Sequence Incrementing by 2 insted of 1

Started by Davidabout 17 years ago10 messagesgeneral
Jump to latest
#1David
david@vanlaatum.id.au

Our internal task database is doing something odd in that the sequence is incrementing by 2 instead of 1 and I can't find any reason why I have checked the
sequence itself to see if it had somehow got set to increment by 2 but no. The table in question has a number of both before and after triggers that do
various bits of business logic as well as some rewrite rules to track field changes. Using 8.3.0. Can anyone think of a reason why?

--

#2Thomas Kellerer
spam_eater@gmx.net
In reply to: David (#1)
Re: Sequence Incrementing by 2 insted of 1

David, 28.04.2009 07:48:

Our internal task database is doing something odd in that the
sequence is incrementing by 2 instead of 1 and I can't find any
reason why I have checked the sequence itself to see if it had
somehow got set to increment by 2 but no. The table in question has a
number of both before and after triggers that do various bits of
business logic as well as some rewrite rules to track field changes.
Using 8.3.0. Can anyone think of a reason why?

My first guess would be rolled back transactions

Thomas

#3A. Kretschmer
andreas.kretschmer@schollglas.com
In reply to: David (#1)
Re: Sequence Incrementing by 2 insted of 1

In response to David :

Our internal task database is doing something odd in that the sequence is incrementing by 2 instead of 1 and I can't find any reason why I have checked the
sequence itself to see if it had somehow got set to increment by 2 but no. The table in question has a number of both before and after triggers that do
various bits of business logic as well as some rewrite rules to track field changes. Using 8.3.0. Can anyone think of a reason why?

Wild guess: there are an other call to nextval() in one of the triggers.
But without the code it is hard to guess...

Regards, Andreas
--
Andreas Kretschmer
Kontakt: Heynitz: 035242/47150, D1: 0160/7141639 (mehr: -> Header)
GnuPG-ID: 0x3FFF606C, privat 0x7F4584DA http://wwwkeys.de.pgp.net

#4David
david@vanlaatum.id.au
In reply to: A. Kretschmer (#3)
Re: Sequence Incrementing by 2 insted of 1

Looked for that can't find anything and there is no reason why someone would have added that.

On Tue, Apr 28, 2009 at 08:24:46AM +0200, A. Kretschmer wrote:

In response to David :

Our internal task database is doing something odd in that the sequence is incrementing by 2 instead of 1 and I can't find any reason why I have checked the
sequence itself to see if it had somehow got set to increment by 2 but no. The table in question has a number of both before and after triggers that do
various bits of business logic as well as some rewrite rules to track field changes. Using 8.3.0. Can anyone think of a reason why?

Wild guess: there are an other call to nextval() in one of the triggers.
But without the code it is hard to guess...

Regards, Andreas
--
Andreas Kretschmer
Kontakt: Heynitz: 035242/47150, D1: 0160/7141639 (mehr: -> Header)
GnuPG-ID: 0x3FFF606C, privat 0x7F4584DA http://wwwkeys.de.pgp.net

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

--

#5Scott Marlowe
scott.marlowe@gmail.com
In reply to: David (#4)
Re: Sequence Incrementing by 2 insted of 1

On Tue, Apr 28, 2009 at 12:38 AM, David <david@vanlaatum.id.au> wrote:

Looked for that can't find anything and there is no reason why someone would have added that.

Well, unless you've got some example code for us to look at, it's
kinda hard to tell you what's wrong.

Often, you see people do something like this:

insert into autoincrementingtable values (DEFAULT, 'a', 500);
select nextval('autoinctable_seq');
yada yada yada

or something like that. i.e. there's an implicit call to the nextval
with the insert, and another explicit one somewhere in your code.

#6David
david@vanlaatum.id.au
In reply to: Scott Marlowe (#5)
Re: Sequence Incrementing by 2 insted of 1

Ok got permission to send the code I think I have got all the relevent bits.

