From: vsevolod afanassiev on
I've done following test:
- Brought database down using 'shutdown abort'
- Renamed temp files to .bak
- Started database using 'startup'
Basically I wanted to know whether Oracle will complain about missing
temp files or it will ignore them
and then I'll be able to add them manually.

Instead Oracle re-created the files:

Thu Jan 21 16:26:57 2010
SMON: enabling tx recovery
Thu Jan 21 16:26:57 2010
Re-creating tempfile /eip/db04/oracle/EIP4T/oradata/EIP4T_temp_02.dbf
Re-creating tempfile /eip/db05/oracle/EIP4T/oradata/EIP4T_temp_01.dbf
Database Characterset is AL32UTF8
Opening with internal Resource Manager plan
where NUMA PG = 1, CPUs = 8
replication_dependency_tracking turned off (no async multimaster
replication found)
Starting background process QMNC

This is 10.2.0.4 on AIX
From: joel garry on
On Jan 21, 12:48 pm, vsevolod afanassiev
<vsevolod.afanass...(a)gmail.com> wrote:
> I've done following test:
> - Brought database down using 'shutdown abort'
> - Renamed temp files to .bak
> - Started database using 'startup'
> Basically I wanted to know whether Oracle will complain about missing
> temp files or it will ignore them
> and then I'll be able to add them manually.
>
> Instead Oracle re-created the files:
>
> Thu Jan 21 16:26:57 2010
> SMON: enabling tx recovery
> Thu Jan 21 16:26:57 2010
> Re-creating tempfile /eip/db04/oracle/EIP4T/oradata/EIP4T_temp_02.dbf
> Re-creating tempfile /eip/db05/oracle/EIP4T/oradata/EIP4T_temp_01.dbf
> Database Characterset is AL32UTF8
> Opening with internal Resource Manager plan
> where NUMA PG = 1, CPUs = 8
> replication_dependency_tracking turned off (no async multimaster
> replication found)
> Starting background process QMNC
>
> This is 10.2.0.4 on AIX

Also see bugs 1641989 and 5861994

In retrospect, this makes perfect sense, as tempfiles don't have
anything needed for recovery.

(Don't feel bad about not noticing this long ago, I'm sure I've seen
posts and ran across this on this along the way but had totally
forgotten about it until your post - RMAN makes things too easy :-)

jg
--
@home.com is bogus.
Tom Snyder and the Clash http://www.youtube.com/watch?v=JVygiX0KEEw
From: hpuxrac on
On Jan 21, 3:48 pm, vsevolod afanassiev
<vsevolod.afanass...(a)gmail.com> wrote:

snip

> I've done following test:
> - Brought database down using 'shutdown abort'
> - Renamed temp files to .bak
> - Started database using 'startup'
> Basically I wanted to know whether Oracle will complain about missing
> temp files or it will ignore them
> and then I'll be able to add them manually.
>
> Instead Oracle re-created the files:
>
> Thu Jan 21 16:26:57 2010
> SMON: enabling tx recovery
> Thu Jan 21 16:26:57 2010
> Re-creating tempfile /eip/db04/oracle/EIP4T/oradata/EIP4T_temp_02.dbf
> Re-creating tempfile /eip/db05/oracle/EIP4T/oradata/EIP4T_temp_01.dbf
> Database Characterset is AL32UTF8
> Opening with internal Resource Manager plan
> where NUMA PG = 1, CPUs = 8
> replication_dependency_tracking turned off (no async multimaster
> replication found)
> Starting background process QMNC
>
> This is 10.2.0.4 on AIX

Yup ... should be documented somewhere in your step by step procedures
to recreate a production database all the way using rman.

My stuff starts with "create diskgroup force" ( since we are using
ASM ) and ends with verifying stuff like this and recreating the
spfile inside the ASM diskgroup.

This was a change from 9 where you had to create the TEMP the way you
wanted it to be ... now it is remembered.