From: gs on
I have a test database I often restore from an online user managed
backup (not rman), the procedure is simple:

shut test database down

copy files from last online backup on prod to test db locations, along
with any archived logs since backup

startup mount the test db, rebuild the controlfile renaming database and
new file locations etc.

recover database until cancel using backup controlfile, then apply redo
etc. cancel and open database with resetlogs, create a tempfile and I'm
good to go


Usually goes without a hitch, but for some reason after I hit enter to
have it apply the archived log it is looking for, I am getting:
===================================================================
ORA-00283: recovery session canceled due to errors
ORA-19906: recovery target incarnation changed during recovery


ORA-01112: media recovery not started
===================================================================

Looking up the 19906 error says its an rman incantation setting, but I
am not using rman at all for this, and I'm stumped why this is happening
now.

I even re-ran the online backup and recopied the files over, with no
success.


Any ideas what I'm missing here?


thanks





From: GS on
ps. database is 10.2.0.4 on windows 64 bit server.
From: joel garry on
On Sep 8, 7:06 pm, GS <g...(a)gs.com> wrote:
> ps. database is 10.2.0.4 on windows 64 bit server.

Wild guess after looking at metalink Note: 824213.1is that you did a
flashback of your prod database at some point, perhaps between when
you took the controlfile backup and the online backup? Any clues in
the prod alert log?

See http://download.oracle.com/docs/cd/B19306_01/backup.102/b14192/flashptr005.htm#sthref646
maybe if you specify an until time you can overcome an SCN ambiguity.

jg
--
@home.com is bogus.
Cash of the Titans: Amazon, Microsoft and Yahoo versus Google:
http://www3.signonsandiego.com/stories/2009/sep/09/sparring-rises-over-googles-book-deal/?uniontrib
From: gs on
joel garry wrote:
> On Sep 8, 7:06 pm, GS <g...(a)gs.com> wrote:
>> ps. database is 10.2.0.4 on windows 64 bit server.
>
> Wild guess after looking at metalink Note: 824213.1is that you did a
> flashback of your prod database at some point, perhaps between when
> you took the controlfile backup and the online backup? Any clues in
> the prod alert log?
>
> See http://download.oracle.com/docs/cd/B19306_01/backup.102/b14192/flashptr005.htm#sthref646
> maybe if you specify an until time you can overcome an SCN ambiguity.
>
> jg
> --
> @home.com is bogus.
> Cash of the Titans: Amazon, Microsoft and Yahoo versus Google:
> http://www3.signonsandiego.com/stories/2009/sep/09/sparring-rises-over-googles-book-deal/?uniontrib

Nope, no flashback done, and controlfile script is standard script I use
using backup controlfile to trace. After digging through the prod alert
log I am seeing a few of these:

Wed Sep 09 10:07:31 2009
Thread 1 cannot allocate new log, sequence 3255
Private strand flush not complete
Current log# 5 seq# 3254 mem# 0: E:\ORACLE\ORADATA\MYDB\MYDB\REDO5.LOG
Current log# 5 seq# 3254 mem# 1: D:\ORACLE\ORADATA\MYDB\REDO5A.LOG
Wed Sep 09 10:07:32 2009
Thread 1 advanced to log sequence 3255 (LGWR switch)
Current log# 6 seq# 3255 mem# 0: E:\ORACLE\ORADATA\MYDB\MYDB\REDO6.LOG
Current log# 6 seq# 3255 mem# 1: D:\ORACLE\ORADATA\MYDB\REDO6A.LOG

so I am wondering if this is somehow related.. looking at ways now to
cure this particular error, it happens when I do a manual logfile switch
as well.
From: gs on
gs wrote:
> joel garry wrote:
>> On Sep 8, 7:06 pm, GS <g...(a)gs.com> wrote:
>>> ps. database is 10.2.0.4 on windows 64 bit server.
>>
>> Wild guess after looking at metalink Note: 824213.1is that you did a
>> flashback of your prod database at some point, perhaps between when
>> you took the controlfile backup and the online backup? Any clues in
>> the prod alert log?
>>
>> See
>> http://download.oracle.com/docs/cd/B19306_01/backup.102/b14192/flashptr005.htm#sthref646
>>
>> maybe if you specify an until time you can overcome an SCN ambiguity.
>>
>> jg
>> --
>> @home.com is bogus.
>> Cash of the Titans: Amazon, Microsoft and Yahoo versus Google:
>> http://www3.signonsandiego.com/stories/2009/sep/09/sparring-rises-over-googles-book-deal/?uniontrib
>>
>
> Nope, no flashback done, and controlfile script is standard script I use
> using backup controlfile to trace. After digging through the prod alert
> log I am seeing a few of these:
>
> Wed Sep 09 10:07:31 2009
> Thread 1 cannot allocate new log, sequence 3255
> Private strand flush not complete
> Current log# 5 seq# 3254 mem# 0: E:\ORACLE\ORADATA\MYDB\MYDB\REDO5.LOG
> Current log# 5 seq# 3254 mem# 1: D:\ORACLE\ORADATA\MYDB\REDO5A.LOG
> Wed Sep 09 10:07:32 2009
> Thread 1 advanced to log sequence 3255 (LGWR switch)
> Current log# 6 seq# 3255 mem# 0: E:\ORACLE\ORADATA\MYDB\MYDB\REDO6.LOG
> Current log# 6 seq# 3255 mem# 1: D:\ORACLE\ORADATA\MYDB\REDO6A.LOG
>
> so I am wondering if this is somehow related.. looking at ways now to
> cure this particular error, it happens when I do a manual logfile switch
> as well.

update, according to metalink note:372557.1 this is normal expected
behavior so..