Re: TXNCOUNT column from V$UNDOSTAT (Oracle 9.2.0.5, RAC on Solaris)

From: Matthias Hoys <idmwarpzone_NOSPAM__at_yahoo.com>
Date: Mon, 28 Nov 2005 16:48:08 +0100
Message-ID: <438b26ba$0$10297$ba620e4c@news.skynet.be>

"Bart the bear" <bartthebear_at_gmail.com> wrote in message news:1133192478.076485.317190_at_g47g2000cwa.googlegroups.com...
>I have an extremely high TXNCOUNT number for the period when
> the database is supposed to be mostly inactive. The fine manual
> defines the column like this:
>
> TXNCOUNT
>
> NUMBER
>
> Identifies the total number of transactions executed within the period
>
> What exactly is counted as a transaction? Below is the data that looks
> extremely suspicious to me:
> GIN_TIME END_TIME UNDOBLKS TXNCOUNT MAXQUERYLEN
> MAXCONCURRENCY SSOLDERRCNT
> 11/28/2005 03:42:02 11/28/2005 03:52:02 4 119287 5 1 0
> 11/28/2005 03:32:02 11/28/2005 03:42:02 1 119137 4 1 0
> 11/28/2005 03:22:02 11/28/2005 03:32:02 0 119090 3 0 1
> 11/28/2005 03:12:02 11/28/2005 03:22:02 11 119074 2 1 0
> 11/28/2005 03:02:02 11/28/2005 03:12:02 3 118924 3 1 0
>
> The reason for this is, of course the SSOLDERRCNT column which shows
> one @#$! ORA-1555 between
> 03:22 and 03:32. What I don't understand is what makes TXNCOUNT so high.
>

A large batch job with a high number of intermediate commits maybe ? Received on Mon Nov 28 2005 - 09:48:08 CST