RE: Question on NEXT_CHANGE#

From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Mon, 30 Dec 2013 11:44:47 +0000
Message-ID: <CE70217733273F49A8A162EE074F64D901DD37A6_at_exmbx05.thus.corp>


I did a couple of log file dumps as well (1.2.0.4), just to confirm that there was a fairly regular gap between the last SCN recorded for a redo change record in the file and next_change# recorded in the file header - and it is a persistent phenomenon.

As PS Thomas pointed out, the sequence of SCNs recorded in the redo log does show gaps (and I've often occasionally why an SCN sometimes jumps by 2 when I've been expecting an increase of 1 - though I've never looked very closely).

One thought that may be worth pursuing is that various actions have to update the control file - taking out CF enqueues while they do so - I don't think you see redo records for such changes, but perhaps the SCN is still incremented because of thses changes. If so the little gap in SCNs at the end of log file may simply reflect the changes made to the control file as Oracle updates the records relating to the old and new log files.

(A simple test - does the size of the gap change with the number of log files per log file group ?)

Regards
Jonathan Lewis
http://jonathanlewis.wordpress.com
_at_jloracle



From: Nassyam Basha [nassyambasha_at_gmail.com] Sent: 30 December 2013 06:12
To: Jonathan Lewis
Cc: ORACLE-L
Subject: Re: Question on NEXT_CHANGE#

Hi Jonathan,

Thanks for the reply. I used dump method over archive log sequence 31 and it has last SCN as 2432672 where am expecting 2432676. From the last redo record the hex value "0x0000.00251ea0" i have converted to decimal and the last value is indeed 2432672. From this method also we can't see any system change records?

--
http://www.freelists.org/webpage/oracle-l
Received on Mon Dec 30 2013 - 12:44:47 CET

Original text of this message