Re: PQendcopy:resetting connection Problem and Cannot insert a duplicate key into unique index

Thanks Scott and Tom.
Yes. I know, I know that I need to upgrade :-)
What would be the newer version of pgsql (I mean a bit higher version of my current version) that provides that capability of telling me what the duplicated key was ?

Migration is always a challenge, especially I'm not sure if I will have any unexpected hiccups when I dump my whole DB. I also need to upgrade the OS. My first priority is to get this problem fixed so that my users will stop IM, email or call me.

Mary

-----Original Message-----
From: Tom Lane [mailto:tgl(at)sss(dot)pgh(dot)pa(dot)us]
Sent: Wednesday, February 03, 2010 2:21 PM
To: Scott Marlowe
Cc: Wang, Mary Y; pgsql-general(at)postgresql(dot)org
Subject: Re: [GENERAL] PQendcopy:resetting connection Problem and Cannot insert a duplicate key into unique index

Scott Marlowe <scott(dot)marlowe(at)gmail(dot)com> writes:
> On Wed, Feb 3, 2010 at 2:16 PM, Wang, Mary Y <mary(dot)y(dot)wang(at)boeing(dot)com> wrote:
>> I still couldn't find that particular line that caused that problem :-(.  Counting was very pain.
>> Is there anyway that I can tell psql just to "ignore" (I mean don't insert it duplicate key into unique index users_pkey) and just keep going without doing the PQendcopy:resetting connection?

> Not really directly. What I'd do is remove the unique constraint,
> insert, then use something like

> select max(row_id) from table t1 join table t2 on
> t1.somefield=t2.somefield and t1.row_id<>r2.row_id;

> to find dupes and remove them.

> Then I'd dump the whole db and migrate to a more modern version of pgsql.

If you were using a more modern version of pgsql, it would tell you what the duplicated key was ;-). So maybe you could try loading the dump file into something newer as a means of debugging the problem.

regards, tom lane