Undocumented Postgres error: "failed to fetch old tuple for AFTER trigger"
I'm running a very large series of commands - mainly DDL but some DML as
well - in a large transaction. I get the following error, which doesn't
seem to be documented:
"ERROR: failed to fetch old tuple for AFTER trigger
: COMMIT"
There are no triggers that I'm aware of. I've gotten this error when
running the transaction on two different instances of the database. I also
did a VACUUM FULL, which didn't help.
Breaking down the commands into a series of about 7 smaller transactions
works fine. Each transaction runs totally fine by itself. But, when I run
them all in one giant transaction, I get errors.
I'm running Postgres 8.2.1 on Windows XP. If anyone can help, I'd be very
greatful.
the table o tables have triggers?
try to use COMMIT
--- On Thu, 6/26/08, Robert James <srobertjames@gmail.com> wrote:
From: Robert James <srobertjames@gmail.com>
Subject: [GENERAL] Undocumented Postgres error: "failed to fetch old tuple for AFTER trigger"
To: "Postgres General" <pgsql-general@postgresql.org>
Date: Thursday, June 26, 2008, 8:28 PM
I'm running a very large series of commands - mainly DDL but some DML as well - in a large transaction. I get the following error, which doesn't seem to be documented:
"ERROR: failed to fetch old tuple for AFTER trigger
: COMMIT"
There are no triggers that I'm aware of. I've gotten this error when running the transaction on two different instances of the database. I also did a VACUUM FULL, which didn't help.
Breaking down the commands into a series of about 7 smaller transactions works fine. Each transaction runs totally fine by itself. But, when I run them all in one giant transaction, I get errors.
I'm running Postgres 8.2.1 on Windows XP. If anyone can help, I'd be very greatful.
"Robert James" <srobertjames@gmail.com> writes:
"ERROR: failed to fetch old tuple for AFTER trigger
I'm running Postgres 8.2.1 on Windows XP.
I seem to remember a bug with this symptom; so your first move ought to
be to update to 8.2.latest. If you can still reproduce it afterwards,
please send a test case to pgsql-bugs.
regards, tom lane