Re: After I remove the online fuzzy tag of the datafile, can it be safe on standby?

Comments embedded.
AnySQL (d.c.b.a) wrote:
> Hi all:
>
> I developed a database file tool,

Really. By definition a tool is a device designed for the reliable performance of a task, and it occurs to me you've not met that criteria.

> when database is open, I do not put table into "begin backup" mode,

I see, so this 'tool' is your idea of a backup utility ...

> just clear the fuzzy tag on data file header, and make sure that no fractured block in the copied datafile,

You dramatically alter the datafile header by clearing the 'fuzzy' tag, without verifying any further information in the file, such as SCN consistency. This sounds more like a data shredder.

> seems like a cold copy, but SCN on some block may be newer than the datafile header's.

And with enough body work, suspension work, powertrain work, upholstery and paint a Yugo can seem like a Rolls-Royce, although it never will actually BE a Rolls-Royce. However, you've skipped straight to the paint work, and foregone any of the other tasks necessary to convert that Yugo to a possible pseudo-Rolls, and yet you're trying to sell it as though you have. Is this file copy made from a bonafide backup? No, by your own admission. Yet you want other people to accept your poor excuse for a product and go merrily on their way, trusting your modified files to contain their production data. I believe you need to return to reality (you remember, the place where people display a solid grasp of conceptual accuity), and possibly undergo some serious therapy, as you're obviously delusional.

> Is it safe for this file to be used in standby?

Personally I wouldn't use the file as a doorstop, much less as a component of a standby database. A standby database is intended to shorten recovery time in the event of a disaster, not create the disaster by using files modified by a 'tool' that only changes the headers without checking for other possible problems. Certainly it's possible to drive screws with a sledgehammer, but that isn't the tool specifically designed for the operation. It's also possible to use a boat anchor as a toothpick, presuming you can hoist it to your mouth and force it between your teeth; of course, that isn't very PROBABLE, for good reason. It isn't the proper implement for the task. Neither is your 'tool' the proper implement to create standby database files. And I know of no one who would jeopardise their service agreement, or their data, with your six-lane superhighway to ruin.

What, exactly, are you trying to prove with this fiasco you call a 'tool', prove yourself smarter than the Oracle developers? You are sorely mistaken if that is your aim. Are you trying to write a tool which performs the same function as RMAN, only without Oracle's 'high licencing costs'? You've missed the mark, sad to say, as you've done nothing except prove you know so little about Oracle as to be considered dangerous. One does not re-write datafile headers, nor does one try to prove himself intelligent by basically writing an automatic hex editor to modify data in a way Oracle did not intend. Such antics only create problems, not solutions. Altering a datafile header without the proper tools is asking for trouble, and possibly a lawsuit.  If you market this wonderous software package, purchase PLENTY of liability insurance. Trust me, you'll be makkng claims on the policy until the carrier cancels it, as an Oracle database isn't safe when this 'tool' is used.

David Fitzjarrell Received on Tue Jul 05 2005 - 07:39:40 CDT