BUG #5798: Some weird error with pl/pgsql procedure

From: "Maxim Boguk" <Maxim(dot)Boguk(at)gmail(dot)com>
To: pgsql-bugs(at)postgresql(dot)org
Subject: BUG #5798: Some weird error with pl/pgsql procedure
Date: 2010-12-21 02:48:09
Message-ID: 201012210248.oBL2m9sp090535@wwwmaster.postgresql.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs


The following bug has been logged online:

Bug reference: 5798
Logged by: Maxim Boguk
Email address: Maxim(dot)Boguk(at)gmail(dot)com
PostgreSQL version: 8.4.4
Operating system: FreeBSD 7.2
Description: Some weird error with pl/pgsql procedure
Details:

While I testing my table compactor utility
(http://code.google.com/p/compacttable/), I sometime getting exeptionally
weird errors from pl/pgsql procedure.

2010-12-21 05:17:19 MSK 17263 postgres(at)hh from 192.168.1.129
[vxid:22/1898656 txid:20671567] [SELECT] ERROR: cannot extract system
attribute from virtual tuple
2010-12-21 05:17:19 MSK 17263 postgres(at)hh from 192.168.1.129
[vxid:22/1898656 txid:20671567] [SELECT] CONTEXT: PL/pgSQL function
"__clear_table_tail" line 50 at FOR over EXECUTE statement
2010-12-21 05:17:19 MSK 17263 postgres(at)hh from 192.168.1.129
[vxid:22/1898656 txid:20671567] [SELECT] STATEMENT:
SELECT * from __clear_table_tail('vacancy_view', 'public', 'count_',
25053, 25)

Now 50th line of the procedure is:
--trying update all tuples in current page range
FOR _new_tid IN EXECUTE 'UPDATE ONLY ' ||
quote_ident(arg_relnamespace) || '.' || quote_ident(arg_relname) || ' SET '
|| arg_field || ' = ' || arg_field || ' WHERE ctid=ANY($1) RETURNING ctid'
USING _current_tid_list LOOP

This error usually happen with another error from Slony side (this error
easy to understand and not an problem at all):
2010-12-21 05:15:08 MSK 24679 slony(at)hh from [local] [vxid:37/4336609
txid:20671079] [UPDATE] ERROR: could not serialize access due to concurrent
update
2010-12-21 05:15:08 MSK 24679 slony(at)hh from [local] [vxid:37/4336609
txid:20671079] [UPDATE] STATEMENT: update only "public"."vacancy_view" set
"count_"='1428' where "vacancy_id"='3568635';
...

I think there some strange concurrency issues with pl/pgsql and slony, and I
don't understand reason of that error.

This error not easy to reproduce but happens for me 3 times during tests on
130GB slave database run with my utility.

Anyone can enlighten me what happens here?

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Sandro Santilli 2010-12-21 10:08:40 BUG #5799: failed to find conversion function from unknown to text
Previous Message Peter Eisentraut 2010-12-20 22:55:21 Re: BUG #5795: 9.0.2 PDF needs editing