012

dropping restore point on db with active data guard



in order to be able go back in case of a big issue with application releases, i always create 
on all db's a guaranteed restore point … both the normal physical standby database as the active data guard database.

typically i drop those after a couple of days

i wanted to do the same thing on the active data guard db 
however it didn't let me do :

received :

SQL>  drop restore point before9; drop restore point before9 *ERROR at line 1:ORA-16000: database open for read-only access
stopping the media recovery didn't help either same error. :

i will need to stop the database and mount it in order to delete the restore point .




Thursday, October 25, 2012

upgrade to 11203 some grains of sand




Yesterday we did the upgrade from 10.2.0.3 to 11.2.0.3 PSU 3 as discussed earlier on this blog using a parallel environment with a logical standby. Just before activating the logical standby after testing that everything was correctly applied i made a guaranteed restore point.

2012-10-24 02:38:23.921000 +02:00
Created guaranteed restore point BEFORE_PRIM


Everything went fine and after testing the switchovers of the 11.2.0.3 database it was the last step that had to be performed increase compatible to 11.2.0. 
First on the standby databases and then on the primary.
as explained here

This went perfectly fine on the standby databases.

Then it was the primary's turn….  


alter system set compatible = '11.2.0' scope=spfile ;


ok and then restart …

oracle complained about the fact that there was a restore point created when db was running with compatible 10.2.0.3….
ok no problem i'll change it back to 10.2.0.3 in spfile and  drop the restore point ….

hmmm the db compatible wasn't changed in the db but was in the spfile… 
ok no problem will create pile ….and adapt that one …

oracle says it cannot do that because the spfile is in 11.2.0 WTF…

ok no problem start in nomount  and 

create pfile from memory


create pile = '/tmp/initfile'  from memory ;

edit the file startup with compatible 10.2.0.3  and drop the restore point. 
remove all the _ parameters you didn't set and make sure that in case of RAC, node specific parameter are back... which wasn't the case ...

had manually define 

instance1.thread
instance2.thread=...
....


HTH if you encounter this issue 

note is was around 4 am in the morning when we encountered this ....

So bottom-line pre compatible version restore points prevent the compatible parameter to be set ....



Thursday, October 4, 2012

Oracle appraisal event

Wow this was neat,altough i missed a lot of the concert, and was send out through the wrong food/catering exit and had to walk 25 min to find my way to the shuttle bus which was apart from overloaded alsov stuck in traffic. ! Here are some pics



Security was nice :)