Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Migrated Rows

Re: Migrated Rows

From: Pete's <empete2000_at_yahoo.com>
Date: 15 Jun 2001 13:16:42 -0700
Message-ID: <6724a51f.0106151216.7c10fe0d@posting.google.com>

Not necessarily, check out metalinks, Oracle Doc id: 122020.1, which is a discussion of the difference between Row Migration and Row Chaining. Most likely, you will only be able to fix Row Migration, not Chaining. Basically, Row Migration is due to an update occuring on a row and there is not enough free space is left in the block. Oracle basically creates a pointer in the original block point to where the row now exists. Row Chaining is where a row does not fit in an entire data block, usually due to the use of Long Raw, Long and such.

HTH,
Pete's

Norman Dunbar <ndunbar_at_lynxfinancialsystems.co.uk> wrote in message news:<F43E6BAE5BB5D411A44C00805FBE740DA14FD4_at_apps.leeds.lfs.co.uk>...
> Howard,
>
> Yeah, I've analysed with compute stats on all the tables and I've
> detected the chained rows ok. I'm led to believe that the V$SYSSTAT view
> tells me that I have migrated rows. I can easily identify and unchain
> the chained ones, but I wanted to know how to identify the migrated ones
> - none of the manuals seem to say how.
>
> However, it would apear that if I fix the chained rows, I'll (probably)
> fix the migrated ones as well, so no worries - apart from corectly
> determining the PCTFREE/PCTUSED stuff which someone (not me !) has set
> to 1% and 90% - which implies that these tables don't update or delete
> rows very often/at all.
>
> Cheers,
> Norman.
>
> -----Original Message-----
> From: Howard J. Rogers [mailto:howardjr_at_www.com]
> Posted At: Thursday, June 14, 2001 10:26 PM
> Posted To: server
> Conversation: Migrated Rows
> Subject: Re: Migrated Rows
>
>
> Assuming you have calculated statistics on your tables in the recent
> past, a
> simple select table_name from dba_tables where chain_cnt <> 0 would do
> the
> trick. That of course will also pick up any tables that have row
> chaining
> as well as row migration, but at least it would be a start.
>
> But I suspect you want something a bit cleverer than that.... in which
> case,
> er... I'm at a loss.
Received on Fri Jun 15 2001 - 15:16:42 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US