Re: Problems with pg_upgrade after change of unix user running db.

On 11/27/2015 08:15 AM, Bruce Momjian wrote:
> On Fri, Nov 27, 2015 at 04:05:46PM +0000, Benedikt Grundmann wrote:
>> > [as-proddb(at)nyc-dbc-001 upgrade-logs]$ tail pg_upgrade_dump_16416.log
>> > pg_restore: creating CHECK CONSTRAINT seqno_not_null
>> > pg_restore: creating CHECK CONSTRAINT seqno_not_null
>> > pg_restore: [archiver (db)] Error while PROCESSING TOC:
>> > pg_restore: [archiver (db)] Error from TOC entry 8359; 2606 416548282
>> CHECK
>> > CONSTRAINT seqno_not_null postgres_prod
>> > pg_restore: [archiver (db)] could not execute query: ERROR: constraint
>> > "seqno_not_null" for relation "js_activity_2011" already exists
>> > Command was: ALTER TABLE "js_activity_2011"
>> > ADD CONSTRAINT "seqno_not_null" CHECK (("seqno" IS NOT NULL)) NOT
>> VALID;
>>
>> I have no idea, but this is a pg_dump bug or inconsistent system tables,
>> rather than a pg_upgrade-specific bug.
>>
>>
>> Any recommendation on how to proceed?
>
> My guess is you are sharing the constraint name "seqno_not_null" with
> multiple tables. I think you are going to have to dig into the system
> tables to see where that is referenced and fix it.
>

In the post below the OP shows the tables involved(they where inherited):

http://www.postgresql.org/message-id/CADbMkNM_y9ewdaWdQ_8DJ1mUC0Z_FGwTyAD2RwCHgExj2jvOHQ@mail.gmail.com

Still I did not think there was a restriction that constraint names be unique
across a database or cluster:

test=> create table tbl_a (fld_1 int CONSTRAINT fld_1_chk CHECK(fld_1 > 0) );
CREATE TABLE

test=> create table tbl_b (fld_1 int CONSTRAINT fld_1_chk CHECK(fld_1 > 0) );
CREATE TABLE

test=> \d tbl_*
Table "public.tbl_a"
Column | Type | Modifiers
--------+---------+-----------
fld_1 | integer |
Check constraints:
"fld_1_chk" CHECK (fld_1 > 0)

Table "public.tbl_b"
Column | Type | Modifiers
--------+---------+-----------
fld_1 | integer |
Check constraints:
"fld_1_chk" CHECK (fld_1 > 0)

--
Adrian Klaver
adrian(dot)klaver(at)aklaver(dot)com