Re: Odd behavior when recovering with backup controlfile

This is expected behaviour when recovering using backup controlfile. The syntactic direction to use backup controlfile implies an incomplete recovery. Since you told it to use a backup controlfile, the recovery mechanism cannot be sure where the end of redo is, so you have to tell it when to stop and then do an "open resetlogs".

This is not a bug at all. The simple workaround is to copy the online redolog as you did and point the recover to it. No sweat. Why didn't you just clone everything, including the non-backup controlfile and just to a normal recovery?

• ricky

Jay wrote:
>
> Hi,
>
> I'm testing recovering with a hot backup, and I'm finding something odd.
>
> Environment:
> HP-UX v11.0.0.0
> DB: 8.1.6.0.0
>
> I want to preface this question with, I want to take my entire backup to
> another machine to perform recovery, I don't want to rely on having access
> to the original files. My hot backup goes thru the normal drill:
>
> alter system switch log file
>
> loop <thru all tablespaces>
> alter tablespace begin backup
> os cp of datafiles for tablespace
> alter tablespace end backup
>
> end loop
>
> backup controlfile to '...'
> backup controlfile to trace
> alter system switch log file
>
> Now I test recovery.
> - Shutdown the database
> - restore all the files from backup (including all archives and backup
> controlfile)
> - startup mount
> - recover database using backup controlfile.
>
> Now the odd behavior:
> During the backup I produce exactly 1 archive log (there is no activity on
> the database) - #467
> during recovery it prompts for #467 & #468 (#468 is the current online redo
> log)
> since #468 has not been archive my recover is unsuccessful.
>
> I AM able to recover:
> if I use the current controlfile (i.e. not replace it with the backup
> controlfile)
> if I point recover to my online redo log (i.e. at the prompt I specify the
> file
> location "/u03/oradata/sbox/redo03.dbf"
> if I use the "backup controlfile to trace" to recreate my controlfile.
>
> I don't understand:
> 1) Shouldn't the backup controlfile know that the #468 is my current online
> redo log?
> 2) How do I use my backup controlfile without having to point to my online
> redo log (I want to recover on a different machine)?
> 3) Is this a bug?
>
> Lastly, I have altered my script to perform the last log switch before
> backing up the controlfile, but I get the same results.
>
> Thank for your help
> Jay
Received on Sat May 12 2001 - 13:46:23 CDT