Re: Oracle Cold Backup

"Md Irfan" <irfan_pk_at_hotmail.com> wrote in message news:c42168e7.0204182301.d0fc147_at_posting.google.com...
> Hi Mark -- Thanks for the suggestion. I was more likely to know if
> Oracle has
> a)commands for cold backup instead of using OS commands. Like for hot
> backup we have alter tbsp begin/end backup and cp the data files

And you know *why* we have those commands, right? Because the O/S is dumb: it can't be sure it grabs the datafile header block first as it performs the copy, and without that minor piece of critical information, we have no reliable way of telling the "age" of a datafile. So those commands were invented to finesse the problem by the simple expedient of locking the header block from update by CKPT.

Now, if the database is closed, and the backup is being performed cold, what is the chance of CKPT ever updating the datafile header block, given that CKPT doesn't actually exist, since this is a cold backup? And if there is no risk of CKPT stuffing things up in the middle of an O/S copy, why on Earth would we need a command to prevent it from doing so?

> b)in your solution for me, how about the control files?
>

Yeah, how about those control files! What about 'alter database backup controlfile to.......'? Stick the replacement for the dots in quotes, you get a binary backup. Replace the dots with the word 'trace' (without the quotes) and you get a script to re-construct the files.

Regards
HJR
> Thanks/Irfan
Received on Fri Apr 19 2002 - 06:23:44 CDT