Re: Problem with ALTER TABLE - occasional "tuple concurrently updated" - Mailing list pgsql-bugs

From Jon Nelson
Subject Re: Problem with ALTER TABLE - occasional "tuple concurrently updated"
Date
Msg-id AANLkTikrmLMPjcZzMo2T-rknoQpt+e-hf=2uB44SNydd@mail.gmail.com
Whole thread Raw
In response to Re: Problem with ALTER TABLE - occasional "tuple concurrently updated"  (Robert Haas <robertmhaas@gmail.com>)
Responses Re: Problem with ALTER TABLE - occasional "tuple concurrently updated"  (Robert Haas <robertmhaas@gmail.com>)
List pgsql-bugs
On Wed, Nov 17, 2010 at 8:57 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> On Tue, Nov 16, 2010 at 10:48 AM, Jon Nelson <jnelson+pgsql@jamponi.net> =
wrote:
>> I have a process which runs in parallel creating tables which, as the
>> /final/ step in the import, gets SQL much like the following applied:
>>
>> ALTER TABLE foo INHERIT bar;
>>
>> Periodically, I get this error: =C2=A0tuple concurrently updated
>>
>> Of course, I googled for the error message and see a bunch of issues
>> involving ANALYZE and even DROP function.
>> Is this the same root cause? Is there a fix? Is there a lock I could
>> take or some other approach that would prevent the error?
>> I thought all ALTER TABLE statements took a big fat lock to prevent
>> such an issue.
>
> The ALTER TABLE generates that error? =C2=A0Is it running concurrently wi=
th
> any other DML? =C2=A0What version of PostgreSQL is this?

Yes, sometimes yes, and 8.4.5.

> That does sound like a bug.

That's what I thought!

--=20
Jon

pgsql-bugs by date:

Previous
From: Craig Ringer
Date:
Subject: Re: BUG #5756: postgres eat up all memory(2GB memory)
Next
From: "Andrey Galkin"
Date:
Subject: BUG #5757: PL/PGSQL composite variable field error reporting