On February 3, 2005 11:08 am, Jan Wieck wrote:
> On 1/31/2005 11:50 AM, Darcy Buskermolen wrote:
> > On January 30, 2005 09:21 am, Hannu Krosing wrote:
> >> Hi
> >>
> >> I had to take replication on one of my DBs down for some days, and after
> >> restarting it I have problems with it cactching up.
> >>
> >> One of the problems I found is that sl_log_1 is cleaned up twice (and as
> >> it results on 2 seqscans over a table of tens of millions of rows, it is
> >> a real problem).
> >>
> >> The query that does that is constructed starting at 
> >> cleanup_thread.c:144 in src of slony1-1.0.5.
> >>
> >> Could it be that the other delete was actually meant for sl_log_2 ?
> >
> > Yes that is correct, and this has been corrected in CVS a few weeks back.
>
> Is that backpatched into the 1.0 branch?
Yes:
----------------------------
revision 1.13.2.5
date: 2005/01/12 03:15:36;  author: darcyb;  state: Exp;  lines: +2 -2

Copy down the delete from sl_log_[1-2] fix that was discoverd while debugging 
James Black's large sl_log table/performance issue.

----------------------------


>
>
> Jan
>
> >> Why it decides to do it using seqscan is another puzzle for me.
> >>
> >> The table is vacuum analysed quite often, but IIRC it has always done
> >> the deleting using seqscan.
> >>
> >> --------------
> >> Hannu
> >>
> >>
> >> _______________________________________________
> >> Slony1-general mailing list
> >> Slony1-general at gborg.postgresql.org
> >> http://gborg.postgresql.org/mailman/listinfo/slony1-general

-- 
Darcy Buskermolen
Wavefire Technologies Corp.
ph: 250.717.0200
fx:  250.763.1759
http://www.wavefire.com