Re: _rollback_segment_count side effects

Greg.
You've just proved my point (though from the opposite direction to the OP). I've stated a couple of POSSIBLE side effects, now you want to know exactly why you're seeing neither.

If the OP had said: "this is what my system looks like and this is what I do, what would the side effects be if I set the -_rollback_segment_count to 5000?" I might have given no answer (because the side-effects I can think of wouldn't apply); or I might have said "because you're doing XYZ then ABC can happen when ...."

What you're now asking is "what are all the reasons why I'm NOT seeing XYZ" - and you didn't even tell me whether or not you had set the parameter.

Regards
Jonathan Lewis

On Wed, 21 Apr 2021 at 14:58, GG <grzegorzof_at_interia.pl> wrote:

> Appologize for kind of takeover but how I'm not seeing this ORA-01628
> having below stats (also db 12.1):
>
> SELECT DISTINCT STATUS, SUM(BYTES)/1024/1024/1024 GB, COUNT(*) FROM
> DBA_UNDO_EXTENTS GROUP BY STATUS
> 2 /
>
> Status GB COUNT(*)
> --------- ---------- ----------
> UNEXPIRED 1601.99518 306973
> EXPIRED .745910645 2169
> ACTIVE 3.40631104 112
>
> SQL>
> SQL> SELECT MAX(TUNED_UNDORETENTION), MAX(MAXQUERYLEN),
> MAX(NOSPACEERRCNT), MAX(EXPSTEALCNT) FROM V$UNDOSTAT;
>
> MAX(TUNED_UNDORETENTION) MAX(MAXQUERYLEN) MAX(NOSPACEERRCNT)
> MAX(EXPSTEALCNT)
> ------------------------ ---------------- ------------------
> ----------------
> 863546 61543 0 26
>
> SQL> SELECT BEGIN_TIME, END_TIME, TUNED_UNDORETENTION, MAXQUERYLEN,
> MAXQUERYID, NOSPACEERRCNT, EXPSTEALCNT, UNDOBLKS, TXNCOUNT FROM V$UNDOSTAT;
>
> BEGIN_TIME Date TUNED_UNDORETENTION MAXQUERYLEN
> MAXQUERYID NOSPACEERRCNT EXPSTEALCNT UNDOBLKS TXNCOUNT
> ------------------- ------------------- ------------------- -----------
> ------------- ------------- ----------- ---------- ----------
> 2021-04-21 09:37:14 2021-04-21 09:46:11 863546 2418
> fcfpwturkq7dm 0 13 153064 193322
> 2021-04-21 09:27:14 2021-04-21 09:37:14 863062 1813
> fcfpwturkq7dm 0 15 137362 690064
>
> Regards.
> Greg
>
>

--
http://www.freelists.org/webpage/oracle-l