Re: consistency check

I've a paper on my web site that says it's the best thing since crusty bread, melted butter and Vegemite swimming around on top (OK, it doesn't *quite* put it like that, but you get the idea). It's a fine tool. Anyone taking O/S backups and not scripting it in immediately afterwards to check what they've just backed up wants their head tested.

Now it's also nice that it can be used against online files... but that's an awful lot of I/O (potentially) to be throwing against a production system. So I tend not to recommend it against the live data files (there's no real harm, of course... just I/O) as a routine. (Having said that, it makes a fine diagnostic against *parts* of a datafile about which you have suspicions for other reasons.)

The only thing I've never liked about it, incidentally, is that if you forget to specify a blocksize parameter, it comes back and tells you something like 'the discovered blocksize 8192 does not match the default of 2048'. Which has always left me wondering why, if it can work out the actual blocksize, why the hell can't it use what it discovers instead of a crummy, out-of-date and utterly useless default like 2K???!

But that's like me quibbling that my grandmother dribbles slightly, instead of concentrating on the fact that she makes *excellent* crusty bread swimming with melted butter and Vegemite! Hopefully not at the same time, though.

Regards
HJR

--
----------------------------------------------
Resources for Oracle: http://www.hjrdba.com
===============================


"Paul Brewer" <paul_at_paul.brewers.org.uk> wrote in message
news:3c6d677d_3_at_mk-nntp-1.news.uk.worldonline.com...


> Howard,

>

> Any recommendation on dbv?

>

> TIA.

> Paul

>

> "Howard J. Rogers" <dba_at_hjrdba.com> wrote in message

> news:1013714253.358780_at_bugstomper.ihug.com.au...

> > Use RMAN to do your backups.  It checks for inconsistency as it does the

> > backups.  If your backup succeeds, you don't have any corruption.

> >

> > You can always consider things like DB_BLOCK_CHECKING, too (throws

errors


> > into the alert log if it encounters corruption).  I'd stay away from

> > DB_BLOCK_CHECKSUM like the plague, if I were you: it imposes a very

> > significant performance overhead.  The CHECKING version is not entirely

a


> > freebie, either, but it's much, much less of a killer than the older

> > CHECKSUM.

> >

> > Personally, if I was using RMAN, I wouldn't be using anything else.

> >

> > Regards

> > HJR

> > --

> > ----------------------------------------------

> > Resources for Oracle: http://www.hjrdba.com

> > ===============================

> >

> >

> > "Bud Socks" <bud_socks_at_gmx.net> wrote in message

> > news:a4gonq$g9p$00$1_at_news.t-online.com...

> > > Hi there,

> > >

> > > we have a 7x24h productive db environment. what would be the most

> > > recommended

> > > method to check block corruption ?

> > >

> > > Thanks

> > >

> > > Bud

> > >

> > >

> > >

> > >

> >

> >

>

>