On Tue, Apr 28, 2009 at 01:01:39AM -0600, Scott Marlowe wrote:

On Tue, Apr 28, 2009 at 12:38 AM, David <david@vanlaatum.id.au> wrote:

Looked for that can't find anything and there is no reason why someone would have added that.

Well, unless you've got some example code for us to look at, it's
kinda hard to tell you what's wrong.

Often, you see people do something like this:

insert into autoincrementingtable values (DEFAULT, 'a', 500);
select nextval('autoinctable_seq');
yada yada yada

or something like that. i.e. there's an implicit call to the nextval
with the insert, and another explicit one somewhere in your code.

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

--

Attachments:

sql.txttext/plain; charset=us-asciiDownload
#7Daniel Verite
daniel@manitou-mail.org
In reply to: David (#1)
Re: Sequence Incrementing by 2 insted of 1

David wrote:

the sequence is incrementing by 2 instead of 1 and I can't
find any reason why I have checked the
sequence itself to see if it had somehow got set to increment
by 2 but no.
The table in question has a number of both before and after triggers
that do various bits of business logic as well as some rewrite rules
to track field changes.
Using 8.3.0. Can anyone think of a reason why?

I'd say rewrite rules are your prime suspect.
Check out for example:
http://archives.postgresql.org/pgsql-sql/2007-03/msg00334.php

Cordialement,
--
Daniel

#8David
david@vanlaatum.id.au
In reply to: Daniel Verite (#7)
Re: Sequence Incrementing by 2 insted of 1

Cool thanks that makes sense was hard to pin down because we didn't notice for a while.

On Tue, Apr 28, 2009 at 09:31:26AM +0200, Daniel Verite wrote:

David wrote:

the sequence is incrementing by 2 instead of 1 and I can't
find any reason why I have checked the
sequence itself to see if it had somehow got set to increment
by 2 but no.
The table in question has a number of both before and after triggers
that do various bits of business logic as well as some rewrite rules
to track field changes.
Using 8.3.0. Can anyone think of a reason why?

I'd say rewrite rules are your prime suspect.
Check out for example:
http://archives.postgresql.org/pgsql-sql/2007-03/msg00334.php

Cordialement,
--
Daniel

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

--

#9Jasen Betts
jasen@xnet.co.nz
In reply to: David (#1)
Re: Sequence Incrementing by 2 insted of 1

On 2009-04-28, David <david@vanlaatum.id.au> wrote:

Our internal task database is doing something odd in that the sequence is incrementing by 2 instead of 1 and I can't find any reason why I have checked the
sequence itself to see if it had somehow got set to increment by 2 but no. The table in question has a number of both before and after triggers that do
various bits of business logic as well as some rewrite rules to track field changes. Using 8.3.0. Can anyone think of a reason why?

possibly a rule involves the the sequence, thus it's value is computed twice.

#10Jasen Betts
jasen@xnet.co.nz
In reply to: David (#1)
Re: Sequence Incrementing by 2 insted of 1

On 2009-04-28, David <david@vanlaatum.id.au> wrote:

Ok got permission to send the code I think I have got all the relevent bits.

CREATE TABLE currentcall (
cadc_taskno integer DEFAULT nextval(('currentcall_cadc_taskno_seq'::text)::regclass) NOT NULL,
coversight character varying(4),

...

CREATE RULE currentcall_insert AS ON INSERT TO currentcall DO INSERT INTO
taskchangelog (tasknum, newstatus, newtech, status, newpriority,
newestcomplete, newclass, newnewtask) VALUES (new.cadc_taskno,

^^^^^^^^^^^^^^^

new.cstatus, new.ctechalloc, new.ccustcomment, new.cfprior,
new.cestcomplete, new.cjobclass, (new.cnewtask = 'Y'::bpchar));

if you're inserting DEFAULT into the cadc_taskno column (or not
explicitly setting that column) the re-writer will substitute

nextval(('currentcall_cadc_taskno_seq'::text)::regclass)

And that will be evaluated once for the insert and again for the rule.

this is documented somewhere.... I think