Обсуждение: Vacuum full error No one parent tuple

I have a summary table that I dump and refresh daily.  The refresh is done
via a stored_proc that inserts and updates this table from numerous other
tables. In the end, there are about 150k records in this table.  I usually
do a vacuum analyze on this table each night, but lately, my refresh
process has creeped up to about 20 mins.  So I tried a Vacuum Full Analyze.
The process stops and I get an error: No one parent tuple was found.  It
seems as though this is the only table I get this error.
What could be causing this?

Pg : 7.2
RedHat 7.2

TIA
Patrick


Patrick Hatcher
Macys.Com
Legacy Integration Developer

"Patrick Hatcher" <PHatcher@macys.com> writes:
> process has creeped up to about 20 mins.  So I tried a Vacuum Full Analyze.
> The process stops and I get an error: No one parent tuple was found.

This is a known bug in 7.2 --- IIRC, vacuum can get confused by some
situations involving rollback after SELECT FOR UPDATE.  It's pretty
harmless other than preventing you from using VACUUM FULL.

You can update to 7.3 to get the fix, or I think that just doing a
dummy SELECT * FOR UPDATE will clear the condition.

7.2.3 fixes some but not all cases of this, so an update to 7.2.3
might or might not help.

            regards, tom lane

thank you

Patrick Hatcher
Macys.Com
Legacy Integration Developer
415-422-1610 office
HatcherPT - AIM





                    Tom Lane
                    <tgl@sss.pgh.p       To:     "Patrick Hatcher" <PHatcher@macys.com>
                    a.us>                cc:     pgsql-novice@postgresql.org
                                         Subject:     Re: [NOVICE] Vacuum full error No one parent tuple
                    12/04/2002
                    12:25 PM





"Patrick Hatcher" <PHatcher@macys.com> writes:
> process has creeped up to about 20 mins.  So I tried a Vacuum Full
Analyze.
> The process stops and I get an error: No one parent tuple was found.

This is a known bug in 7.2 --- IIRC, vacuum can get confused by some
situations involving rollback after SELECT FOR UPDATE.  It's pretty
harmless other than preventing you from using VACUUM FULL.

You can update to 7.3 to get the fix, or I think that just doing a
dummy SELECT * FOR UPDATE will clear the condition.

7.2.3 fixes some but not all cases of this, so an update to 7.2.3
might or might not help.

                               regards, tom lane