On May 15, 9:38 pm, Mladen Gogala <mgogala.SPAM_ME...._at_verizon.net>
wrote:
> On Tue, 15 May 2007 15:15:16 -0700, Ronny wrote:
> > I don't see a chicken and egg situation here: The standby database is
> > started in "nomount" state,
>
> Really? I was under the impression that you need to create
> a special controlfile by something like this:
>
> SQL> alter database create standby controlfile as '/tmp/oracle.ctl';
>
> Database altered.
>
> SQL>
>
> So, if the database is in "nomount" state, I have two questions:
> 1) How does it know where are the data files to recover? What is
> the starting SCN and which archive is it contained in?
> 2) What the heck do I need a standby controlfile for? Should I
> send it to Joel or Sybrand?
>
> The truth of the matter is that the database MUST be mounted for recovery
> to take place. In version 9, it was necessary to use "ALTER DATABASE
> MOUNT STANDBY DATABASE" command. Version 10 is better, it recognizes the
> standby controlfile and automatically mounts a standby if you execute
> "startup mount" command. If your database is not mounted, you have no
> standby. Of course, the command
> "alter database screw up managed standby database disconnect from session"
> must still be entered manually, for the recovery to start.
>
> --http://www.mladen-gogala.com
Truth is that some kind of remote copy could be easily implemented
(see rac installation) and made available/hidden through/behind RMAN.
I don't know why it is not implemented. Maybe because of ASM ?;) -
although even ASM should not be a huge obstacole.
Received on Tue May 15 2007 - 21:02:28 CDT
